Im 2

Novel Design and Measurement Methodologies of
Millimeter Wave Smart Antennas

Ghulam Ahmad
Submitted for the Degree of

Doctor of Philosophy
from the
University of Surrey
Surrey Space Center

Faculty of Engineering and Physical Sciences
University of Surrey
Guildford, Surrey GU2 7XH, U.K.
April 2018

c Ghulam Ahmad 2018
To my father for his great love, and memories which he left behind!
Summary
Next generation wireless communication systems are expected to support
unprecedented extremely high data transfer rates. This objective requires wider
bandwidths which are presently only available at the millimeter waves (mm-waves)
spectrum (30–300 GHz). Due to stringent propagation impairments, mm-waves mainly
rely on the line of sight communication links which require high gain and wide angle
beamsteeering smart antennas to maintain their performance. Owing to the complexity
and losses in array beamformers, the realization of a high gain wide angle electronic
beamsteering antenna solution at mm-waves becomes a key challenge.
This research provides a potentially competing novel high gain electronic beamsteering
antenna solution for mm-waves in the form of a phase quantized smart reflectarray
consisting of high performance reconfigurable unit cells. Novel contributions of this
research are: (a) Analysis of mm-wave reflectarray unit cells including the effects of
fringing fields, surface waves, finite metal conductivity and metal surface roughness. (b)
New measurement techniques for mm-wave reflectarray unit cells to ease the alignment,
orientation, and DC biasing issues. (c) Characterization of PIN diodes at 10 GHz and
60 GHz for their ON/OFF state models extraction from measurements. (d) Design of
three state implicit phase shifter reflectarray unit cell at 60 GHz, reduction in its DC
bias lines, and an optimization technique to improve polarization purity of a multi-state
reconfigurable unit cell. (e) A fast algorithm to prepare the electromagnetic simulation
model of large reflectarrays. (f) Conception and measurement based validation of phase
quantized reflecarrays and their performance matrix. (g) Conception and measurement
based analytical solution of low DC power consuming smart reflectarrays. The resulting
solution is agile, simple to implement, do not necessarily require multiple RF chains,
enables wide angle electronic beamsteering (±78◦ ), is scalable for any gain/frequency
requirements, can be made foldable for smaller satellite platforms, is very reliable, and
consumes low DC power. This smart reflectarray platform can implement any phase
only synthesis technique for radiation pattern control including single/multiple pencil
beams, contoured beams, and their scanning over wider angles. Findings of this research
would potentially benefit next generation terrestrial/air/space communication systems
and radars.
Key words: Reflectarray, printed antennas, high gain antenna, smart antenna, unit
cells, reflection phase, mm-waves, reconfigurable antennas
Email: g.ahmad@surrey.ac.uk; gasajid48@gmail.com

WWW: http://www.eps.surrey.ac.uk/
Acknowledgements
In the name of ALLAH, the most beneficent, the most merciful. My eternal submission
to Almighty ALLAH for everything I have in my life. I can never be grateful enough
to YOU for YOUR uncounted blessings upon me. I am, from the deepest of my heart,
thankful to YOU for granting me a beautiful life, guiding me to the true path, and
providing me the strength to complete this research. I seek YOUR forgiveness in this
world and the world hereafter!
I am extremely thankful to my supervisors Dr. Tim Brown, Prof. Craig Underwood and
Dr. Tian Hong Loh for their encouragement, great support and guidance throughout
the project without which this research would have not been possible. Thank you for
your always welcoming attitude and extended help throughout the PhD. I strongly
believe that words are too less to acknowledge your contribution in this research.
I would like to take this opportunity to thank Dr Peter Aaen and his whole team at
N3M laboratory at Advanced Technology Institute (ATI) for their support during the
measurements. The support offered by Zhengrong Tian, Martin Salter, Dr. Chong Li,
and Prof. Nick Ridler during the measurements at National Physical Laboratory (NPL)
was a great help in obtaining the reliable results. During manufacturing and test phases
of the project the staff in University of Surrey technical services unit was extremely
supportive in manufacturing the project hardware and test structures. Thank you John
Mouat, Richard Clarke, Graham Webb, Eric Worpe, Andy Walker, Bill Backhouse, and
Stefen for your great work to support this research. I highly admire the support offered
by Karen Collar and Louise Quigley for making my research at Surrey Space Center a
memorable journey. Thank you FEPS and university IT teams for your support.
I would like to thank NPL and Surrey Space Center (SSC) for funding this research. A
special thanks to National Space Agency of Pakistan for allowing me to proceed for this
PhD research. I would greatly acknowledge the support of my colleagues there which
made me able to stay focused. Many thanks to Atif Noushad for his always available
guaranteed help. Thank you Muhammad Yousuf Khan, Dr Muhammad Riaz Suddle
(SI), Dr Ismatullah, Dr Muhammad Amin, Shahnaz Yasir, Muhammad Latif, Nadeem
Faisal, Shakeel Zahid, SRDC-L colleagues, and the whole training division team for
your extended support.
I would like to thank all of my colleagues in SSC, ATI, 5GIC, and Cobbett Hill
Earthstation for their moral support. Thank you Dr Adnan Akbar, Dr Atta Mustafa,
Dr Hasan Malik, Dr Areeb Riaz, Dr Sameed Husain, Dr Adnan Zafar, Dr Haris, Dr Zoha
Ahmed, Dr Muhammad Awais, Dr Junaid Mir, Dr Arslan Saeed, Haris Votsi, Jonas
Urbonas, Sean Glipsi, Dr. Jay Mody, Zabih Khan, Richard Reid, Wael Alkhalidi,
Pragateesh, Ben Mohammadi, Adrian Redfern, Mike Wilson, Paul O’Brien, John
Bulmer and Mel Smith for your moral support and great ideas during our discussions.
I greatly acknowledge the love, support, and prayers of my parents, uncles, brothers,
and the sister. Finally, I highly appreciate the love and support of my wife, daughter
and the son. Thank you for your patience to let me complete this research.
Contents
Nomenclature xviii
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Aims and Objectives of Research . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Major Research Contributions . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2 Background to Millimeter Wave Beamsteering and Reflectarrays 13

2.1 Antenna Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Millimeter Wave Spectrum and Implications . . . . . . . . . . . . . . . . 14
2.3 Operation of Antenna Arrays . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Smart Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.1 Beamforming Techniques in Smart Antennas . . . . . . . . . . . 19
2.4.1.1 Switched Beam Smart Antennas . . . . . . . . . . . . . 19
2.4.1.2 Adaptive Beam Smart Antennas . . . . . . . . . . . . . 20
2.4.2 Beamformer Implementation . . . . . . . . . . . . . . . . . . . . 25
2.4.3 Phase Shifters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5 Available Smart Antenna Solutions . . . . . . . . . . . . . . . . . . . . . 30
2.6 Reflectarrays and Their Operation . . . . . . . . . . . . . . . . . . . . . 32
2.7 Reflectarray Unit Cell Characteristics . . . . . . . . . . . . . . . . . . . 34
2.8 Reflection Phase Control in Unit Cells . . . . . . . . . . . . . . . . . . . 36
2.8.1 Passive Reflection Phase Control . . . . . . . . . . . . . . . . . . 37
2.8.1.1 Comparison of Passive Phase Control Techniques . . . . 38
i
Contents ii
2.8.2 Phase Control by Electrical/Electronic Means . . . . . . . . . . . 39

2.8.2.1 Comparison of Reconfiguration Technologies . . . . . . 42
2.9 Control Element Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.10 Unit Cell Performance Tests . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.11 Phase Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.11.1 Continuous Phase Variations . . . . . . . . . . . . . . . . . . . . 45
2.11.2 Discrete Phase Steps . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.12 Reflectarray Cross Polarization Discrimination . . . . . . . . . . . . . . 48
2.13 Reflectarray Power Dissipation . . . . . . . . . . . . . . . . . . . . . . . 50
2.14 Research Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.15 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3 Characterization of Passive Reflectarray Unit Cells 56

3.1 Unit Cell Selection for mm-waves . . . . . . . . . . . . . . . . . . . . . . 57
3.2 Design and Analysis of mm-wave Passive Unit Cells . . . . . . . . . . . 58
3.2.1 Unit Cell Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.2.1.1 Modeling of an RLC Resonator . . . . . . . . . . . . . 60
3.2.1.2 Waveguide Coupled Unit Cell Resonator Modeling . . . 63
3.2.1.3 Application to Millimeter Wave Reflectarray Unit Cell: 64
3.3 Parametric Study of Reflectarray Unit Cells . . . . . . . . . . . . . . . . 72
3.4 Electromagnetic Simulation of Unit Cells . . . . . . . . . . . . . . . . . 86
3.5 Passive Unit Cells’ Fabrication and Measurements . . . . . . . . . . . . 90
3.6 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4 Characterization of Active Reflectarray Unit Cells 97

4.1 PIN Diode Characterization . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.1.1 A Description of the Measurement Setup . . . . . . . . . . . . . 100
4.1.2 PIN Diode Parameters Extraction . . . . . . . . . . . . . . . . . 103
4.2 PIN Diode Power Consumption and Insertion Loss . . . . . . . . . . . . 106
4.2.1 Series SPST PIN Diode Switch . . . . . . . . . . . . . . . . . . . 106
4.2.2 Insertion Loss of a PIN Diode Switch . . . . . . . . . . . . . . . 107
4.2.3 PIN Diode Series Switch Measurements . . . . . . . . . . . . . . 111
Contents iii
4.3 Active Unit Cell Characterization . . . . . . . . . . . . . . . . . . . . . . 117

4.3.1 Active Unit Cell Design . . . . . . . . . . . . . . . . . . . . . . . 117
4.3.2 Active Unit Cell Simulations: Using PIN Diodes . . . . . . . . . 121
4.3.3 Cross Polarization Isolation in a Dynamic Environment . . . . . 125
4.3.4 Fabrication and Measurements . . . . . . . . . . . . . . . . . . . 131
4.4 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5 Reflectarray Design and Analysis 144

5.1 Design and Analysis of Reflectarrays . . . . . . . . . . . . . . . . . . . . 145
5.1.1 Reflectarray Physics . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.1.2 Aperture Efficiency of a Reflectarray . . . . . . . . . . . . . . . . 149
5.1.2.1 Feed Horn and Unit Cell Radiation Pattern Models . . 150
5.1.2.2 Spillover Efficiency . . . . . . . . . . . . . . . . . . . . . 151
5.1.2.3 Illumination Efficiency . . . . . . . . . . . . . . . . . . 154
5.1.2.4 Parametric Study of Reflectarray Efficiency . . . . . . . 155
5.1.3 Radiation Pattern of a Reflectarray . . . . . . . . . . . . . . . . 161
5.2 Phase Discretization in Reflectarrays . . . . . . . . . . . . . . . . . . . . 171
5.3 Reflectarray EM Simulation Model Building . . . . . . . . . . . . . . . . 177
5.4 Unit Cells’ Phase State Selection in Phase Discretized Reflectarrays . . 181
5.5 DC Power Consumption in Large Reflectarrays . . . . . . . . . . . . . . 188
5.6 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
6 Reflectarray Measurements 194

6.1 Design of Reflectarray Passive Demonstrators . . . . . . . . . . . . . . . 195
6.2 Millimeter Waves Antenna Measurement Setup . . . . . . . . . . . . . . 199
6.3 Measurement of Passive Demosntrators . . . . . . . . . . . . . . . . . . 203
6.3.1 2D Radiation Pattern and Bandwidth Response . . . . . . . . . 210
6.3.2 3D Radiation Pattern . . . . . . . . . . . . . . . . . . . . . . . . 220
6.4 Active Reflectarrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
6.5 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
7 Conclusion and Future Research 226

7.1 Thesis Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
7.2 Future Research Directions and Limitations of this Research . . . . . . . 231
Contents iv
A Passive and Active Unit Cells 234

A.1 Reflection Phase Control in Unit Cells . . . . . . . . . . . . . . . . . . . 234
A.1.1 Passive Reflection Phase Control . . . . . . . . . . . . . . . . . . 235
A.1.1.1 Stub Loaded Patches . . . . . . . . . . . . . . . . . . . 235
A.1.1.2 Patches with Variable Geometric Features . . . . . . . . 235
A.1.1.3 Compound Variable Geometry Patches . . . . . . . . . 236
A.1.1.4 Phoenix Reborn Cells . . . . . . . . . . . . . . . . . . . 237
A.1.1.5 Multi-Scale Cell Splitting: Variable Length Slots . . . . 238
A.1.1.6 Element Rotation for Circular Polarization . . . . . . . 239
A.1.1.7 Comparison of Passive Phase Control Techniques . . . . 239
A.1.2 Phase Control by Electrical/Electronic Means . . . . . . . . . . . 240
A.1.2.1 Electronic Reconfiguration . . . . . . . . . . . . . . . . 242
A.1.2.2 Optical Reconfiguration . . . . . . . . . . . . . . . . . . 244
A.1.2.3 Reconfiguration Using Smart Materials . . . . . . . . . 244
A.1.2.4 Physical Structural Reconfiguration . . . . . . . . . . . 245
A.1.2.5 Comparison of Reconfiguration Technologies . . . . . . 246
A.2 Reconfigurable Unit Cells . . . . . . . . . . . . . . . . . . . . . . . . . . 246
A.2.1 Single Versus Dual Polarized Unit Cells . . . . . . . . . . . . . . 247
A.2.2 A Dual Polarized Tuneable Unit Cell . . . . . . . . . . . . . . . . 248
A.2.3 A Single Polarized Frequency Selectable Tuneable Unit Cell . . . 250
A.2.4 Monolithic MEMS Based Single Polarized Unit Cell . . . . . . . 250
A.2.5 A Polarization Flexible Unit Cell . . . . . . . . . . . . . . . . . . 251
B Unit Cell Reflection Phase and DC Biasing 252
Bibliography 256
List of Figures
1.1 High gain antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.2 Reflectarray antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1 Millimeter wave spectrum. Currently under consideration mm-wave

bands are indicated in green color. . . . . . . . . . . . . . . . . . . . . . 15
2.2 Wavefront and array phase shift . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Smart antenna classification. . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Switched and adaptive beam smart antennas . . . . . . . . . . . . . . . 20
2.5 Adaptive beamforming in smart antennas. . . . . . . . . . . . . . . . . . 22
2.6 Spatial-temporal processing in adaptive arrays . . . . . . . . . . . . . . 24
2.7 Beamforming implementation techniques . . . . . . . . . . . . . . . . . . 25
2.8 Printed reflectarray antenna on a grounded substrate . . . . . . . . . . . 33
2.9 Various types and shapes of passive unit cells . . . . . . . . . . . . . . . 37
2.10 Antenna reconfiguration technologies and their ingredients . . . . . . . . 41
2.11 Through-reflect-line (TRL) calibration kit and PIN diode mounted for
test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.1 A reflectarray unit cell (front and side views). . . . . . . . . . . . . . . . 60

3.2 RCL equivalent circuit of a reflectarray unit cell (a) Lossless case, (b)
Lossy case, loss represented by a resistor in series with capacitor to
model the increase in loss due to an increase in frequency, (c) A unit cell
resonator coupled to the waveguide with an external excitation source
(dominant mode waveguide incidence), the incident wave is represented
by s+ (t), the reflected wave is represented by s− (t), and the reflection
coefficient of the unit cell is represented by Γ. . . . . . . . . . . . . . . . 61
3.3 Radiated wave efficiency = η as a function of the substrate thickness
for various values of relative dielectric constants. For each r case the
substrate thickness is limited to allow only the first order mode of surface
waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
v
List of Figures vi
3.4 Quality factors Qo and Qext as a function of the substrate thickness.

When Qext ≤ Qo is the over coupled region. . . . . . . . . . . . . . . . . 73
3.5 Reflection coefficient as a function of the substrate thickness at 60 GHz.
Γ (fring smooth) considers the effect of fringing fields in microstrip and
copper conductivity without its surface roughness. In Γ (fring rough)
the effect of conductor surface roughness is also considered in addition
to fringing fields. While in Γ (fring rough SW) the effects of fringing
field, conductor roughness and surface waves are considered. . . . . . . . 75
3.6 Phase swing ∆Φ (deg) as a function of substrate thickness. Two cases
are shown where patch length was changed by 20% and 100%. . . . . . 75
3.7 Reflection coefficient (Γ) magnitude versus frequency for various
substrate thickness values considering the effects of fringing fields,
conductor roughness, and surface waves. . . . . . . . . . . . . . . . . . 77
3.8 Reflection coefficient’s (Γ) phase versus frequency for various substrate
thickness values considering the effects of fringing fields, conductor
roughness, and surface waves. . . . . . . . . . . . . . . . . . . . . . . . 77
3.9 Reflection coefficient’s (Γ) magnitude versus change in patch length (L)
for various substrate thickness values. . . . . . . . . . . . . . . . . . . . 78
3.10 Reflection coefficient’s (Γ) phase versus change in patch length (L) for
various substrate thickness values. . . . . . . . . . . . . . . . . . . . . . 79
3.11 Reflection coefficient’s (Γ) magnitude versus change in patch width (W )
for various substrate thickness values. . . . . . . . . . . . . . . . . . . . 79
3.12 Reflection coefficient’s (Γ) phase versus change in patch width (W ) for
various substrate thickness values. . . . . . . . . . . . . . . . . . . . . . 80
3.13 Reflection coefficient’s (Γ) magnitude versus change in substrate relative
dielectric constant(r ) for various substrate thickness values. . . . . . . 80
3.14 Reflection coefficient’s (Γ) phase versus change in substrate relative
dielectric constant(r ) for various substrate thickness values. . . . . . . 81
3.15 Reflection coefficient’s (Γ) magnitude versus change in substrate loss
tangent (tanδ) for various substrate thickness values. . . . . . . . . . . 82
3.16 Reflection coefficient’s (Γ) phase versus change in substrate loss tangent
(tanδ) for various substrate thickness values. . . . . . . . . . . . . . . . 82
3.17 Reflection coefficient’s (Γ) magnitude versus change in metal
conductivity (σ) for various substrate thickness values. Metal
conductivity is plotted as logarithm to show the effect of lower
conductivity values leading to sever losses in unit cell for thin substrates. 83
3.18 Reflection coefficient’s (Γ) magnitude versus change in length (L) of a
square patch for various substrate thickness values. . . . . . . . . . . . 84
3.19 Reflection coefficient’s (Γ) magnitude versus frequency (f ) of a square
patch for various substrate thickness values. . . . . . . . . . . . . . . . 84
List of Figures vii
3.20 Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch

for various substrate thickness values. Φres is plotted only without
considering the effects of reflections from un-occupied area of the unit cell. 85
3.21 An RLC equivalent circuit of a mm-wave unit cell on a 10 mil RO5880
at 59 GHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.22 Unit cell on a grounded substrate simulated in CST using Floquet ports. 87
3.23 Unit cell on a grounded substrate simulated in CST using metallic
waveguide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.24 A comparison of unit cell simulated using metallic ports; and floquet
ports with various wave incident angles. Incident angle at 60 GHz is
around 48◦ in metallic waveguide which matches very well with the
floquet port simulation for the same angle. . . . . . . . . . . . . . . . . . 89
3.25 Unit cell types, (a) to be tested by placing inside the waveguide, (b) unit
cell to be mounted to waveguide by screws, and (c) unit cell with extra
copper on top surface to be mounted to waveguide with screws. . . . . . 90
3.26 Unit cell on a grounded substrate under test, front and rear views. . . . 91
3.27 Waveguide test setup with a unit cell. . . . . . . . . . . . . . . . . . . . 91
3.28 Fabricated unit cells under test. . . . . . . . . . . . . . . . . . . . . . . . 92
3.29 Reflection coefficient’s (Γ) magnitude versus frequency (f ) of a square
patch on a 10 mil thick substrate. A comparison of measured unit cells’
results, CST prediction and analytically calculated results . . . . . . . . 93
3.30 Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch
on a 10 mil thick substrate. A comparison of measured unit cells’ results,
CST prediction and analytically calculated results . . . . . . . . . . . . 93
3.31 Reflection coefficient’s (Γ) magnitude versus patch length (L) for a
square patch on a 10 mil thick substrate at 59 GHz. . . . . . . . . . . . 94
3.32 Reflection coefficient’s (Γ) phase versus patch length (L) for a square
patch on a 10 mil thick substrate at 59 GHz. . . . . . . . . . . . . . . . 94
4.1 Equivalent circuits of a PIN diode in ON (forward bias) and OFF (reverse
bias) states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.2 V band PIN diodes test structures with and without DC isolation
capacitors (FB: finger break). . . . . . . . . . . . . . . . . . . . . . . . . 101
4.3 X band PIN diodes test structures with and without DC isolation
capacitors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.4 V and X band PIN diode fabricated PCBs for PIN diode characterization
and their corresponding test setups. . . . . . . . . . . . . . . . . . . . . 102
4.5 Characterization of MACOM PIN diodes MA4AGBLP912 and
MA4GP905 at V band. . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
List of Figures viii
4.6 Characterization of MACOM PIN diodes MA4AGBLP912 and

MA4GP905 at X band. . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.7 Forward resistance under 5 mA forward current for MACOM PIN diode
MA4AGBLP912 at V band. . . . . . . . . . . . . . . . . . . . . . . . . 106
4.8 A PIN diode series switch . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.9 Forward resistance (Rs ) as a function of forward bias current (If ). . . . 109
4.10 PIN diode current and voltage drop in forward bias. . . . . . . . . . . . 110
4.11 Insertion loss as a function of forward bias current. . . . . . . . . . . . . 110
4.12 Fabricated PIN diode switch circuit PCB, and measurement setup. . . . 112
4.13 Measured insertion loss of a series PIN diode switch with variable DC
voltage and fixed values of the current limiting resistor. . . . . . . . . . 113
4.14 Measured insertion loss and corresponding forward resistance at 10.33
GHz with variable DC voltage and fixed values of the current limiting
resistor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.15 Measured insertion loss of a series PIN diode switch with fixed DC
voltage and selectable variable values of the current limiting resistor. . . 114
4.16 Measured insertion loss and corresponding forward resistance at 10.33
GHz with fixed DC voltage and selectable variable values of the current
limiting resistor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.17 PIN diode current and voltage drop in forward bias calculated based on
corresponding values of Rs from the measurements. . . . . . . . . . . . . 116
4.18 DC power consumption and insertion loss of a series PIN diode switch. 116
4.19 Active unit cell using PIN diodes. . . . . . . . . . . . . . . . . . . . . . 119
4.20 Various views of a mm-wave active unit cell using PIN diodes for
reconfiguration to implement a three state integrated RF plane implicit
phase shifter on a 2.7 mm square lattice. Via holes are shared by
neighboring unit cells when placed in an array, therefore only half portion
is shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.21 Reflection loss magnitude response of an active unit cell using PIN
diodes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.22 Reflection phase response of an active unit cell using PIN diodes. . . . 124
4.23 Co and cross polarization reflection magnitudes when reconfiguration
takes place. Only single polarization results are shown due to symmetry. 127
4.24 Co and cross polarization reflection coefficient behaviour for various
values of offsets ∆X and ∆Y in mm: D2X , D2Y ON. Only single
polarization results are shown due to symmetry. . . . . . . . . . . . . . . 129
4.25 Reflected co and cross polarized field magnitude response of an active
unit cell using PIN diodes after optimization. . . . . . . . . . . . . . . . 130
List of Figures ix
4.26 Various views of a mm-wave active unit cell for manufacturing. Unit
Cell simulated in CST using Floquet ports. . . . . . . . . . . . . . . . . 132
4.27 Reflection coefficient’s magnitude for the active unit cell, simulated in
CST for manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.28 Reflection coefficient’s phase for the active unit cell, simulated in CST
for manufacturing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.29 Reflection coefficient’s XPD for the active unit cell, simulated in CST
for manufacturing. XPD can be further optimized for OFF state. . . . . 133
4.30 Various views of a mm-wave active unit cell for manufacturing. Unit
Cell simulated in CST using waveguide ports. The unit cell structure is
made big to be compatible with the waveguide test setup. . . . . . . . . 134
4.31 Incident electric field for two modes (TE10 and its orthogonal) and unit
cell surface currents for D1X and D1Y ON. Red color represents maximum
while minimum is represented by blue. . . . . . . . . . . . . . . . . . . . 135
4.32 Reflection coefficient’s magnitude and phase for a loss less and lossy full
structure of the unit cell. For lossy case, the substrate (tan δ = 0.004)
is lossy while metallic surfaces are considered as PEC. . . . . . . . . . . 137
4.33 Cross polarization discrimination (XPD) for the active unit cell’s full
structure. A similar trend was observed for lossy and loss less cases. . . 137
4.34 Fabricated unit cell photos (full structure and zoomed by microscope). . 138
4.35 Reflection coefficient’s phase for the active unit cell in passive state; prior
to PIN diodes’ mounting . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.36 Measurement setup for active unit cells. The DC probe feeds the DC
voltage to unit cell. Current limiting resistors, current meter, and diodes
are all in series connected to the DC power supply. . . . . . . . . . . . . 139
4.37 Reflection coefficient’s phase for the active unit cell with PIN diodes. . . 140
4.38 Reflection coefficient’s phase for the active unit cell with PIN diodes: a
zoomed view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.39 Reflection coefficient’s magnitude for the active unit cell with PIN diodes.141
5.1 Co-ordinate system for reflectarray analysis. . . . . . . . . . . . . . . . . 147

5.2 Feed horn radiation pattern and directivty for various values of
parameter qf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.3 A prime focus circular aperture reflectarray and its spillover efficiency
calculation (F = 70 mm, D = 94.5 mm, qf = 4, qe = 1.6, α = 1 to 34
degree). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.4 Illumination and spillover efficiency values versus F/D for a fixed
diameter, (D = 94.5 mm, qf = 4, qe = 1.6, focal length (F ) 10 to
400 mm , offset angle =0 (center fed case), feed pointing (0,0,0) ). . . . 156
List of Figures x
5.5 Efficiency ηs × ηill as a function of array side length, focal length and qf
where color bars represent the value of efficiency. . . . . . . . . . . . . 157
5.6 Illumination and spillover efficiency values versus unit cell and feed cosine
function powers (qe , qf ), (D = 94.5 mm, qf = 4, qe = 1.6, focal length
(F ) 70 mm , offset angle =0 (center fed case), feed pointing (0,0,0) ). . 158
5.7 Efficiency ηs × ηill versus feed cosine power qf and offset angles, (D =
94.5 mm, F = 70 mm, qe = 1.6 ). . . . . . . . . . . . . . . . . . . . . . . 158
5.8 Offset angle versus efficiency predictions for various offset angles. The
feed beam point (xo , yo ) is adjusted to obtain the maximum efficiency for
the configuration. It is found that feed beam point should be offset by
an amount yo from the center of a reflectarray fed in offset configuration
to achieve the maximum efficiency. (D = 94.5 mm, qf = 4, qe = 1.6,
focal length (F ) 70 mm). The color bars represent the value of efficiency. 159
5.9 Efficiency ηs × ηill versus feed location for qf = 4 and 8, (D = 94.5 mm,
qe = 1.6), where color bars represent the value of efficiency. . . . . . . . 160
5.10 Efficiency ηs × ηill versus feed and unit cell cosine powers qf & qe . (D =
94.5 mm, F = 70 mm), where color bars represent the value of efficiency. 161
5.11 Continuous phases distribution over reflectarray aperture for beam
direction in (θ, φ) = (0, 0). The colors represent the required phase
at the center of a unit cell’s location, and the color bar represents the
phase in degrees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.12 Power patterns for unit cell and feed horn, where a higher value of the
parameter q results in a narrower beamwidth. . . . . . . . . . . . . . . . 165
5.13 Distance of each unit cell from the reflectarray center, dmn (mm),
where colors represent the distance of the center of a unit cell from the
reflectarray aperture center, and the color bar represents this distance
in mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.14 Distance of each unit cell from the feed horn’s phase center, rmn (mm). 166
5.15 Unit cells’ angle from cos θe (deg), where colors represent the angle at
the unit cell center at each location, and the color bar represents angle
in degrees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.16 Incident angle for unit cells from feed horn (deg). . . . . . . . . . . . . . 167
5.17 Illumination tapering over reflectarray aperture by feed horn (dB). . . . 168
5.18 Phase of the incident field on unit cells from the feed horn(deg). . . . . 169
5.19 Normalized radiation pattern of a 35 × 35 element reflectarray. . . . . . 170
5.20 Directivity loss in quantized reflectarrays with various number of phase
states per unit cell simulated in CST Microwave Studio. . . . . . . . . . 173
5.21 Directivity loss in reflectarrays with scan angle simulated in Matlab.
Directivity for each case is normalized with the directivity of continuous
phase case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
List of Figures xi
5.22 Normalized directivity loss in reflectarrays with scan angle simulated in

Matlab. Directivity for each case is normalized with the directivity of
continuous phase case at each scan angle. . . . . . . . . . . . . . . . . . 175
5.23 Desired beam pointing angle versus achieved beam pointing angle in
reflecarrays, simulated in Matlab. . . . . . . . . . . . . . . . . . . . . . . 175
5.24 Normalized radiation pattern of phase quantized reflectarrays, simulated
in CST Microwave Studio. . . . . . . . . . . . . . . . . . . . . . . . . . . 176
5.25 Reflectarray construction algorithm. . . . . . . . . . . . . . . . . . . . . 179
5.26 A sample of a four states phase quantized reflectarray constructed
through algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
5.27 Normalized radiation pattern of a three states phase quantized
reflectarray with various phase state selections. . . . . . . . . . . . . . . 182
5.28 Phase state scattering in phase discretized reflectarrays. . . . . . . . . . 183
5.29 Phase state scattering in phase discretized reflectarrays simulated in CST.186
5.30 Normalized radiation pattern of two, three states, phase quantized
reflectarrays with phase scattering simulated in CST. . . . . . . . . . . . 186
5.31 Normalized directivities of two, three states, phase quantized
reflectarrays with phase scattering. @ FLx = at focal length x. . . . . . 187
5.32 DC power consumption of large phase quantized reflectarrays at X and
V bands using PIN diode test data. Results for two and three states
phase quantization for single and dual polarized unit cells are shown. . . 191
6.1 A 3 states phase quantized reflectarray at 0◦ and 55◦ beam pointing

angles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
6.2 Phase quantized fabricated reflectarrays. Each reflectarray was mounted
on a plastic support to provide physical strength to the 10 mil thick
RO5880 grounded substrate. . . . . . . . . . . . . . . . . . . . . . . . . . 199
6.3 Antenna measurement setup. . . . . . . . . . . . . . . . . . . . . . . . . 200
6.4 Antenna measurement chamber setup . . . . . . . . . . . . . . . . . . . 202
6.5 AUT support structures. . . . . . . . . . . . . . . . . . . . . . . . . . . 202
6.6 Measured loss in gain with achieved beam pointing angles for
reflectarrays having various number of phase qunatization states. Values
are normalized w.r.t. the gain of 8 states case at 0◦ . Measurement points
are shown by geometric symbols at the frequency of 61.5 GHz. . . . . . 207
6.7 Measured beam pointing angles for reflectarrays having various number
of phase qunatization states. The continuous phase case is plotted for
reference as was calculated using Matlab R
. Measurement points are
shown by geometric symbols at the frequency of 61.5 GHz. . . . . . . . 209
List of Figures xii
6.8 Response of a V band 3 states phase quantized reflectarray designed

to point its beam at 0◦ , a comparison of measured results with CST
simulated results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
6.19 Measured normalized gain versus frequency for a 3 states phase quantized
reflectarray at various pointing angles to show bandwidth response at
pointing angles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
6.20 A comparison of the transformed far field, and the near field for a 3 states
phase quantized reflectarray designed for 55◦ beam pointing. As the
distance between antennas is just equal to the start of far field distance,
this measured near field is termed as “near field at just far field distance”. 219
List of Figures xiii
6.21 Measured 3D radiation pattern at 60 GHz of a 3 states phase quantized

reflectarray designed for (θ, φ) = (55◦ , 0◦ ). The color bar represent the
measured gain (dBi). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.22 2D radiation pattern in θ plane (called Plane 1). . . . . . . . . . . . . . 222
6.23 2D radiation pattern in φ plane (called Plane 2) at θ = 55◦ . . . . . . . . 222
6.24 2D radiation pattern in θ and φ planes (called Plane 1 and 2 respectively).223
A.1 Various types and shapes of passive unit cells . . . . . . . . . . . . . . . 235

A.2 (a) A reflectarray made up of double cross elements, (b) Simulated
incident field, (c) Near field measured over the reflectarray aperture:
(color indicates the field strength with red being the maximum), at
places where phase transition from 360◦ to 0◦ happens field can be
seen vanishing in (c) due to unit cell sudden geometric transitions. . . . 238
A.3 Antenna reconfiguration technologies and their ingredients . . . . . . . . 242
A.4 Electronic reconfiguration technologies: (a) Varactor diode loaded cell,
(b) Phase shifter loaded unit cell, (c) PIN/RF-MEMS switch loaded cell. 243
A.5 Optical reconfiguration technologies: (a) Optical switching through
non-integral fiber optic, (b) Optical switching through integral fiber
optic, (c) Directly integrated LASER without a fiber optic. . . . . . . . 244
A.6 Operation of liquid crystals: (a) Liquid crystal dipole/molecules
allignment w.r.t. bias voltage , (b) resulting change in (εr ) and loss
tangent (tan δ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
A.7 (a) A split ring unit cell, (b) Rotation of split ring by an angle ψ, (c)
MEMS based dual band implementation of rotation based technique. . . 246
A.8 Inter-cell spacing for (a) dual polarized unit cells and (b) single polarized
unit cells. Dual polarized unit cells offer a compact solution as the
inter-cell
√ spacing is d as compared to single polarized unit cells where it
is 2 d. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
A.9 Various reconfigurable unit cells (a) Dual polarized unit cell
reconfigured by a combination of varactor and PIN diodes, (b) Single
polarized frequency reconfigurable unit cell using PIN switches and
varactor tuning, (c) Reconfiguration using RF-MEMS, (d) Polarization
reconfiguration using PIN diode switches, (e) MEMS based circular
polarized unit cell based on rotation principle. . . . . . . . . . . . . . . . 249
B.1 Continuous reflection phase required from unit cells in a reflectarray of

35 × 35 unit cells. Top row and left most column indicate the number o
unit cell in x and y directions. Similarly unit cells’ central locations are
listed in mm for both axis. . . . . . . . . . . . . . . . . . . . . . . . . . 253
List of Figures xiv
B.2 Discrete 3 states quantized reflection phase required from unit cells in a
reflectarray of 35 × 35 unit cells. Top row and left most column indicate
the number o unit cell in x and y directions. Similarly unit cells’ central
locations are listed in mm for both axis. . . . . . . . . . . . . . . . . . 254
B.3 DC bias voltage for discrete 3 states quantized reflection phase in a
reflectarray of 35 × 35 unit cells. Top row and left most column indicate
the number o unit cell in x and y directions. Similarly unit cells’ central
locations are listed in mm for both axis. Here, 0 V, 5 V and -5 V are
shown. However, these diodes can be operated by a ±1.5 V supply
instead of ±5 V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
List of Tables
2.1 Unlicensed mm-wave band and maximum transmit power in various

regions of the world . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Insertion loss for various phase shifter technologies . . . . . . . . . . . . 29
2.3 Smart antennas, a comparison . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4 Unit cell properties and expectations . . . . . . . . . . . . . . . . . . . . 35
2.5 A comparison of passive phase control techniques . . . . . . . . . . . . . 38
2.6 Reconfigurable properties of an antenna . . . . . . . . . . . . . . . . . . 40
2.7 A comparison of reconfiguration technologies . . . . . . . . . . . . . . . 42
2.8 Number of phase states and directivity loss at 30 GHz . . . . . . . . . . 47
2.9 Performance of a 64 element phased array with various number of phase
shifter bits at 30 GHz . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.1 A comparison of unit cell measurement techniques/structures. . . . . . . 92
4.1 A comparison of pin diode switches . . . . . . . . . . . . . . . . . . . . . 108
5.1 Reflectarray formulation definitions . . . . . . . . . . . . . . . . . . . . . 148

5.2 Performance of phase quantized reflectarrays . . . . . . . . . . . . . . . 172
5.3 Effect of phase state selection in a three states phase quantized reflectarray.182
5.4 Effect of phase state scattering around the chosen phase states in a three
states phase quantized reflectarray. . . . . . . . . . . . . . . . . . . . . . 185
5.5 DC power consumption of reflectarrays at X band. . . . . . . . . . . . . 189
5.6 DC power consumption of reflectarrays at V band. . . . . . . . . . . . . 189
5.7 Expected gain of reflectarrays at V band. . . . . . . . . . . . . . . . . . 190
6.1 Passive demonstrators of reflectarrays at fixed beam pointing angles. (F

= 70 mm, L = W = 94.5 mm, qe = 1.6, qf = 4) . . . . . . . . . . . . . 196
xv
List of Tables xvi
6.2 A comparison of CST simulated and measured results for passive

demonstrators of reflectarrays at fixed beam pointing angles. (35 ×
35 elements, inter-element spacing = 2.7 mm, F = 70 mm, L = W =
94.5 mm, feed horn model LB-15-10-A from A-Info, frequency for CST
results = 60 GHz, frequency for test results = 61.5 GHz) . . . . . . . . 205
6.3 A comparison of CST simulated and measured results for passive
demonstrators of reflectarrays at fixed beam pointing angles. (35 ×
35 elements, inter-element spacing = 2.7 mm, F = 70 mm, L = W =
94.5 mm, feed horn model LB-15-10-A from A-Info, frequency for CST
results = 60 GHz, frequency for test results = 61.5 GHz, sidelobe level
= SLL, cross polarization discremination = XPD.) . . . . . . . . . . . . 206
6.4 Gain reduction versus scan angle . . . . . . . . . . . . . . . . . . . . . . 208
A.1 A comparison of passive phase control techniques . . . . . . . . . . . . . 239

A.2 Reconfigurable properties of an antenna . . . . . . . . . . . . . . . . . . 241
A.3 A comparison of reconfiguration technologies . . . . . . . . . . . . . . . 247
Nomenclature
Acronyms
5G Fifth Generation cellular communication
BJT Bipolar Junction Transistor
C, X, Ku, Ka Frequency bands
CAD Computer Aided Design
CMOS Complementary Metal Oxide Semiconductor
CP Circular Polarization
CST Computer Simulation Technologies
DAC Digital to Analog Conversion
dBi Decibels in comparison to an Isotropic source
DC Direct Current
DoA Direction of Arrival
DSP Digital Signal Processing
DUT Device Under Test
EIRP Effective Isotropic Radiated Power
EM Electromagnetic
ESPAR Electronically Steerable Parasitic Array Radiator
ESPRIT Estimation of Signal Parameters via Rotational Invariance Technique
FET Field Effect Transistor
FRIC Radio Frequency Integrated Circuit
FW-EM Full Wave Electro-Magnetic
G/T Antenna Gain over system noise Temperature
GHz Giga Hertz = 109 cycles/second
HFSS High Frequency Structure Simulator
HPA High Power Amplifier
HPBW Half Power Beam Width
IC Integrated Circuit
IF Intermediate Frequency
IoT Internet of Things
MEMS Micro Electro-Mechanical Switch
MIMO Multiple Input Multiple Output
mm-waves Millimeter Waves
MMIC Monolithic Microwave Integrated Circuit
MUSIC Multiple Signal Classifier
nLoS non-Line of Sight
OFDM Orthogonal Frequency Division Multiplexing
xvii
List of Tables xviii
PCB Printed Circuit Board

RF Radio Frequency
RLC Resistor, Inductor and Capacitor
SAGE Space Alternating Generalised Expectation
SAR Synthetic Aperture Radar
SDMA Space Division Multiple Access
SIW Substrate Integrated Waveguide
SNR Signal to Noise Ratio
Tbps Tera bits per second
TE Transverse Electric
TRL Thru-Reflect-Line
WRC World Radio-communication Conference
XPI/XPD Cross Polar Isolation/Discrimination
Chapter 1
Introduction
A continuous quest to achieve unprecedented extremely high data transfer rates in

the order of Tera bits per second (Tbps) using wireless medium is a key driver for a
shift towards millimeter waves (mm-waves). Such high data rates require wider and
preferably contiguous bandwidths, which are currently only available at mm-waves.
Therefore, next generation communication systems are going to unleash the potential
of currently underutilized mm-waves spectrum (30 - 300 GHz). In a communication
system an antenna acts as an interface between radio frequency (RF) electronics and
the free space to radiate RF signals. A theoretical antenna which radiates equally in all
directions is called an omni-directional antenna. The ability of a directional antenna
to focus the radiated signal in a given direction (antenna boresight) as compared to
omni-directional antenna is called the antenna gain. In practice all the antennas are
somewhat directional.
At mm-waves the communication path loss increases rapidly. Therefore, to maintain

a reliable wireless communication link over a longer distance under the scenarios
of regulated upper bound on transmit power makes the antenna gain a primary
determinant of the system performance. When antennas are made high gain, they
become very directional which results in very narrow radiation beams therefore, the
target tracking becomes an important concern to maintain a communication link.
Through the use of smart antennas, the radiation characteristics (e.g. frequency,
radiation pattern, or polarization etc.) are made adaptive to suit the surrounding
1
1.1. Motivation 2
RF environment. For high throughput in terms of data transfer rates and ubiquitous
coverage it is envisioned that high gain smart antennas would be a primary constituent
of mm-wave communication systems.
In a radiation beamsteerable smart antenna two key parameters are its gain and
beamsteering range. High antenna gains and wide angle beamsteering features enable
mm-wave antennas for their integration into terrestrial, space, airborne and radar
applications. In terrestrial communication these antennas are highly desired for 5G
as well as satellite earth stations. Satellite-earth and inter-satellite links can benefit
from mm-wave smart antennas. Similarly, passenger and military aircrafts need smart
antennas for their connectivity with a satellite constellation or the ground terminals.
Millimeter waves also enable better precision of target detection when applied in ground
or airborne radar applications. Due to numerous applications at mm-waves, currently V
band (around 60 GHz) is under heavy investigations for the realization of high capacity
line of sight communication systems implementing antenna tracking.
This research is focused at realizing aesthetic, inexpensive, agile, and scalable high
gain smart antenna solutions by considering various antenna configurations, antenna
elements, and low-cost phase tuning mechanisms. To combine the advantages of
reflector antennas, microstrip patch arrays, and smart antenna technologies is the main
innovation of this research. Findings of this research will lay foundations for further
developments to enable new cutting edge applications in air/space borne synthetic
aperture radars for earth observations as well as in terrestrial/space communication
systems.
This chapter presents the motivation behind this research, its objectives, major research
outcomes, and the thesis outline.
1.1 Motivation
A rapid growth of the wireless communication market, the advent of internet of

things (IoT), a fast rising demand for extremely high quality real world wireless
videos, and futuristic applications yet to be commercialized require large contiguous
1.1. Motivation 3
bandwidths. A continuously increasing high resolution demand in observation systems

including satellite remote sensing, earth’s weather monitoring, radio astronomy, solar
system and deep space explorations is pushing the limits of technology even further.
The mm-waves spectrum is a potential candidate solution for the next generation
communication, tracking, and observation systems. On one hand it is capable of
providing contiguous large bandwidths to support higher data rates. While on the
other hand, due to smaller wavelengths it enables a better target detection resolution.
However, these charming benefits come at a cost of an increased system complexity.
The propagation environment is very harsh for mm-waves. Millimeter waves suffer
an increased atmospheric loss at certain bands (60 GHz in particular due to oxygen
absorption) and are prone to rain fades.
To overcome these propagation losses, mm-waves systems demand high values of

effective isotropic radiated power (EIRP) and antenna gain over system noise
temperature (G/T). Both of these system level parameters are directly controlled by
the gain of an antenna. Presently, it is extremely hard to build mm-wave high power
amplifiers (HPA) [1]. Additionally, there are certain safety regulations which put an
upper bound on the mm-wave transmit power. This further increases the significance
of high antenna gains. Therefore, high gain antennas are unavoidable in mm-wave
communications over relatively longer distances [2–6]. Millimeter wave communication
links rely on a clear line of sight requirement. When mm-waves antennas are made
high gain they become highly directional, and their radiated beams become very narrow.
Although this reduces interference at mm-waves, however to maintain a communication
link in a mobile environment becomes very challenging. To provide a ubiquitous
coverage and seamless connectivity to communication terminals the high gain mm-wave
antennas are required to be agile beamsteerable at wider angles.
From antenna design perspective, the simultaneous requirements of high gain and wide
angle agile beamsteering at mm-waves present a serious challenge. Two traditional
antenna techniques to achieve high gains are parabolic reflector antennas and phased
arrays as shown in Fig. 1.1. However, both of these technologies suffer significant
disadvantages when applied to mm-waves. While parabolic reflectors can achieve very
high gains, their beamsteering capability by electronic means is very limited due to
1.1. Motivation 4
their focused beams. Due to curved nature of parabolic surfaces, these reflectors are
difficult to manufacture with reasonable accuracy especially at mm-wave frequencies [7].
Presently, communication satellites (Fig. 1.1 (a) [8]) benefit from contoured beam
antennas to limit their radiations in a specified geographical region. These contoured
beam reflector antennas made of carbon fiber re-inforced polymers [9] require invar
molds for their manufacturing which is extremely costly and time consuming to build.
As a result the satellite antenna coverage is bound to be frozen well before its launch
[10]. Moreover, during launch phase of a satellite the accommodation of curved surfaces
in launch fairing becomes a problem. To accommodate a large parabolic reflector
in a small satellite platform is almost impossible unless made piece-wise deployable.
Additionally, in large parabolic reflector antennas, the beamsteering achieved by
mechanical means which physically moves a heavier antenna structure is very slow
e.g. earth station antennas, and antennas shown in Fig. 1.1 (b [11] & c [12]).
(a) Satellite antennas (b) FAST observatory (c) ALMA observatory
(d) Phased array radar (e) Phased arrays control
Figure 1.1: High gain antennas
In contrast, high gain phased arrays can achieve wide angle electronic beamsteering
e.g. in a phased array radar as shown in Fig. 1.1 (d) [13]. Due to electrical design
requirements the gain of a mm-wave tiny individual antenna element is only a few
1.1. Motivation 5
decibels compared to the isotropic antenna (dBi) . Therefore, a large number of these
antenna elements are required in an array to achieve high gains. In a phased array
usually the excitation phase (and amplitude) of each antenna element is controlled
by employing phase shifters (and power amplifiers). Due to a very large number
(thousands) of individual antenna elements required at mm-waves to achieve high
gains, the implementation of phased arrays is extremely difficult. The performance of
mm-wave large phased arrays is deteriorated due to their extremely lossy beamforming
networks [14]. These higher losses may necessitate compensatory amplifiers, thereby
increasing the complexity even further [15, 16]. Additionally, the implementation of
phased arrays is very expensive due to a large number of RF chains required as
shown in Fig. 1.1 (e) [17]. As a result the overall system complexity and cost
becomes prohibitively high which prevents their adoption in widespread applications.
Moreover, such a high complexity raises questions on system’s reliability. Therefore,
due to associated disadvantages, these high gain enabling technologies of the past
are not directly applicable to mm-wave bands without significantly compromising the
performance. Futuristic smart antenna systems demand inexpensive aesthetic agile
solutions which are capable of providing high gains, wide angle electronic beam steering,
and are scalable.
One possible solution to achieve very high gains as well as wide angle electronic
beamsteering functionality is the combination of reflector and phased array
technologies; called a reflectarray. Reflectarrays are a wonderful genetic hybrid of
reflector and phased array technologies, which benefit from spatial feeding of reflectors
and has planar structure as of phased arrays [18–20] as shown in Fig. 1.2 (a). A
reflectarray works on the principle that a constant phase is realized (through the
reflected field phase control by means of a large number of individual array elements
(unit cells) at the reflectarray aperture) in a plane orthogonal to the radiated main
beam direction. While planar in profile and aesthetic in appearance; a reflectarray
can simultaneously achieve high gains and wide angle beamsteering. Due to spatial
feeding of comprising array elements, a reflectarray eliminates complex and lossy feeding
network of phased arrays. Reflectarrays provide a platform for high performance
beamforming at the RF plane of a mm-wave antenna. When operating in transmit
1.1. Motivation 6
mode, the array elements (called unit cells) in a reflectarray control the phase of the
reflected field from their location which ultimately controls the far field. Similarly, in
receive mode the phase of incoming field is modified at each unit cell’s location. The
overall cumulative effect is observed at the feed horn. Therefore, the signal processing
required in smart antennas gets implemented implicitly directly at the RF plane of a
reflectarray antenna. This eliminates the need of any beamforming chains required in
phased arrays. Similarly, it eliminates signal processing for calculating beamforming
weights. One only need to configure the unit cells to produce the required reflection
phases based on the required beam pointing. Wide angle beamsteering is achieved by
controlling individual unit cells in a reflectarray.
Unit Cell
Feed Horn
1000s of Unit Cells
Printed Reflectarray Surface
(a) Reflectarray geometry (b) Folded reflectarray on a cube sat
Figure 1.2: Reflectarray antennas
Another attractive advantage of reflectarrays is their ability to be made foldable [21] to

get accommodated in small platforms and to conform the launcher’s fairing to largely
facilitate a satellite launch. The folded reflectarrays can be deployed while in orbit
to accomplish high gain requirements. A foldable reflectarray mounted on a 3U (1U
= 10 cm) cube sat [22, 23] is shown in Fig. 1.2 (b). This folding capability equip
designers to accomplish very large antenna reflecting area where required to achieve
the communication payload performance objectives. Modern reflectarrays benefit from
printed microstrip technology which makes them a low cost competing solution for
the new age of antennas. Due to a large number of unit cells in a reflectarray it is
less sensitive to manufacturing tolerances and can still achieve the performance objects
1.1. Motivation 7
significantly when few unit cells go wrong during the printed circuit board (PCB)
printing process. Due to mold free manufacturing, satellite antenna footprints can be
altered just before manufacturing [10] without any significant costs involved.
When the unit cells of a reflectarray are made reconfigurable a range of very versatile
functionalities can be achieved. This includes pencil beamsteering, multiple beams
generation, contoured beams, implementation of pattern synthesis techniques; all
achieved at the run time. Reflectarrays facilitate the implementation of beamforming
at RF plane of an antenna so greatly reduce the system losses and complexity of signal
processing. It also eliminates the requirement of a large number of RF chains therefore,
reducing the systems complexity and cost.
Major disadvantages of reflectarrays are their limited bandwidth [24–27] and low
aperture efficiency [28]. Although, in the past significant efforts have been made to
improve their bandwidth performance. Nevertheless, the bandwidth performance of
reflector antennas remained unbeaten. However, not all the bandwidth offered by
reflector antennas is practically being used in a particular application concurrently.
Therefore, for practical systems operating in a limited bandwidth the reflectarrays are
a significantly simplified solution. The gain reduction due to low aperture efficiency
can be compensated by slightly increasing the aperture size where applicable.
In spite of these disadvantages, reflectarrays are still considered as a potentially

compelling solution for next generation communication systems requiring high gain
and wide angle beamsteering capabilities. In the past reflectarrays are not widely
adopted due to a lack of analytical tools for their performance prediction. Similarly,
the reliability of reflectarrays for a long life satellite mission was questioned and people
were not comfortable in taking a risk due to no prior space heritage of reflectarrays.
In very recent years a significant progress in modeling, optimization, and testing of
fixed reflectarrays have been reported by TICRA, Denmark [29–31]. Similarly, a
folding reflectarray has been used on a small space mission [32, 33] for technology
demonstration. It is believed that reflectarrays would prove as an emerging technology
in the coming future. Therefore, this research in mainly focused at realizing high
gain wide angle beamsteering reflectarrays. In this research the terms, beamsteerable
1.2. Aims and Objectives of Research 8
reflectarrays and smart reflectarrays are used interchangeably.
1.2 Aims and Objectives of Research
This research shall investigate innovative design concepts of mm-wave smart antennas
and their measurement techniques. In particular, the printed reflectarray antenna;
which is planar in structure, compact in size, easy to fabricate and low cost, therefore
attractive for commercial applications. This research shall lead to the development
of new capabilities and skills in design approaches and measurement metrology of
mm-wave smart antennas. Keeping in view these broad objectives, this research has
the following objectives in particular.
O1: To critically review the existing smart antenna techniques with objectives of high
gain, wide angle electronic beamsteering, reliability, power consumption, ease
of manufacture, compactness, scalability, and cost to select the most suitable
technique.
O2: To comprehensively examine the available reconfiguration technologies and select

the most suitable for mm-waves.
O3: Characterization of the selected reconfiguration technology at mm-waves.
O4: To study the complexity versus performance trade-off for mm-wave smart
antennas.
O5: Design novel solutions suitable for mm-waves smart antennas.
O6: Implementation and measurements of novel designs to demonstrate the proposed

concepts.
O7: Development of methods and procedures to ease the design, performance analysis
and implementation of smart antennas.
O8: To propose suitable and cost effective measurement techniques for the constituents
of smart antennas.
1.3. Major Research Contributions 9
O9: To propose considerations to economize power consumption of the selected

technology.
O10: Finally, to lay foundations for further research and collaboration in the field of
smart reflectarray antennas.
1.3 Major Research Contributions
Research work presented in this thesis brings following contributions.
RC1: Formulation for the analysis and RLC model extraction of the most fault tolerant
passive reflectarray unit cells at mm-waves. Two new measurement techniques to
ease the alignment, orientation, and DC biasing issues of mm-wave unit cells.
RC2: Selection of PIN diodes based on a critical review of reconfiguration technologies

for mm-waves. PIN diodes’ characterization at X and V bands. Measurements
based extraction of their equivalent RLC models. Analysis and measurements
for their DC power consumption and RF insertion loss.
RC3: A novel design of three states mm-wave phase quantized unit cells. Reduction
in DC bias lines and a novel technique to optimize the cross polarization of unit
cells.
RC4: A fast algorithm to prepare the electromagnetic (EM) simulation model of large
reflectarrays.
RC5: Conception and measurement based validation of phase quantized reflecarrays

and their performance matrix.
RC6: Conception and measurement based analytical solution of low DC power

consuming smart reflectarrays.
1.4. Thesis Outline 10
1.4 Thesis Outline
The rest of the thesis is organized as follows:
Chapter 2 presents basic essential knowledge to comprehend the research work

presented in this thesis. A comprehensive and critical review of the available smart
antenna techniques and their underlying technologies is presented in this chapter. It is
justified why reflectarrays are the most optimum candidate to simultaneously achieve
very high gains and reconfiguration/smart functionalities at mm-waves. A comparison
criteria to choose the optimum technology to implement mm-wave smart reflectarrays is
discussed. Various ways to implement unit cells and their reconfiguration are discussed.
This chapters ends with the selection of suitable antenna topology and reconfiguration
technology for the smart antenna to be designed in this research.
Chapter 3 serves two purposes. First, the characterization of the selected unit cell is
presented. An optimally suitable selected unit cell is characterized using analytical
procedures and 3D EM simulations. The effects of surface waves, metal surface
roughness and metal finite conductivity are studies in this chapter. A parametric
analytical model was prepared to characterize the unit cells w.r.t. its geometrical and
material parameters. A lumped equivalent circuit model for unit cells was extracted
based on the analytical solution. This completes the unit cell selection and analytical
parts of the research contribution RC1. Second, two new measurement techniques
for unit cells were proposed as part of research contribution RC1. Unit cells were
fabricated and measurements were carried out. The results from 3D EM simulator,
analytical model, and circuit based equivalent model were compared with the test data.
Through a series of unit cells fabrication good confidence was achieved in analytical
procedure and the fabrication facility.
In the first part of Chapter 4 the selected reconfiguration technology (PIN diodes);
based on critical review of existing technologies in Chapter 2, was characterized at
X and V band for extracting the PIN diode parameters at both bands. Through
various series of measurements of multiple fabricated boards two different PIN diodes
were characterized. Similarly, through multiple measurements the power consumption
1.4. Thesis Outline 11
of PIN diodes and their RF loss was measured. A significant insight of power
consumption versus RF loss was achieved. This accomplishes research contribution
RC2. In its second part the selected unit cell was made reconfigurable. After trying
various techniques for reconfiguration the most optimum topology was selected for
reconfiguration. Multiple 3D EM simulations were performed to attain this goal. These
reconfigurable unit cells were fabricated in an external facility due to their multi-layered
structure. The unit cells were measured and a significant insight of their operation was
achieved. This completes research contribution RC3.
Design and analysis of reflectarrays is the focus of Chapter 5. Reflectarray analytical

solution is presented in this chapter. It is followed by efficiency predictions in
reference to reflectarray’s design variables. The procedure for building the reflectarray
EM simulation model is discussed here as research contribution RC4. The concept
of phased quantized reflectarrays is presented in this chapter. The performance
characterization of these phase quantized reflectarrays is explained. The design and
fabrication of various level of phase quantized reflectarrays at different pointing angles
is discussed in this chapter as the part of research contribution RC5. Based on
measured results of PIN diodes, the DC power consumption calculations and trade-off
with functionality are discussed in this chapter as part of RC6.
Chapter 6 presents the measurement of these phase quantized reflectarrays as part of

the research contribution RC5. Various measurements are presented here for these
arrays. Gain, pointing accuracy and sidelobes performance parameters are compared
with the simulated design parameters.
Chapter 7 concludes this thesis by summarizing major results achieved in this

research and highlights important lessons learned during the execution of this project.
Limitations of this research, and some possible directions for novel futuristic research
in mm-waves smart reflectarrays domain are presented in this chapter.
1.5. Publications 12
1.5 Publications
The work carried out in this thesis has resulted in following publications.
C1: Ghulam Ahmad, T. W. C. Brown, C. I. Underwood and T. H. Loh,

“How coarse is too coarse in electrically large reflectarray smart antennas?”,
2017 International Workshop on Electromagnetics: Applications and Student
Innovation Competition, London, 2017.
C2: Ghulam Ahmad, T. W. C. Brown, C. I. Underwood and Tian Hong Loh, “An
efficient algorithm for electrically large reflectarray antenna design automation”,
2017 International Workshop on Electromagnetics: Applications and Student
Innovation Competition, London, 2017.
C3: Ghulam Ahmad, Tian Hong Loh, T. W. C. Brown, and C. I. Underwood,
“On the phase selection of millimeter wave quantized reflectarray”, International
Applied Computational Electromagnetics Society Symposium, 01 Aug - 04 Aug
2017, National University of Singapore Suzhou Research Institute, Suzhou, China,
2017.
C4: Ghulam Ahmad, T. W. C. Brown, C. I. Underwood and Tian Hong Loh, “Role
of surface waves in the reflection properties of a millimetre wave reflectarray unit
cell”, LAPC IET, Loughborough, UK, 2017.
C5: Ghulam Ahmad, T. W. C. Brown, C. I. Underwood and Tian Hong Loh,
“Millimetre wave reflectarray antenna unit cell measurements”, LAPC IET,
Loughborough, UK, 2017.
“Millimetre wave high gain reflectarray smart antennas”, 5th IET Colloquium
on Antennas, Wireless and Electromagnetics. London, 2017.
C7: Ghulam Ahmad, T. W. C. Brown, C. I. Underwood and Tian Hong
Loh, “Millimetre wave high gain smart antenna”, Festival of Radio Science
(FRSci-2015), International Union of Radio Science (URSI), Manchester, UK,
2015.
“High gain millimetre wave antenna solution for next generation communication
systems”, Post Graduate Research Conference, University of Surrey, 2016.
“Implementation perspectives of millimetre wave high gain antennas”, Doctoral
College Conference, University of Surrey, 2017.
J1: Ghulam Ahmad, T. W. C. Brown, and C. I. Underwood, Tian Hong Loh, “An
investigation of millimeter wave reflectarrays for small satellite platforms”, Acta
Astronautica, 2018.
Chapter 2
Background to Millimeter Wave

Beamsteering and Reflectarrays
Millimeter waves offer large contiguous bandwidths to enable high capacity links for
terrestrial and space communication applications. When applied in radar applications
mm-waves enable high resolution of target detection. These benefits come at the cost
of an increased path loss at these frequencies. To overcome this path loss one need to
have high gain antennas to benefit from the attractive features of mm-waves. When
antenna have high gains, their beamwidths become narrow. At mm-waves the links are
mainly line of sight. The performance of a link using narrow beam antennas degrade
when beam is not pointed at the target. This can happen due to shadowing effects as
well as in nomadic applications. Therefore, in addition to high gains, one need to have
tracking antennas to provide a ubiquitous coverage without service interruptions. This
chapter is about the background of high gain tracking antennas and their beamforming
principles.
This chapter presents the background knowledge essential to understand the research
work presented in this thesis. It starts by developing an understanding about antenna
terminologies, classification and operation of smart antennas. Then a critical review of
the state-of-the-art in reflectarrays and their implementation technologies is presented.
Various challenges faced in a mm-wave reflectarray smart antenna implementation in
reference to the research objectives are identified. Finally the chapter concludes with
13
2.1. Antenna Fundamentals 14
the selection of a suitable topology and optimum reconfiguration technology for the
implementation of a high gain smart reflectarray antenna at mm-waves.
2.1 Antenna Fundamentals
An antenna is a piece of hardware that acts as a transducer to interface RF electronics

with the free space for the purpose of EM energy exchange. Antennas exist in a variety
of shapes and can be as simple as a mono-pole or as sophisticated as a high speed
tracking antenna. Majorly, antennas behave as reciprocal devices which means they
exhibit the same properties while working in transmit or receive mode. Important
characteristics of an antenna are their radiation pattern, directivity, gain, beamwidth,
polarization, bandwidth, and return loss. In this research the antenna boresight; the
direction of main radiations, is aligned with the z axis, while xz plane is the E-plane,
and yz plane is the H-plane. For details on the antenna coordinate system one can refer
to [34, 35]. The range of a communication link can be determined using [36, 37] with
free space path loss given by [38]. For a detailed description of scattering parameters
and their de-embedding one can refer to [39].
2.2 Millimeter Wave Spectrum and Implications
Millimeter wave spectrum as shown in Fig. 2.1, until recently remained less attractive
due to its associated propagation characteristics and lack of mm-wave technologies.
Millimeter waves suffer higher free space path loss, rain attenuation, and penetration
losses. At certain bands oxygen and water vapor absorption plays a dominant
role. Moreover, mm-waves signals are significantly effected by blockage, have reduced
diffraction, and weak reflections. To maintain a communication link mm-waves require
strict line of sight availability. Similarly, to manufacture and accurately characterize
mm-wave components is highly ambitious. In the past due to less commercial
applications in this spectrum the mm-wave technology is still in its developing stage.
Recently the great potential of mm-waves spectrum is realized and their propagation
impairments are going to be exploited. Millimeter waves can permit more densely
2.2. Millimeter Wave Spectrum and Implications 15
packed short range communications links. Therefore, they enable very efficient
spectrum utilization by frequency reuse. Additionally, due to higher associated
losses, mm-waves can increase security of communication transmissions by limiting the
distribution to a specific area. Millimeter waves spectrum provides continuous large
bandwidths to support emerging applications. World radio-communication conference
(WRC-15) has supported 24.25 - 27.5 GHz (called 26 GHz pioneer mm-wave band)
for next generation communication [40]. Various distinct bands in the range of 31.8 -
86 GHz are one of the agenda for WRC-19 to consider their application for earth to
space and vice versa communications. Despite significant efforts the next generation
communication systems are still in the standardization stages.
Wavelength
10 cm 1 cm 1 mm
3 GHz 30 GHz 300 GHz
Frequency
Figure 2.1: Millimeter wave spectrum. Currently under consideration mm-wave bands
are indicated in green color.
Table 2.1: Unlicensed mm-wave band and maximum transmit power in various regions
of the world
Region Bandwidth Tx Power EIRP Antenna Gain

USA 7 GHz (57 - 64) 500 mW 43 dBm -
Canada 7 GHz (57 - 64) 500 mW 43 dBm -
Japan 7 GHz (59 - 66) 10 mW - 47 dBi
Australia 3.5 GHz (59.4 - 62.9) 10 mW 51.76 dBm -
Europe 9 GHz (57 - 66) 20 mW 57 dBm 37 dBi
A license free band around 60 GHz (57 - 66 GHz) is a keen interest of research to enable
unprecedented high data rates. Table 2.1 lists the unlicensed bandwidth with maximum
permitted transmit power and EIRP for various regions of the world [1]. In addition
2.3. Operation of Antenna Arrays 16
to stringent maximum transmit power limit there are significant associated losses at
mm-waves. The atmospheric absorption can reach upto 18 dB/km at frequency (f )
= 60 GHz. Similarly, the rain attenuation at a moderate rain rate of 25 mm/h is 10
dB/km [41, 42].
At 60 GHz, the free space path loss for a distance (R) of 1 km is 128 dB using (2.1) [38].

LdB = 32.4 + 20 log10 f (M Hz) + 20 log10 R (km) (2.1)
A possible countermeasure to overcome higher signal losses in scenarios of upper

bounded transmit power is to increase the antenna gain. Free space path loss
= (4πR/λ)2 is ∝ to f 2 regardless of the distance between a transmitter and its receiver.
When antenna electrical size (in terms of λ) is kept constant; with an increase in
frequency the effective aperture scales with λ2 /4π therefore, antenna physical aperture
becomes smaller at higher frequencies. However, here is an interesting fact about
mm-waves and antennas. If the antenna physical aperture is held constant at one side
of the link as the frequency increases, the free space path loss remains unchanged.
Moreover, if physical apertures of antennas at both sides of a communication link are
held constant, then the free space path loss actually diminishes with f 2 and can be
exploited as power gain to counter the higher noise floor associated with a broader
signal bandwidth at mm-wave frequencies.
Due to electrical design reasons it is desirable to preserve the electrical size of an

individual antenna therefore, maintaining its physical aperture simultaneously is made
possible by utilizing aggregated individual antenna apertures to form arrays. An
individual antenna element shrinks with frequency, therefore more elements can be
accommodated in the same fixed original aperture to achieve high gains to overcome
the associated losses at mm-waves.
2.3 Operation of Antenna Arrays
At mm-waves a single antenna element do not provide enough gain therefore, one need
to use antenna arrays. Due to multiple antenna elements in an array, the overall
radiation pattern is a multiplication of single antenna element’s radiation pattern and
2.3. Operation of Antenna Arrays 17
the array factor. In Fig. 2.2 a uniform linear array with element spacing d is shown.
In far field the incoming wave’s wavefront is planar. This planar wavefront makes an
angle θ with array boresight. Therefore, incoming wave travels an additional distance
of d sin θ to reach to each successive antenna element. In free space this would result
in a time delay of ∆ t = d sin θ /c. When expressed in terms of phase shift it becomes
∆ φ = 2 π f ∆ t = 2 π d sin θ /λ. Therefore, in an n element array, this results in a phase
distribution of: nd
∆ φn = 2 π f sin θ (2.2)
c
To state in another way, if the phase distribution of (2.2) is implemented in a uniform
n element array with element spacing d, then the beam can be steered at an angle θ at
frequency f .
W
a
inc vefr
θ om ont
ing o
wa f
ve
θ
in
ds
Antenna element
d
Figure 2.2: Wavefront and array phase shift
The array factor of a one dimensional array (AF1D ) discussed above is given by [35]:
N
X
AF1D = an e j (n−1)( k d sin θ + β) (2.3)
n=1
where an is the excitation amplitude of the nth element, k is the wave number, and β is
the progressive phase of elements with spacing d. When beam scanning is required in
two dimensions, one can benefit from two dimensional array. A two dimensional array
whose elements are aligned in x and y axis with inter element spacing of dx and dy
respectively has the array factor given by [35]:
N
X M
X
j (m−1)( k dx sin θ cos φ + βx )
AF2D = In Im e e j (n−1)( k dy sin θ sin φ + βy ) (2.4)
n=1 m=1
where βx and βy are the progressive phases in x and y directions, and excitation
amplitudes are represented by Im and In respectively.
2.4. Smart Antennas 18
In most cases the phase distribution of the array elements is controlled to synthesize a
desired array factor to point the radiated beam in a desired direction. Antenna arrays
possess directional capabilities making antennas capable of focusing radiated energy in
specific directions. However, to configure these arrays to productively transmit/receive
is a challenging task. Although, high antenna gains can overcome the propagation
losses however, the blockage and line of sight requirements need further solutions. To
conquer the issues arising due to signal blockages and strict line of sight requirements
of mm-wave signals, these high gain antennas are required to be made steerable.
2.4 Smart Antennas
An antenna which is capable of adjusting its radiation characteristics e.g. frequency,

radiation pattern, or polarization etc. in accordance with its surrounding RF
environment is known as a smart antenna. For high throughput in terms of data
transfer rates and ubiquitous coverage it is envisioned that high gain smart antennas
would be a primary constituent of mm-wave communication systems. Currently,
mm-wave bands are under heavy investigations for the realization of these systems.
A smart antenna system consists of multiple Antenna elements with Smart signal
processing capabilities which enables the system to respond to the surrounding
radio/electromagnetic environment. It is achieved by modifying certain characteristics
(radiation pattern/individual element weights). Generally smart antennas can improve
a specified performance metric in terms of increased communication link signal to noise
ratio (SNR) and/or system’s immunity to the interference. Usually, the operating
frequency and polarization are fixed in communication links, therefore radiation pattern
characteristics are the key interest in mm-wave smart antennas. Millimeter wave smart
antennas are expected to demonstrate their beamsteering to wider angles without a
significant reduction in their gains.
There exist multiple ways to classify smart antennas as shown in Fig. 2.3. This is mainly
due to the effective means for beamforming, signal processing, and usage of techniques
for the performance enhancement. Here only the brief details of beamforming methods
in smart antenna is presented. Interested reader is referred to [43] for more details on
diversity and MIMO techniques.
Smart Antennas
Technique Diversity Beamforming MIMO
Beamforming Switched Beam Adaptive Beam
Signal Spatial Space Time Space Time

Processing Processing Processing Detection
Figure 2.3: Smart antenna classification.
2.4.1 Beamforming Techniques in Smart Antennas
Based on beamforming techniques, smart antennas can be classified as switched or

adaptive beam as discussed in the following sections.
2.4.1.1 Switched Beam Smart Antennas A switched beam smart antenna

system forms spatially selective finite predetermined fixed beams. Any of these beams
can be switched in based on requirements [44]. These antennas are simple to implement,
and are low cost in comparison to adaptive beam antennas. A switched beam antenna
with pre-fixed beams may not be optimally pointing towards the user [45]. As a user
moves away from the beam center and crosses over the beam periphery, the received
signal becomes weaker. This causes an intra-cell (within a cell) handover and a switched
beam antenna switches from one beam to another as shown in Fig. 2.4 a. Due to spatial
selectivity, the interference is reduced as compared to simple sectored antennas. A
switched beam antenna neither distinguish between a desired signal and an interference
nor reconfigure itself to place nulls in the direction of an interferer. Therefore, if an

interference signal is approximately at the beam center then it becomes an enhanced
additive noise and the intended user quality gets degraded. A switched beam antenna
can serve a single user or multiple users at a time by switching in multiple beams at
different spatial locations.
Interference 2
Interference 1
(b) Adaptive beam with

(a) Switched beam interferences (c) Multiuser adaptive beam
Figure 2.4: Switched and adaptive beam smart antennas
Phased arrays or multibeam antennas are categorized as switched beam antenna [45].
This is due the fact that such configurations consist of either a number of fixed beams
(with one beam turned on towards the desired signal), or a single beam formed by
phase only adjustment (which is steered towards the desired signal). In a generalized
case of switched beam antennas (e.g. dynamic phased arrays, where received power is
maximized), a direction of arrival (DoA) algorithm is embedded in the system. Different
parameters are tuned in order to track an intended roaming user within the antenna
beam range [44] by means of beamsteering. Therefore, when an intra-cell handover
occurs, the intended users signal is received with an optimal gain. Dynamic phased
arrays is a step towards adaptive arrays however, not truly adaptive in the strictest
sense.
2.4.1.2 Adaptive Beam Smart Antennas An adaptive antenna system [46] can
utilize the additional dimension of space to adapt to a changing RF environment or
spatial origins of the signals. The antenna main beam is directed towards an intended
signal while interferers are suppressed by adapting the radiation pattern in real time.
An adaptive antenna system can customize an appropriate radiation pattern for each
individual user as indicated by Fig. 2.4 b. In case of multiuser support, multiple
radiation beams are adapted in a way that each beams peak is pointing towards an
intended user and its nulls or minimum gain points are towards the other user/interferer
as shown in Fig. 2.4 c. This is maintained by constantly adapting to the changes for
roaming users in the coverage area.
Adaptive antenna systems are capable to locate and track signals including intended
users as well as interferers. These systems dynamically adjust the antenna pattern to
enhance an intended user reception while minimizing the interference. It is achieved
using sophisticated digital signal processing (DSP) algorithms. A functional block
diagram of adaptive array beamforming in smart antennas is shown in Fig. 2.5 [44].
Received signals at each antenna are passed through individual receive chains, down
converted to base band signals and are digitized. A DoA algorithm computes the
direction of arrival of all signals by computing time delays among antenna elements.
Afterwards an adaptive algorithm; using a cost function, computes the appropriate
weights that result in an optimum radiation pattern. This is done continuously to
track the signal of interest and interferer by dynamically changing complex weights.
Various well known DoA algorithms [47–49] available are MUSIC [50], ESPRIT [51–
55], or SAGE [56–58] etc. A detailed investigation of DoA algorithms is beyond the
scope of this research. A continuous adaptation of the radiation pattern (adaptive
beamforming), has two main strategies [43].
1. The first technique is based on an assumption that the desired signal is already
known partially through a training sequence. This known signal is compared with
the received signal to adjust weights to minimize the mean square error between
them. This approach optimizes the signal to interference ratio and is applicable to
non-line of sight (nLoS) environments. Since the weights are updated according
to incoming signals, this reduces interference as well as mitigates the multipath
fading.
2. The second technique is based on calculating DoA for all incoming signals, and
then complex weights are adjusted to produce a maximum towards the desired
angle and a null towards the interferer. This technique may fail in cases where
there are too many DoAs due to multipaths especially in non-line of sight cases
where wider spread of angle of arrival is expected due to many local scatterers.
W1
DSP
Rcvr,wD/C ADC
Antennawarray
W2 Array
∑
output
Rcvr,wD/C ADC
WN
Rcvr,wD/C ADC
Rcvrw=wReceiver
D/Cw=wDownwConversionwtowbaseband Adaptivew
ADCw=wAnalogwtowDigitalwConversion Algorithm
WNw=wComplexwWeights
DoAw=wDirectionwofwArrival
DSPw=wDigitalwSignalwProcessing
DoA
Figure 2.5: Adaptive beamforming in smart antennas.
Adaptive smart antennas are divided into three categories based on the involvement of
spatial and temporal processing of the received signals [59] which are spatial processing,
space time processing, and space time detection.
1. In spatial processing/filtering the received signal of each separate receive chain

is multiplied with a complex weight. These weighted signals of all the chains are
summed up and the resulting output is processed like any signal from a normal
antenna. Space only processing works best if each antenna shows the same time
dispersion. However, in reality there exist multipaths and temporal dispersion of
received signals which degrades the performance of space only processing.
2. Space time processing method combines received signals from individual chains
sampled in time and space in order to incorporate signal components that
have been temporally dispersed. Spatial components from individual chains are
multiplied by complex spatial weights similar to space only processing (spatial
equalization). In addition to spatial equalization time equalizers are employed to
realign delayed copies of the intended signal to overcome the effects of multipaths.
A typical simplified space time processing structure is shown in Fig. 2.6. Here it
is assumed that the down conversion and analog to digital conversion etc. step
has taken place after the antenna. Here spatial filtering is achieved using antenna
arrays while temporal filtering is obtained by using finite impulse response filters
at the output of each antenna element. To form a temporal filter, each antenna
is followed by K taps where δ indicates the time delay between successive taps.
A combination of several antenna elements each followed by temporal taps form
the space time equalizer. Space time weights are represented by wmk , where
1 ≤ k ≤ K is temporal index and 1 ≤ m ≤ M is the spatial/antenna element
index. If M receiver chains, and time equalizer of length K are utilized, then total
structure has M spatial and K temporal complex weights leading to a complexity
of M × K. If an antenna element output is represented by ym (t), then the array
output after space time processing is given by:
M X
X K
Array output = ym (t − k δ) wmk (2.5)
m=1 k=1
When K = 1, the spatial processing is realized, whereas setting M = 1 realizes

temporal filtering. In space time equalization, spatial and temporal weight vectors
can be calculated jointly. This joint space time processing exploits both spatial
and temporal characteristics of the incoming signal in a single complex filter.
Although this provides processing gain however, it is computationally extensive.
In sequential space time equalization, several temporal equalizers followed by a
single spatial beamformer realize multipath and interference in two steps therefore
resulting in a reduced computational complexity. The array output signal is
finally fed to a decision device to recover the bit stream.
3. In the most complex form of adaptive smart antennas, the detection of bit
streams can be done jointly with space time equalization. It offers the best
performance however, has the highest degree of complexity as compared to rest
of the techniques.
Antenna 1
Spatial processing
Temporal processing
δ δ δ
w11 w12 w13 w1k
Antenna m
Array
δ δ δ ∑ output
wm1 wm2 wm3 wmk
Figure 2.6: Spatial-temporal processing in adaptive arrays
At mm-waves line of sight link mainly dominates the communication whereas, multipath
reflected signals are of very low magnitude in comparison. Therefore, there would be
a less need for temporal equalization and beamforming would mainly benefit from
spatial equalization. Although, adaptive smart antenna systems are theoretically very
attractive due to their excellent characteristics however, the practical realization of
these systems is highly infeasible as would be clear from beamformer implementation
complexities.
2.4.2 Beamformer Implementation
Beamforming is a very powerful technique employed in communication systems for

SNR enhancement, multipath mitigation, and interference rejection. Due to shorter
wavelength at mm-waves, a large number of antennas can be accommodated in a
small volume therefore, enabling large gains through beamforming. When employed
in transmitter and/or receiver it can significantly improve the signal strength and
suppress the interference, which results is much higher throughput as compared to
simple sectorization techniques currently being used in mobile networks. Where in the
communication chain, a beamformer should be implemented and whether it should
be digital or analog are the key questions which dictate certain complexity issues.
Beamforming can be implemented at various levels in the chain. It can be digital or
analog in form. Its implemented can be at baseband or directly at RF level. Salient
advantages and drawbacks of each are discussed below.
DAC HPA HPA
DAC
DAC HPA HPA
(a)vDigitalvbasebandvbeamforming (b)vAnalogvbasebandvbeamformingv
Antenna HPA
Beamformingv DAC
Upconverter weights
Base
HPA
bandv
pre
HPA cod-
HPA
ing
DAC
DAC
HPA HPA
(c)vRFvbeamforming (d)vHybridvbeamforming
Figure 2.7: Beamforming implementation techniques

1. Digital baseband beamforming: In digital baseband beamforming as

shown in Fig. 2.7 (a) the beamforming weights are applied is digital domain
before the digital to analog conversion (DAC). Individual data streams are
upconverted, amplified by HPA and transmitted through an antenna. Through
this architecture, it is possible to simultaneously transmit multiple data streams.
Therefore, enabling space division multiple access (SDMA) and multiple input
multiple output (MIMO) transmission schemes at the cost of multiple RF chains.
However, the operation of amplitude and phase correction has to be performed on
each subcarrier. Normally this is accomplished in digital beamforming chipsets
e.g. Broadcom, Marvell, and Mediatek chipsets [60]. Recently IMEC has
announced a low power beamforming chipset for IEEE 802.11ad for 60 GHz
operation [61]. Currently, these chipsets are based on orthogonal frequency
division multiplexing (OFDM). The adoption of waveforms other than OFDM
is under consideration in future networks which would required new chipsets.
Theoretically, digital beamforming can have as many data streams as the number
of RF chains therefore, is capable of supporting extremely high data rates.
However, due to low gain of an individual antenna in each RF chain, the
performance would be limited. Additionally, to implement a large number of
RF chains becomes practically difficult and expensive. At the same time a high
number of power hungry DACs for a large number of chains would demand high
operating power and increase the system complexity.
2. Analog baseband and RF beamforming: Analog baseband and analog

RF beamforming implementations are shown in Fig. 2.7( b & c). respectively.
Here, the beamforming weights are applied to the same signal (usually a single
data stream) which is transmitted through multiple antennas. Beamforming
weights are applied prior to upconversion at an intermediate frequency is case
of analog baseband beamforming. Due to low frequency operation it provides
a better control of signals and beamforming losses can be compensated by
proving additional gains at the intermediate frequency. However, there is a lot of
hardware involved in analog baseband implementation. Whereas, in case of RF
beamforming the beamforming weights are applied after unconversion. There are
few hardware units involved in direct at RF implementation. However, working at

RF frequencies make designs very challenging and often introduce significant RF
losses which are compensated by HPA power and/or the antenna gain. To design
low loss and stable phase-amplitude control RF hardware at these frequencies is
a significant challenge. Both of these techniques significantly reduce the number
of RF chains. However, implementation for multiple data streams is extremely
challenging. Therefore, multiplexing capabilities are almost absent in case of
analog beamforming.
3. Hybrid beamforming: Due to the fact that next generation antenna

technologies will be deploying massive antenna structures to ensure the coverage
and capacity demands [62]. Beamforming in analog or digital alone would not
provide the optimum performance. Therefore, a hybrid approach; where analog
and digital beamforming work jointly, would result in an optimum performance.
In a hybrid architecture, analog beamforming (using phase shifters) compensate
for the large path loss and digital beamforming (pre-coding) provides necessary
flexibility to accomplish the advanced multi-antenna techniques like multiuser
MIMO [62, 63]. A simplified hybrid architecture is shown in Fig. 2.7 (d) where
multiple independent data streams are manipulated by digital base band part of
beamforming, whereas the RF beamforming compensates for propagation losses.
Particularly at mm-waves where one would need thousands of antenna elements in the
array to achieve high gains. It is computationally extremely challenging to compute
thousands of beamforming weights in a time efficient manner. Simultaneously, to
implement such a large number of RF chains is practically impossible due to real estate
problems and associated losses at mm-waves. Therefore, the complexity and cost of
such systems remains a major hindrance in their widespread adoption.
Customarily, RF beamforming make use of phase shifters and amplifiers/attenuators

to adjust the complex weights of antenna elements in an array. In next section the
performance of various phase shifter technologies is presented.
2.4.3 Phase Shifters
Phase shifters provide signal’s phase control required for beamforming/beamsteering

in electronically steerable antenna arrays. Various types and implementations of phase
shifters are available. Active phase shifters provide gain and are non-reciprocal, while
passive phase shifters introduce insertion loss and are usually reciprocal. Reciprocal
phase shifters can be used as transmit and/or receive. However, this is not the case for
non-reciprocal phase shifters; where separate phase shifters are required for transmit
and receive chains. Analog phase shifters provide continuous variable phase shift or
time delay. While digital phase shifters provide a discrete set of phase shifts or time
delays. The phase shifter may or may not be based on semiconductor technology. For
example, non-semiconductor passive phase shifters are based on ferrites, ferro-electrics,
piezoelectric, RF MEMS, liquid crystals, phase change materials, and graphene. While
semi-conductor technology based phase shifters can benefit from BJT, FET, MMIC,
RFIC, or Optical ICs to implement active phase shifters or can make use of PIN
and Varactor diodes for passive implementations. Phase shifters can be classified as
loaded line, reflection type, and switched line phase shifters based on their design. Key
parameters to characterize a phase shifter are gain/loss, linearity, noise figure, power
handling, switching time, reliability, number of phase states, DC biasing, DC power
consumption, and size etc. Table 2.2 presents the insertion loss for various types of
phase shifters. In this table and throughout the thesis, the number of phase shifter bits
are related to achievable phase states by: number of states = round(2(number of bits) ),
where the function round() provides the nearest integer, e.g. round(21.5 ) = 3.
In general, phase shifters have a significant amount of insertion loss which is unavoidable
as long as they exist in the chain. Vanadium dioxide based solutions are temperature
sensitive which makes them not suitable for space applications where temperature
varies significantly. Elastomer actuation based phase shifters although have lower
insertion loss, however require very high operating voltage which is not desired for
space applications. The minimal loss is for waveguide based phase shifters, which
is a manual phase shift solution, is very expensive and bulky. Its manual operation
and cost prohibits its use in electronic beamsteering. To implement only few phase
shifters may not be a problem. However, at mm-waves one need to implement

in the order of thousands which complicate the phase shifter based beamforming
solutions. Additionally, these phase shifter solutions are physically big and may not
get accommodated behind the tiny mm-wave individual antenna element therefore,
additional connectivity losses gets accrued when phase shifters are placed far from the
antenna elements. The best way to get rid of issues with phase shifters (insertion loss,
complexity, and accommodation etc.) is not to have them at all in the communication
chain. Therefore, some other novel means of effectively achieving higher antenna gains
instead of phase shifter based beamforming are highly desirable.
Table 2.2: Insertion loss for various phase shifter technologies
Technology Type Frequency Ins. Loss Comment Ref.

Phase change Switched 50 GHz very 2 - 4 dB (for Ultra low [64]
material line 2 bit wideband 90 to 360 noise,
(Vanadium deg) frequency
dioxide) independent
MEMS Switched 17.25 GHz 5.4 dB Very lossy [65]
line 5 bit even at Ku
band
MEMS Switched 2.23 GHz 1 - 2.5 dB Lossy even at [66]
line 4 bit low frequency
Graphene Switched 1 - 2 THz 0.8 - 4 dB Graphene for [67]
line 4 bit, (90 to 360 THz
Loaded line deg) frequencies
CMOS Reflection 60 GHz 5.6 dB Lossy [68]
type
CMOS Reflection 60 GHz 6.25 dB Lossy [69]
type
CMOS Active 4 bit 43 GHz 5 dB CMOS chip [70]
based
MEMS High 75 GHz 3 to 7.5 dB Complex [71]
tuneable impedance design
surface
Active IF - 71 GHz 6 dB Gain Multiple RF [72]
baseband chains
Dielectric - 35 GHz < 2 dB 1.4 KV for [73]
Elastomer Electrostatic
Actuation field
Waveguide - 60 GHz 0.3 dB manual, 1 [74]
based item = £2030
2.5. Available Smart Antenna Solutions 30
2.5 Available Smart Antenna Solutions
This section provides a comprehensive comparison of various smart antennas available

in the literature. Table 2.3 lists various parameters of the majority of smart antennas
available. Only the first two; namely Ruckus and Adant, are available commercially
whereas, rest of them are only research based. Majority of them operate at 2.4 GHz
to provide Wi-Fi services. Switching devices; PINs and MEMS, are the main source to
achieve reconfiguration. Few of these antennas are implemented using meta materials
in addition to reactance control devices. The main focus here is to comprehend the
available techniques rather than their operational frequency band or applications.
Table 2.3: Smart antennas, a comparison
Antenna Topology Freq. Gain Beam Comment Ref.

Ruckus Beam 2.4, 5 3 - 9 dBi 360◦ Az Linear pol.
R-7962 switched, GHz diversity, Max. [75–
printed yagi, ratio combiner 77]
PIN diodes
Adant Leaky wave 2.4, 5 6 dBi 180◦ Az Linear pol,
antenna meta material, GHz surface current [78–
PIN, Varactor, density control 81]
MEMS
Aruba Omni antennas 2.4 GHz <5 dBi - Tx [82]
AP-130 beamforming
chipset
Cisco Omni antennas 2.4 GHz <5 dBi - MIMO, not [83]
Aironet fully
Lap implemented
1142N
Substrate Beam switched 60 GHz 13 dBi 66◦ El, 8 sectors of [84,
integrated 360◦ Az 22.5◦ each in 85]
waveguide Az, linear pol.
antenna
SIW Beam 60 GHz 10 - 13 ±30◦ IF baseband [86]
antenna switched, slot dBi control, linear
antenna with pol
butler matrix
ESPAR Parasitic 2.54 10 dBi 0◦ , 180◦ , linear pol.,
, reactance GHz ±90◦ selectable four [87–
inverted control, PIN, beams 90]
F Varactor
Continued on next page ...
2.5. Available Smart Antenna Solutions 31
Continuation of Table 2.3

Antenna Topology Freq. Gain Beam Comment Ref.
Leaky Meta material 3.34 >0 dBi −50◦ to Linear pol,
wave and varactors GHz 24◦ poor Z match [91–
antenna 94]
Radial switched beam, 2.4 GHz 10 dBi 0◦ , 180◦ , four selectable [95]
line slot microstrip ±90◦ beams
antenna slots, PIN
diodes
Electronic Switched 2.4 6.1 dBi ±50◦ 10◦ steps, [96]
phase beam, phased GHz, circular pol.
array array, PIN
diodes
Compou- Beam 2.4 GHz 4 dBi ±30◦ linear, circular
nd switched, pol, freq tune [97–
reconfig. parasitic pixel 101]
array surface, PIN
diodes
None of these antennas is an adaptive smart antenna. Similarly, neither of these can
provide continuous beam scanning. All of these antennas are switched beam concepts
implemented in one way or the other. Moreover, these antennas are very low gain
antennas. All of these implementations are not a suitable candidate for mm-wave high
gain smart antennas. This is due to the fact that at mm-waves to attain high gains one
would need to use thousands of antenna elements. If implemented via these existing
techniques, the system gets too complex to be manageable
As was discussed in section 1.1 that mm-waves need high gains and wide angle
beamsteering capabilities simultaneously. The high gain solution of the past including
reflector antennas and phased arrays are not a suitable solution for mm-waves due
to their associated disadvantages of functionality, complexity, reliability, and cost.
Therefore, next generation high gain smart antennas need novel solutions. One such
possible solution is the hybrid of reflector and phased array technologies to enable
high gain and wide angle beamsteering capabilities simultaneously in the form of
reflectarrays. Various advantages of reflectarrays were discussed in section 1.1, therefore
are not repeated here.
2.6. Reflectarrays and Their Operation 32
2.6 Reflectarrays and Their Operation
The concept of reflectarrays was conceived in early 1960s by utilizing short ended
waveguide elements [102] to demonstrate co-phased far field radiations. This idea
of reflectarrays did not receive significant attention due to large and bulky hardware
operating at relatively lower microwave frequencies until late 1980s. Microstrip elements
were for the first time introduced in reflectarrays by [103, 104] in the year 1978. In the
same year, the first analysis effort was made by employing an infinite array approach
[105, 106]. Research in the field of reflectarrays and their advantages largely remained
ignored until the time when planar microstrip technology became fairly mature. In
late 1980s and early 1990s, a need for lighter, inexpensive, compact, and low profile
antennas revived the concept of reflectarrays, and as a result the printed reflectarray
technology was significantly developed [25]. Several techniques to achieve localized
reflection phase control of reflectarray elements were developed. These include variable
patch size [107], addition of stubs to reflecting element [108], element rotation [109],
and use of miniaturized motors for element rotation [24]. In the year 1996, two early
mm-waves reflectarrays were introduced. The first design demonstrated wide angle
electronic beam scanning (±45◦ ) at 94 GHz using one-bit PIN diode phase shifters in
a multilayered wafer printed monolithic technology [110], while the second achieved
beam scanning capability of ±25◦ using three bit ferrite phase shifters at 35 GHz [111].
In the 21st century, the availability of commercial full wave electromagnetic solvers led
to further developments in an accurate modeling and characterization of reflectarray
unit cells. Recent developments in reflectarrays are focused at realizing unit cells which
can integrate reconfiguration techniques to make versatile functionalities of reflectarrays
possible. These include dynamic reconfiguration of: operating frequency, polarization,
radiation pattern, contoured beams, multiple pencil beams, and beam scanning [18].
Due to less applications at mm-wave bands until very recently, the reflectarray research
was mainly limited to relatively low frequency bands e.g. C, X, Ku, and Ka bands
[112, 113] with only few developments reported at mm-wave bands [27, 114].
The geometry of a basic reflectarray [18, 115, 116] is shown in Fig. 2.8. Reflectarrays
have mainly planar reflecting surfaces, although they can also be made slightly curved.
2.6. Reflectarrays and Their Operation 33
ane
onal Pl
rthog
O ( 0, 0, F)
Y
Z
X
Figure 2.8: Printed reflectarray antenna on a grounded substrate
In its basic form, a reflectarray surface is designed with an objective to collimate the
scattered field in its far field. The number of elements / unit cells in a reflectarray is
determined by the required gain. Usually, a reflectarray is composed of thousands of
unit cells. A feed source antenna having its phase center coinciding with array focal
point (0, 0, F) spatially illuminates these unit cells. Radiated field from feed is scattered
by these unit cells. At the location of each unit cell a phase correction is applied to
the scattered field. Therefore, the parabola effect is artificially engineered by scattered
field phase correction from the entire reflectarray surface. The phase of the reflected
field in a plane orthogonal to the direction of radiated beam is constant as follows [18]:
~ mn · r̂b ) − 4φmn = 2πN

k0 (rmn − R (2.6)
where k0 = 2π/λ is the free space wave number for wavelength λ, ~rmn is the position
~ mn is the position vector
vector of mnth unit cell w.r.t. array focal point (0, 0, F), R
of mnth unit cell relative to the array center/origin of coordinate system (0, 0, 0),
r̂b is the direction of radiated pencil beam, F represents array focal length, and N
is an integer. Each mnth unit cell introduces a phase shift equal to 4φmn between
the incident and scattered field. By controlling this localized phase at each unit cell
location, the pencil beam can be scanned or a contoured beam can be synthesized. In
a well-designed reflectarray, it is mainly the unit cell behavior which determines its
overall characteristics. Therefore, significant design efforts are focused to develop an
optimum unit cell.
2.7. Reflectarray Unit Cell Characteristics 34
2.7 Reflectarray Unit Cell Characteristics
This section highlights the role of unit cell properties in a reflectarray design and
provides guidelines for the unit cell selection. One of the most important steps in a
reflectarray design is to realize suitable unit cells. Unit cell characteristics are the key
determinant of the reflectarray’s overall performance. Unit cells can be passive or active.
An active unit cell can integrate amplifiers for amplitude enhancement of the reflected
field. However, to implement these active unit cells is very complicated. A wide
variety of reflectarray implementations rely on passive unit cells. A unit cell essentially
performs the key function of phase corrected incident field reflection as dictated by far
field requirements. Unit cells introduce insertion loss in the reflected field. To minimize
this insertion loss in a unit cell, the reflected field’s magnitude is desired to be as close
as possible to that of the incident field. A selected unit cell topology should provide
flexibility to synthesize a large phase range to accommodate required reflection phases
from a certain geometric distribution of unit cells in a reflectarray [25]. To achieve a
wide operational frequency bandwidth the unit cell reflection phases are required to
be as linear as possible with the frequency [117]. A unit cell structure is designed by
taking the polarization(s) of intended incident wave(s) into account. It should reflect
the incident wave of a particular polarization without deteriorating its polarization
purity.
In modern reflectarrays, the reconfiguration trend is rapidly increasing. A unit cell has
to facilitate the reconfiguration. Geometry of a unit cell may vary depending on the
reconfiguration property required e.g. polarization, frequency and bandwidth, radiation
pattern beamwidth, and scanning in Elevation/Azimuth planes [18, 118–120]. Control
elements e.g. PIN diodes switches, MEMS, and/or Varactors should be minimized to
avoid biasing network complexity and associated RF losses. Additionally, reflectarray
unit cells are selected by considering ease of manufacture and mechanical tolerances
of the manufacturing process. For a unit cell the change in phase per unit change in
length (◦ /µm) should be low enough to accommodate manufacturing inaccuracies.
Reflectarray analysis mainly relies on a local periodicity assumption which gets easily
violated due to sharp geometric transitions of unit cells throughout the reflectarray.
2.7. Reflectarray Unit Cell Characteristics 35
These transitions can be significant when reflected phase is required to jump after
a complete 360◦ cycle [121]. As a consequence of these transitions, the radiation
pattern significantly gets deteriorated. This issue is concerned mainly with large
contoured beam reflectarrays [21]. Therefore, it is preferred to avoid those unit cells
which undergo severe geometrical changes when providing the required reflection phase.
Well-engineered unit cell properties are paramount to a superb reflectarray design.
Table 2.4 lists the unit cell properties and expectation as a guide to quality unit cell
designs [122–124].
Table 2.4: Unit cell properties and expectations
Unit cell property Expectation

Reflection phase Large range: ≥ 360◦
Reflection magnitude 1
Bandwidth High
Phase ∼ frequency As linear as possible to ensure large bandwidth
curve
Polarization Should be capable to reflect the required polarization
Polarization purity Should be high, not to deteriorate the intended
polarization
Reconfiguration Should facilitate the required reconfiguration e.g.
Polarization, frequency, beamwidth, scanning etc.
Control elements Minimum to avoid complex biasing networks and RF
losses
Geometric Low, when providing even a large phase range to avoid
transitions radiation pattern deterioration
(◦ /µm) Low, to accommodate manufacturing tolerances
Simplicity Should be simple to get manufactured
Incident angle ∼ Phase variations should be a weak function of incident
reflection phase angle
2.8. Reflection Phase Control in Unit Cells 36
2.8 Reflection Phase Control in Unit Cells
This section is about the features of various reflection phase control techniques
implemented in unit cells. Numerous techniques have been applied in literature and
can be categorized as passive or active phase control. Passive phase control techniques
benefit from changing the geometrical features of a unit cell while active phase control
techniques make use of control devices to reconfigure the surface currents in a unit cell
which then results in a desired reflection phase. Various types of passive and active
phase controls are discussed as follows:
1. Passive phase control by variable geometrical shapes
(a) Element loaded with variable stub lengths [24]
(b) Variable geometry elements [25]
(i) Single shape elements [107, 123, 124]

(ii) Compound shapes elements
• Double square ring, double cross and rectangle cross elements [125–
127]
• Double concentric ring [128]
• Concentric split ring embedded with a square [129]
• Double petal loops [130]
(c) Phoenix cell structures [21, 121]
(d) Variable length slot & multi-scale cell splitting [131]
(e) Patch physical rotation for circular polarization [109]
2. Phase control by electrical/electronic means in active unit cells
(a) Element electronic rotation for circular polarization [132]
(b) Varactor loaded cells, phase shifters, PIN diodes and MEMS switched stubs
[18]
(c) Liquid crystal cells [133] and elastomers [134] (change of εr )
(d) Reconfigurable Meta-materials

2.8.1 Passive Reflection Phase Control
The original reflectarray concept based on short ended waveguides was revived by
developments in printed microstrip technology when the advantages of reflectarrays
were actually realized. To find alternative reflection phase control solutions gained a
significant interest. As a result various passive techniques were developed to produce
the required amount of reflection phase. Most popular passive reflection phase control
techniques include stub loaded patches, variable geometrical shapes, cell rotation, multi
scale cell splitting, and phoenix unit cells. Here, only the classification of passive unit
cells is listed as shown in Fig. 2.9. The principle of operation for the majority of passive
phase control techniques used in reflectarrays is explained in Appendix A.
Figure 2.9: Various types and shapes of passive unit cells

2.8.1.1 Comparison of Passive Phase Control Techniques Table 2.5 provides

a comparison of the common passive phase control techniques. Although, this table is
very comprehensive however, certain parameters are dependent on the selected shape(s)
therefore, few entries may seem to be off range.
Table 2.5: A comparison of passive phase control techniques
Property Patch Variable Variable Patch Phoenix Varibale

1 with simple compound rotation cells slots
stubs shapes shapes
|M | 0.78 0.9 0.85 0.9 0.85 0.8
∆Φ ≥ 360◦ ≈ 340◦ ≥ 600 ◦ 360◦ ≈ 360◦ ≥ 600◦
Φ∼F medium high 6-13 medium high low high
rad/GHz
Φ∼ 0.08 0.38 0.05 high low high
∆L ◦ /µm
Φ∼ high high high high medium high
incident
angle ◦ /◦
Geomtery medium high medium low low low
Transition
Bandwidth low low medium low medium low 10%
25% 27%
Co-pol any any any only any mainly
circular linear
XPD poor excellent excellent good good good
mm-waves yes yes no yes yes/no no
fabrication
Comment stub thick & multiple only difficult difficult
matching, multi resonances, circular mm-waves mm-waves
lossy, layering smoother polarized fabrication fabrication
Φ ∼ ∆L smooths Φ ∼ ∆L,
non-linear resonances difficult
Φ ∼ ∆L mm-waves
non-linear fabrication
2
|M | = reflection magnitude, ∆Φ = achievable phase range, Φ ∼ F = phase versus frequency
dispersion (rad/GHz), Φ ∼ ∆L = phase change per unit change in dimensions (length, radius etc.)
(◦ /µm) , Φ ∼ incident angle ◦ /◦ = change in phase versus wave incident angle , Co-pol = achievable
co-polarization , XPD = achievable level of cross polarization discrimination.
2.8.2 Phase Control by Electrical/Electronic Means
This section describes the reconfigurable properties of an antenna and techniques to

facilitate it. The electromagnetic field radiated by an antenna is determined by its
surface current distribution. By intentionally altering this current distribution, antenna
characteristics can be changed. Reconfiguration is the deliberate change in operating
frequency, polarization, and/or radiation characteristics of an antenna by redistribution
of its surface currents. These reconfigurable antennas are becoming important in
modern communication systems by providing additional degrees of freedom for system
optimization, not available in fixed antennas. Due to the fact that the antenna
characteristics are not independent of each other, as a consequence changing one
property provokes changes in others to some extent. Therefore, while designing a
reconfigurable antenna following aspects are given due consideration [120].
1. Which properties of antenna are required to be made reconfigurable (frequency,

polarization, radiation pattern, or a combination of these)?
2. How to reconfigure antenna elements to achieve a desired reconfiguration

property?
3. Which reconfiguration technique minimally impacts the other characteristics

which are not being reconfigured in a particular reconfiguration?
4. Availability of selected technique at the frequency of operation?
Table 2.6 lists reconfigurable properties for an antenna.
While the reconfigurable antennas enable multitude of functionality and automation to

fairly adapt to the operating requirements and environments, these capabilities comes
at the expense of following design issues.
• Complexity associated with biasing network to control the switches. It is

particularly problematic in arrays, where a large number of wires are required
to be routed, and results in a performance degradation.
Table 2.6: Reconfigurable properties of an antenna

Property Possibilities
Continuous scan in a band
Frequency Band switching
Multi-band operation
Linear: Horizontal, Vertical, Slant
Polarization
Circular: Left hand CP, Right hand CP
Continuous scan
Fixed bean switching
Radiation Pattern Beamwidth control
Contour beam(s)
Single or multiple pencil beams
Combination of above A suitable combination of above
• Isolation of biasing lines with radio frequencies signal lines.
• Requirement of rapid tuning to ensure the system functionality.
• Increased power consumption due to integration of a huge number of active

elements. This leads to heat dissipation problems.
• Operation at high RF power may be problematic for devices.
• Generation of harmonics and inter-modulation products.
• Capacitance modulation for Varactors.
• Increased system cost and complexity.
Antenna reconfiguration can be invoked by either of the following technologies as shown

in Fig. 2.10.
1. Electrical: By switching and tuning techniques e.g. RF-MEMS, PIN diode

switches, and Varactor diodes
2. Optical: By photoconductive switches
3. Smart materials: Ferrites, liquid crystals, and elastomers (by changing εr ),

reconfigurable Meta-materials
4. Physical structural changes: By means of electric motors and re-orienting the

constituting elements of an antenna
Antenna Reconfiguration
Figure 2.10: Antenna reconfiguration technologies and their ingredients
A comprehensive discussion is made in Appendix A on these methods of

reconfiguration including electrical/electronic, optical, smart materials and element
rotation. Similarly, in Appendix A reconfigurable unit cells are explored in more
details including polarization selection for unit cells, varactor and PIN diode based unit
cells, polarization reconfigurable unit cells, frequency tunable unit cells, and monolithic
MEMS based unit cells.
2.8.2.1 Comparison of Reconfiguration Technologies When it comes to choose

a reconfiguration technology, there is no single answer to which one is the best. The
best is the one which is most suitable to constraints of the application in consideration.
One needs to consider the technology maturity, performance (loss), availability, power
handling, switching time, device linearity, and application to certain frequency bands
and the operating environment. Other design complexities e.g. ease of integration and
biasing control are given due consideration. Table 2.7 compares major reconfiguration
technologies.
Table 2.7: A comparison of reconfiguration technologies 1
Property MEMS PIN Optic Varactor Ferrite LC Graphene

Freq. GHz <40 <110 <20 >70 >1000
Voltage V 20-100 3-5 1.8-1.9 0-30 >100 - <35
kV
Current 0 3-20 0-87 100
mA
PDC mW 0.05-0.1 5-100 0-50 0-300 Low Low
Switching 1-200 1-100 3-9 uS nS mS S
speed uS nS
Isolation Very High High Medium
High
Loss dB 0.05-0.2 0.3-1.2 0.5-1.5 High 0 GHz - GHz, - GHz,
- THz + THz + THz
Linearity + 0 - 0 -
Analog / D D D A A A A
Digital
Reliability 0 + + + 0 0 -
Integration + - - - + 0 +
Biasing
Cost Medium Low High High Low 0
Available Foundry COTS COTS Special Special
Design facility facility
mm-waves - + - - - - -
application
1
+ = good, 0 = neutral, - = poor, A= analog, D = digital, COTS = commercial off the
shelf, empty box = data not available
2.9. Control Element Modeling 43
2.9 Control Element Modeling
To accurately model the control element used in a unit cell for electronic reconfiguration
is highly desirable for its incorporation into EM solver, post processing, and hybrid EM
solver-numerical optimizers. Normally, the reflecting unit cell is first modeled in EM
solver using ports at locations where control elements are expected. Then the measured
and corrected impedance models of control elements are used as loading elements in
post processing. To generate an accurate impedance model for control element e.g.
a PIN diode requires that the modeling circuitry should be a true representative of
the conditions where these devices are intended to operate. The impedance model for
control element should be prepared for all operating states, e.g. for a PIN diode or
MEMS in ON and OFF states. The measurement is carried out by connecting the
control element on a gap of a microstrip line in the same configuration (substrate, line
width, and gap length, current) as in the reconfigurable unit cell. Thru-Reflect-Line
(TRL) calibration is performed to place the reference planes at the boundaries of the
gap. A PIN diode under measurement is shown in Fig. 2.11 with TRL calibration kit.
The impedances in OFF and ON states are required to be extracted [135] [136].
MEMS can be directly modelled in EM simulator for an accurate analysis. However,

the simulation time may increase due to very small size of the component requiring
dense mesh setting. By deriving the parameters from EM simulation of MEMS and
afterwards using in post processing will speed up the process.
Figure 2.11: Through-reflect-line (TRL) calibration kit and PIN diode mounted for test
2.10. Unit Cell Performance Tests 44
2.10 Unit Cell Performance Tests
Waveguide simulators were developed to characterize the performance of phased arrays

by characterizing its few elements [137]. The same setup is employed to economically
test unit cells of a reflectarray by placing a few number of cells inside a rectangular
waveguide to simulate the periodicity assumption. While performing the waveguide
simulator tests, the impinging field’s incident angle on unit cells is an important
consideration. Dimensions of the waveguide simulator at a certain test frequency
decide the maximum number of cells that can be placed inside it. The test frequency
bandwidth is decided by the first two modes of the particular waveguide setup.
In active unit cells DC biasing lines are oriented orthogonal to the polarization of
interest in order to minimize perturbations due to biasing control lines. Lumped
inductors are used as RF chokes. Biasing wires are routed through the trenches milled
in the ground plane [117] while maintaining an electrical contact between ground plane
and the test waveguide. The bias control wires are taken out from the waveguide
simulator by drilling a small slot in rectangular waveguide [138]. To fit the geometry
of unit cells, a rectangular to square waveguide transition can be used [10]. However,
the excitation is limited to TE10 mode here too.
For dual polarized unit cells, a sophisticated measurement setup as suggested by [139]
is required. However, if the unit cell is symmetrical along both linear components,
then the characterization in only one linear polarization is sufficient [138]. It is well
established that the waveguide measurement technique as discussed above is capable
of rigorously characterizing the unit cell behavior [117, 137, 140].
2.11 Phase Quantization
Owing to the fact that reflectarrays contain a relatively large number of unit cells which
necessitate the need to have a simplified phase control mechanism for cost reduction
and to facilitate manufacturing. In case a complex control circuitry is designed, the
control DC bias lines would increase significantly which would result in a degradation
2.11. Phase Quantization 45
of the reflectarray performance. The manufacturing of a complex design incorporating

high number of bias lines is extensively demanding task particularly at mm-wave
frequencies due to reduced physical space available. On one hand it is required to
have a simplified design, whereas on the other hand, the acceptable performance is the
core design concern. This situation raises the question of what is acceptable simplified
control circuitry design for a reflectarray. This section will try to consider the main
performance parameters and their trade-offs when a simplification takes place. In a
reconfigurable reflectarray, the phase of reflected field is controlled at the reflecting
surface made up by unit cells. There are two options to control the phase:
• Continuous phase variations

• Discrete phase steps
2.11.1 Continuous Phase Variations
Designs involving a continuous phase variation mainly make use of voltage dependent
capacitance characteristics of varactor diodes by varying the applied DC bias in an
analog manner. Therefore, a particular required phase at a particular unit cell can
be synthesized in relation to its DC bias. It is highly required feature to have a
continuous range of phase variations to synthesize a scanning beam with low sidelobes
and good directivity towards a specific pointing direction. However, the capacitance
characteristics and the operating frequency range of Varactor diodes are limited. The
required capacitance to tune a particular phase using varactor diodes at mm-waves is
extremely small which is not supported by available varactor diodes. Another aspect
to consider is the large range of analog voltages to be made available for tuning the
varactors. On top of it, the increased number of elements in a reflectarray necessitate
to use a higher number of varactors. Therefore, having thousands of varactors would
be a significant challenge to control, and the accumulative power dissipation would be
another serious concern. A varactor diode based continuous phase shift unit cell design
is not preferred at mm-waves due to following reasons:
• A large range of analogue voltages required to control the array implemented using
varactors. When resistor based voltage dividers are used; the power consumption
and complex routing of the DC bias would be a problem. In case of digital

Integrated Circuit (IC) based voltage dividers are used still the bias circuitry
would lead to a significant challenge and power consumption.
• Availability of varactors at mm-waves, mainly not available.
• Capacitance value and its range at mm-waves, required capacitance is too small
to be achieved by a single varactor and there is no physical space available to
incorporate a series combination.
• Non linearity of the varactor diode capacitance [141]
• Capacitance modulation at the signal frequency during high power operations

e.g. in a satellite transmit band.
The use of Liquid Crystal based unit cells is avoided due to their lossy behavior at
lower mm-wave bands and non linear phase variations. However, they can be good
candidates for higher mm-wave bands.
2.11.2 Discrete Phase Steps
Discrete reflection phase shift steps can be implemented by employing switching devices
with microstrip lines. The switching devices can be configured in series or shunt
subject to the affordable RF loss and switch’s operating frequency. Series configuration
introduces a direct loss due to switching device whereas, the shunt switching may
introduce extra losses in microstrip lines which are considerable at mm-wave frequencies
but has the advantage of using low frequency devices as switch at higher frequencies as
the RF signal do not flow through the switching device. These discrete devices operate
in two modes either ON or OFF. Therefore, there is no continuous phase shift available.
Only a discrete set of limited phase shift values as determined by implementation is
offered to be selected from. Switching the reflection phase in steps introduces phase
quantization error. The higher the number of phase states, the lower the quantization
error. Using the experience of phased arrays and reflectarrays it has been established
that too coarse quantization e.g. 1 bit phase shifter, having only two phase states,
results in an excessive degradation of directivity. The reduction in array directivity

(directivity loss) versus the number of phase shifter bits per array element in a phased
array antenna can be found through the data presented in [142–146]. This is listed in
Table 2.8.
Table 2.8: Number of phase states and directivity loss at 30 GHz
No. of Phase No. of Phase Phase Step Directivity Loss

Shifter Bits States (Deg) (dB)
1 2 180 3.9
2 4 90 0.9
3 8 45 0.23
4 16 22.5 0.06
5 32 11.25 0.01
The values shown in the Table 2.8 are reasonable approximations as the radiation
pattern and directivity are complex functions of array parameters including the size
and inter-element spacing. It is clear from the Table 2.8 that beyond 3 bit (8 states) the
performance degradation is practically negligible and is acceptable for most practical
purposes.
Other mainly affected parameters by phase quantization are sidelobe levels and scan
angle deviation from the required pointing angle. Until recently the analogies with
phased arrays were used, which are not very accurate for reflectarrays as shall be shown
later in this research. The performance of a 64 element linear phased array assuming
point sources using various number of phase shifter bits resolution is listed in Table
2.9 [145] averaged over a 60◦ scan range.
Scan angle deviation is a function of the number of phase shifter bits and number of
array elements. Increasing the number of bits in a phase shifter results in an increased
number of possible phase distributions which finally leads to a reduction in scan angle
deviation. The scan angle deviation also decreases with an increase in number of
elements in an array due the fact that number of possible phase distributions in the
array also increases [145].
2.12. Reflectarray Cross Polarization Discrimination 48
The preferred number of bits for a phase shifter in a reflectarray is 3 as suggested

by [146] which results in almost negligible degradation for most practical purposes.
However, when it comes to mm-waves, the required gain dictates using reflectarrays
with a large number of elements, and incorporating 8 phase states in a single unit cell
really creates complexity in the large system consisting of several thousand elements.
To simplify the solution while still ensuring a reasonably optimal performance, a good
compromise among complexity, cost and implementation is to use two bit phase shifter
having four reflection phase states. The resulting directivity loss can be compensated
by manufacturing a relatively physically bigger reflectarray where possible. However,
the physical space available in a unit cell and physical dimensions of control devices
may dictate even fewer bits (states) of phase control.
Table 2.9: Performance of a 64 element phased array with various number of phase
shifter bits at 30 GHz
No. of Phase Shifter Bits

Property
2 Bit 3 Bit 4 Bit 5 Bit
Scan angle deviation, ∆θ (deg) 0.064 0.02 0.014 0
∆θ/HPBW (%) 3.28 1.01 0.73 0
Maximum side lobe level (dB) -9.18 -12.72 -12.97 -13.13
Directivity loss (dB) 0.9 0.22 0.054 0.013
2.12 Reflectarray Cross Polarization Discrimination
When an incident field of a certain pure polarization is incident on the reflectarray

aperture, it is required to preserve the polarization of the field after its reflection from
the reflectarray unit cells (unless a reflectarray is intentionally designed for polarization
conversion). Therefore, the unit cells should reflect the incoming polarized field without
degrading its polarization purity. However, there is always a coupling from co to cross
polarized fields which deteriorate the polarization purity of the reflected field. The
ratio of co to cross polarized fields in the far field in the main beam of antenna is
2.12. Reflectarray Cross Polarization Discrimination 49
defined as cross polarization discrimination/isolation (XPD/XPI) . In this thesis, the

terms XPD/XPI and polarization purity are used interchangeably. XPD/XPI is a
serious concern in dual polarized reflectarrays particularly when transmission is at
higher power levels. Pencil beam reflectarrays can achieve good cross polarization
at boresight. However, the cross polarization issue is of great concern when dealing
with contoured beam reflectarrays, where the cross polarization level is required to be
reduced over a wider satellite footprint [147]. In reflectarrays the main contribution to
cross polarization are:
1. Cross polarization level of the feeding source (usually a horn antenna)
2. Cross polarization level of the individual unit cells
3. Angle of incidence for unit cells, resulting in coupling of co-to-cross polarization
The contribution to cross polarization level by a feeding source i.e. the horn can be
made negligible by selecting a source with very good cross polarization characteristics.
Cross polarization can be controlled at element level by controlling the phase and
amplitude of reflected field by applying parallelogram transformations and trapezoidal
transformations to the reflecting patch [147]. It is also observed that the cross
polarization level depends on the incident angle of EM fields. For unit cells close
to reflectarray edges, this angle can be considerably large. An increase in focal length
is helpful; although constrained by other factors, to bring the incident angle somewhat
close to the surface normal.
When the reflection phase of a reconfigurable unit cell is changed, the cross polarization
level gets changed to some extent too. It is required to maintain an acceptable cross
polarization level when the unit cell is reconfigured which might put an additional
constraint on the operation of a unit cell, thereby limiting it to a certain range only.
By selecting the unit cell structure judiciously may alleviate the problem to some extent.
It is preferred to have unit cell structure symmetries along E and H planes to cancel
cross polarization components.
Another factor that contributes to cross polarization in multiband reflectarrays is

the coupling of inter-band elements. An optimization of the reflectarray geometry
2.13. Reflectarray Power Dissipation 50
is required to minimize such effects. At one hand the distance among unit cells for a
particular frequency band is limited to avoid grating lobes; while on the other hand
it is desirable to increase this distance to minimize the coupling effects among various
elements of different frequency bands. An optimum solution is required for which
grating lobes and coupling effects are minimized to acceptable levels.
Cross polarization can be controlled at the reflectarray level by globally mirroring unit
cells in each of the four quadrants [148] or locally mirroring unit cells in a similar fashion
at sub-array level [149].
2.13 Reflectarray Power Dissipation
The power dissipated by a reconfigurable reflectarray depends on its reconfiguration

technology. In a large reflectarray, consisting of one PIN diode per unit cell, it was
observed experimentally that the number of PIN diodes in ON state were approximately
equal to the number of PIN diodes in OFF state [142, 143]. Hence a probability of
having 50% PIN diodes actually consuming the DC power in this design. Each PIN
diode draws about 10 mA of current when in ON state, and negligible current in OFF
state. Therefore, the control circuit has to supply a current at specified voltage as:
IT = NON × IPIN (2.7)
where IT is the total current to be provided by the control circuitry, NON is the total
number of PIN diodes in ON state, and IPIN is the ON state current of one PIN diode.
It is not a matter of just supplying this much power; to transfer the heat due to this
dissipated power to a sink adds another difficulty. The excessive power requirement
of PIN diodes operation is a key concern where the available power is limited e.g.
satellite operation. It is well known that MEMS require much less power almost zero
watts for their operation, making MEMS based solution an optimal choice to reduce
power requirements. However, MEMS technology is not well established yet for space
applications. The space charging effect in MEMS makes them less reliable for such
applications. Therefore, the selection of control device is again a concern, and is based
on reflectarray electrical specifications and its operational environment. Further it was
2.14. Research Challenges 51
learnt that MEMS are not available at mm-wave bands and foundries need user designs
to manufacture and it was not an economical solution.
2.14 Research Challenges
Although modern reflectarrays have significant lucrative benefits, their wide

implementation at mm-wave bands is impeded by certain challenges demanding
simplified solutions.
C1: The availability of reconfiguration technologies at mm-waves for their application

in reflectarrays is a serious challenge. Presently, reconfiguration devices are either
not available, too lossy, or oversized when applied to mm-waves. To find devices
and their accurate characterization for integration in mm-wave reflectarrays is
an exhaustive business. Integration of multiple reconfiguration devices left alone;
presently the individual devices are much bigger in size than a mm-wave unit
cell can accommodate. A designer is often forced to decide to use the least
number of these devices. Only a few available devices which are individually
physically fit for tiny mm-wave unit cells; are not characterized at mm-waves.
Manufacturers only measure and provide data in lower frequency bands. For
an accurate characterization of a mm-wave reconfigurable unit cell (also called
active unit cell), these devices are required to be fully characterized before their
use in a unit cell. Moreover, the device characterization should be in the same
configuration (substrate thickness, mounting gap, and mounting technology etc.)
in which it is to appear in the actual reflectarray.
C2: In a well designed reflectarray, it is the unit cells’ characteristics which

demonstrate a dominance of behavior at reflectarray level. An accurate modeling
of a mm-wave unit cell is problematic as one need to consider the effects of
substrate surface waves, metal roughness, and finite metal conductivity which
are insignificant and usually ignored at lower frequency bands. Similarly, to
manufacture and accurately characterize reflectarray unit cells at lower bands is
easier where complicated unit cell shapes can be exploited to attain higher phase
ranges. However, at mm-waves the geometry of unit cells become very small
therefore, creating difficulty in their manufacturing. This is mainly because the
same manufacturing tolerances have greater impacts on a tiny mm-wave unit cell
whilst at lower frequencies their effect is negligible. At mm-waves the accurate
manufacturing often limits the unit cells to more fundamental shapes with a
limited reflection phase range. In a reflectarray design ideally all the phase values
from 0◦ - 360◦ are required. However, with limited phase range the reflectarray
performance is compromised in terms of gain and bandwidth. Accurately
characterizing mm-wave unit cells is an ambitious task where a slight imperfection
can lead to greater errors. During tests it becomes significantly difficult to identify
whether it is the manufacturing or test setup causing misleading measured results.
C3: Due to a large number of reconfigurable unit cells in a smart reflectarray, the
power consumption becomes a concern. This is not only due to the amount of
power being consumed but also due to the concern of heat generated. In small
spacecrafts with limited amount of on board available power, the issue of power
consumption by smart antennas gains even higher attention. There is a need to
find the least power consuming options at mm-waves.
C4: When a reconfigurable unit cell have multiple phase states; it gives rise to two
simultaneous problems. First is the provision of DC biasing for control devices
used to make reconfiguration possible. With multiple devices in a single tiny
unit cell, the accommodation of multiple DC bias lines is a big challenge if
not impossible. It can not be achieved without significantly compromising the
performance of a unit cell and finally the reflectarray. It is worth mentioning here
that low frequency methods (unit cell slicing, gaps, and finger break capacitors
etc.) used to isolate DC bias for multiple control devices are not suitable
at mm-waves as their dimensions are bigger than a mm-wave unit cell can
accommodate. Moreover, the minimum manufacturable sizes by the PCB printing
facility often causes constraints. A dimensional reduction in these low frequency
methods is either impossible due to manufacturing or introduce severe losses
which are not acceptable in mm-wave unit cells. Using them with normal optimum
size, results in a bigger unit cell which introduces grating lobes in a reflectarray.
Even if one were able to overcome the first issue there is a second issue. It is
related to the magnitude of the reflected field. It is desired that cross polarized
reflected field w.r.t. to the reflected field in the desired polarization should be
minimum. However, with multiple reconfigurable states it becomes a challenge
to maintain a low level of the reflected cross polarized field. It often results in
cross polarization degradation at reflectarray level and introduces losses in the
reflected field of the desired polarization.
C5: To design a reconfigurable unit cell with the above mentioned constraints at
mm-waves in itself becomes a challenge. On one hand the designer is limited in
choices and on the other hand it is required to attain higher performance. To
achieve an optimum design requires a very comprehensive critical understanding
of the available technologies and design options.
C6: To gain confidence in reflectarray design, it is required to perform 3D EM

simulations prior to manufacturing. There are no commercially available
softwares which can efficiently build the EM simulation model of the reflectarray
as a whole. Therefore, designer have to build this model in a 3D EM simulator
by designing each individual unit cell to provide the required reflection phase in
a reflectarray. There are thousands of elements in the reflectarray, to build the
model manually is a potentially difficult task and prone to human errors. This
issue becomes serious if the designer have to do multiple design iterations. It is
a very time consuming laborious task yet to be done accurately to achieve the
required reflection phase distribution over the reflectarray aperture. Further, the
EM simulation of a reflectarray containing a huge number of elements requires
significantly high end computing resources.
C7: To achieve a continuous reflection phase shift for reflectarray phase

reconfiguration is almost impossible at mm-waves. Therefore, to identify what
level of phase control is sufficient to achieve certain performance in reflectarrays
is necessary. Current studies benefited from the phased array analogy. However,
the phased arrays usually have a much lower number of elements which make
2.15. Chapter Summary 54
these analogies inaccurate. There is a need to establish new performance versus

complexity trade-off matix for reflectarrays. Due to much higher efforts required
to build multiple reflectarray EM simulation models this area has not been
extensively explored in the past.
One would have realized by now that mm-wave smart reflectarrays have potential
design challenges and significant implementation implications. However, the flexibility
and scalability offered by the successful designs is huge and will enable unprecedented
opportunities in mm-wave communication systems.
2.15 Chapter Summary
At mm-wave frequencies the individual antenna elements become very tiny, whereas
mm-wave signals suffer a significant amount of propagation losses. This necessitates
the use of high gain antennas to attain high throughput communication links.
When antenna gain is high, its beamwidth becomes very narrow. In a mobile
environment (where transmitter or receivers can move, mobile communication, satellite
communication, radar tracking etc.) one would experience intermittent connectivity
loss if antenna beams do not track. Therefore to maintain a reliable communication
link these high gain antennas are required to be smart to track/steer their beams
towards the intended direction.
The potential of mm-waves remained mainly ignored until recently, therefore such
high gain smart antenna solutions at these bands are not commonly available. To
achieve high gains two well know solutions are phased arrays and parabolic reflector
technologies. Phased arrays can achieve high gains as well as wide angle beam-steering.
However they are very lossy at mm-wave spectrum due to their complex feeding
structure. Similarly, to implement a large number of RF chains in a beamformer is
very expensive. For a wide spread adoption of smart antennas at mm-waves they
are required to be simple as well as cheaper. The reflector technologies can achieve
high gains but their electronic beam-steering is very limited. Therefore, both of these
solutions are not suitable for smart antennas intended to operate at mm-wave spectrum.
A possible solution to achieve high gain as well as wide angle electronic beam-steering
is to combine reflector and phased array technologies. This generates a wonderful
genetic hybrid of these technologies which is called a reflectarray. Reflectarrays benefit
from spatial feeding of reflector technology thereby eliminating any complexities in the
feeding mechanism of phased arrays. The reflecting surface in a reflectarray benefits
from the printed microstrip technology which makes it a low cost planar solution. The
reflecting surface of a reflectarray is made up of a large number of elements called unit
cells. These unit cells can be made reconfigurable to accomplish versatile functionalities
e.g. beam-steering, single or multiple tracking pencil beams, or contoured beams.
However to implement a reconfiguration mechanism at mm-waves is not an easy

task. Reconfiguration technologies capitalize on electronic components (PIN, varactor
diodes), MEMS, and liquid crystals etc. The geometry of individual unit cells
become too small to integrate multiple electronic components. Therefore, various
reconfiguration solutions available at lower frequency bands are not applicable
to mm-waves. Millimeter wave spectrum need novel simplified solutions for
implementation of high gain smart antennas.
The purpose of this research is to realize a simplified novel solution for high gain smart
antennas in the form of reflectarrays at mm-waves. This chapter presented smart
antenna techniques, technologies and reflectarrays. Various aspects of reflectarray were
covered with due consideration for implementation perspectives. Among the available
reconfiguration technologies only the PIN diodes are considered as a potential candidate
at mm-waves. Significant challenges for a reflectarray implementation at mm-waves
were identified and transformed to objectives of this research presented in section 1.2.
Chapter 3
Characterization of Passive
Reflectarray Unit Cells
A reflectarray consists of a large number of unit cells spatially illuminated by a feed

horn. The presence of these unit cells provide a powerful control of the overall
characteristics of a reflectarray in terms of gain, beamwidth, beam scanning, and
contour beam generation etc. [27, 150]. To synthesize optimum unit cells for a
reflectarray is one of the major design steps. A detailed review of the existing passive
phase control techniques can be found in Appendix A. This chapter presents the
characterization of passive unit cells. The selection of a reflectarray unit cell’s geometry
in relation to fabrication at mm-waves is discussd in section 3.1. A numerically efficient
analytical modeling technique for microstrip reflectarray unit cells placed inside a
dominant mode waveguide based on coupled waveguide resonator theory is presented
in section 3.2. The unit cell’s reflection loss is incorporated in the model using the
perturbation technique. Effects of surface roughness, fringing fields and surface waves
on the reflection properties of a unit cell are considered in the analysis to ensure high
accuracy for millimeter wave bands. Analytical expressions for a unit cell lumped
element equivalent model are derived. Parametric analyses of unit cell reflection
properties in relation to the design variables are presented in section 3.3. Parametric
analyses using analytical approach provide a numerically efficient yet accurate initial
design therefore, primarily reducing the load of 3D electromagnetic (EM) simulations
56
3.1. Unit Cell Selection for mm-waves 57
to reach an optimum design solution.
The analytical procedure was applied to design a mm-wave unit cell. EM simulations
of the unit cell are presented in section 3.4. A large number of unit cells were
fabricated and measured as discussed in section 3.5. Two new measurement techniques
are proposed to ease the measurement process. Results from the analytical solution,
EM solver, and measurements are compared. Additionally, the measurement results
of traditional measurement technique and proposed techniques are compared. An
excellent agreement was found in all the cases. Finally, this chapter concludes with
the basic design of a unit cell ready for reconfiguration in the next chapter.
3.1 Unit Cell Selection for mm-waves
It was shown in section 2.8.1 and Appendix A, there exist multiple ways to achieve
the reflection phase control in passive unit cells. Among the presented techniques and
shapes of unit cells, only the very basic shapes are suitable for mm-waves. This is due
to the fact that complex shapes are very prone to fabrication tolerances at mm-waves.
The minimum achievable gap and track widths on the PCB were 76 µm for an out
sourced PCB facility while 0.5 mm for our in house facility. The tolerances were 25
µm and 100 µm in both cases respectively. Plated through hole’s positioning accuracy
for the out sourced facility was 0.1 mm, with a minimum hole size of 0.15 mm over a
minimum pad of 0.3 mm. A majority of cheap high end commercially available PCB
printing facilities are in a similar or even worse range of parameters for fabrication.
We deliberately avoided the use of very expensive mm-wave fabrication facility due to
project cost constraints. The whole purpose of this exercise was to gain a sufficient
confidence in the passive unit cell performance before it is used for reconfiguration
purposes.
Among basic shapes rectangular and square shapes are the most tolerant to fabrication
tolerances. Even in the case where the fabricated unit cell’s dimensions gets changed
due to tolerances, it would still be able to sufficiently achieve the performance
at a frequency in the vicinity of its original design frequency. However, when
complex/compound shapes are selected it may not be the case. Dual polarized unit cells
3.2. Design and Analysis of mm-wave Passive Unit Cells 58
were preferred in the design to accommodate both linear polarizations (horizontal and
vertical). Therefore, a square shaped patch for the mm-wave unit cell was selected. In
addition to being fabrication friendly, the square/rectangular unit cells can be modeled
analytically using relatively simpler formulation which is a very desirable feature for
the parametric analysis.
3.2 Design and Analysis of mm-wave Passive Unit Cells
In a well-designed reflectarray, the overall performance is mainly determined by its

comprising unit cell(s) [25]. The reflectarray design and performance prediction has
attracted significant attention [30, 151] and has matured to a certain level. However,
until recently, most of the unit cell designs mainly relied on 3D full wave electromagnetic
(FW-EM) simulations. Widely used simulation packages including CST Microwave
Studio [152–157] and HFSS [158–160] have been employed to design and characterize
the unit cells. Although such simulations, when performed carefully, can predict
comparable results to measurements, nevertheless they are very time consuming. Due
to multiple degrees of freedom in design, the designer has to bear heavy simulation
loads in terms of optimizations and parameter sweeps to reach an optimum design
that can achieve the required performance. The complex process of EM simulator
based design necessitates preparing a mathematical model which can reliably predict
the unit cell performance over a range of its multiple parameter variations in contrast
to a fixed set of simulation parameters in a full wave solver. From the results of this
mathematical model, the designer can choose unit cell design parameters followed by
only a single or few EM simulations to converge on the optimum design. Additionally,
having an analytical model facilitates computer aided design (CAD) software modeling
of a reflectarray’s unit cell.
The reflection response (magnitude and phase) of a unit cell is the primary determinant
of its behavior over a frequency range. For a microstrip based unit cell the reflection
response is determined by substrate dielectric constant (r ), loss tangent (tanδ),
substrate thickness (h), metal conductivity (σ), and the reflecting surface geometry.
Due to multiple variables, there may exist multiple combinations leading to the same
resonant frequency but completely different reflection properties. To synthesize an

optimum design one should make best use of the available degrees of freedom to
achieve the desired performance. Therefore, it is critical to completely characterize
and understand the properties of a unit cell in relation to its parametric variables to
avoid accidental misbehavior in the reflection response.
Only a few research efforts have been reported in the literature even at lower frequency
bands where an attempt to analytically characterize the unit cell behavior has been
presented. For instance [161] identified the issue of phase misbehavior using equivalent
circuit model based on poles and zeros analysis. Bozzi et al. [124] used numerical
modeling to reach to an equivalent circuit model of the unit cell. Very recently, Kalyan
et al. [162] introduced the idea of preparing an analytical model for the unit cell.
However, their analysis does not consider the effects of surface roughness, fringing fields
and surface waves, which play a significant role in reflection properties at mm-wave
bands. Their phase range estimation does not consider the contribution from the
unit cell area not occupied by the reflecting patch therefore, resulting in an over
prediction of the phase range for thick substrates. Moreover, it lacks in driving an
analytical equivalent model of a unit cell for its further extension at array level. The
mm-wave band has remained largely ignored until recently due to unfavorable inherent
propagation properties [163, 164]. Therefore, relatively little research has been carried
out in the field of mm-wave reflectarrays and their characterization.
A unit cell can be represented by an equivalent RLC resonator. The modeling

of a waveguide coupled resonator under the assumption of small perturbations to
approximate the energy decay in a resonator is presented here. Theory developed
in [165] for the prediction of reflection properties of a resonator in terms of various
quality factors (Q) is presented in relation to a reflectarray unit cell. The model is then
applied to the case of a mm-wave reflectarray unit cell placed inside a dominant mode
waveguide. Analytical expressions for reflection magnitude, reflection phase, phase
gradient at resonant frequency and bandwidth are derived. Closed form mathematical
expressions to extract the values of unit cell equivalent lumped component circuit model
are subsequently derived.
3.2.1 Unit Cell Modeling
A unit cell is shown in Fig. 3.1. It consists of a reflecting patch on a grounded substrate.
It is well known (as discussed in section 2.10) that the unit cell reflection properties are
measured using a waveguide setup as it provides periodic boundary conditions required
for the unit cell. A reflectarray’s unit cell placed inside a dominant mode waveguide
is modeled based on the waveguide coupled resonator theory [165]. A reflectarray unit
cell is represented by an equivalent RLC resonator circuit. Mathematical equations for
lossless and lossy RLC resonators are derived. The perturbation technique [166] was
used to approximate the mode amplitude decay rate. Then the model progresses to the
coupling analysis of a lossy RLC resonator to the waveguide. Certain quality factors
(Q) are calculated and manipulated to characterize the unit cell behavior.
Waveguide ‘ ’
Length Ground plane
Patch
Patch
Grounded substrate
Figure 3.1: A reflectarray unit cell (front and side views).
3.2.1.1 Modeling of an RLC Resonator In a lossless parallel LC resonator as

shown in Fig. 3.2 (a) the current i(t) and voltage v(t) are defined by well-known
expressions:
d i(t)
v(t) = L (3.1a)
dt
d v(t)
i(t) = −C (3.1b)
dt
i( ) i( ) +(t) i( )
+ -(t)
vs
v( ) v( ) - Waveguide v( )
Zg Γ
(a) (b) (c)
Figure 3.2: RCL equivalent circuit of a reflectarray unit cell (a) Lossless case, (b) Lossy
case, loss represented by a resistor in series with capacitor to model the increase in loss
due to an increase in frequency, (c) A unit cell resonator coupled to the waveguide with
an external excitation source (dominant mode waveguide incidence), the incident wave
is represented by s+ (t), the reflected wave is represented by s− (t), and the reflection
coefficient of the unit cell is represented by Γ.
A time dependent complex variable a(t) is defined as the resonant mode amplitude of
a positive frequency component in an LC resonator [165]:
r r
C L
a(t) = v(t) + j i(t) (3.2)
2 C
Differentiating (3.2) with respect to t; noting the values of voltage as v(t) = j ωo L i(t)
and current as −i(t) = j ωo C v(t) results in the following:
d a(t)
= j ωo a(t) (3.3)
dt
√
where ωo = 1/ LC. The solution to (3.1) can be derived starting with voltage as a
cosine wave as follows:
v(t) = |V | cos (ωo t + φ) (3.4a)

r
C
i(t) = |V | sin (ωo t + φ) (3.4b)
L
Using (3.4) in (3.2) results: r
C
a(t) = |V | ej ωo t (3.5)
2
where V = |V | ej φ , and φ = arg(V ) is the phase. It can be observed from (3.5) that
a(t) has ej ωo t dependence. The energy stored (Ws ) in a parallel LC circuit can be
related to mode amplitude a(t) by:
C L
Ws = |V |2 = |a(t)|2 = |I|2 (3.6)
2 2
In a parallel LC circuit, the losses can be effectively represented by a resistor as

shown in Fig. 3.2 (b). The unit cell reflection loss increases with an increase in
operating frequency. At higher frequencies the loss tangent of the material increases
too. Similarly, an increased conductor surface roughness due to an increase in skin
effect causes a decrease in surface conductivity. As the operating frequency increases,
capacitive reactance in the equivalent RLC circuit decreases, causing more current to
flow through the resistor, which finally results in more power dissipation. Therefore,
this loss is modeled by a resistor in series with the capacitive branch [124]. This second
order lossy resonant circuit expressed as Laplace transform has the solution of es t form,
where its complex roots are:
s 2 r
R 1 R 1 1
s1, 2 =− ±j − =− ±j ωo2 − (3.7)
2L LC 2L τ τ2
In this lossy resonant circuit the amplitude decay rate is determined by τ −1 = R/(2 L).
This decay rate can be computed in another way through perturbation theory [32] in
the limits where loss is small and can be considered as a perturbation. Making use
of the perturbation technique simplifies the introduction of small losses in a lossless
solution without resorting to the complex procedure of an exact solution. Introduction
of small losses in (3.3) results in:
d a(t) d ar (t) 1
→ = j ωo ar (t) − ar (t) (3.8)
dt dt τo
where ar (t) is the perturbed mode amplitude due to the addition of a small resistance,
and a temporarily introduced variable τo represents the decay rate due to small
loss/perturbation. From (3.8) ar (t) has a time variation of ej ωo t e−t/τo . Therefore,
|ar (t)|2 decays as e−2 t/τo with time. Using this in (3.6), the time rate of stored energy
decay which is also equal to power dissipation (Pd ) in the circuit becomes:
d Ws 2
= − Ws = −Pd (3.9)
dt τo
Assuming the perturbation due to R is small; therefore the current and voltage can
still be approximated by (3.4). In this situation the time averaged power dissipated in
the circuit of Fig. 3.2 (b) is:
Pd = |I|2 R/2 (3.10)
Using the value of |I|2 in terms of |a(t)|2 from (3.6) and making use of (3.9) and (3.10)
in the limit of small loss one can find:
R
Ws = τo |ar (t)|2 (3.11)
2L
Comparing (3.6) and (3.11) it is found that τo = 2 L/R = τ .
Therefore, perturbation technique significantly simplifies the modeling of decay rate as

expressed by the exact solution in (3.7). However, the correction to ωo term due to loss
in exact solution (3.7) is not modeled by this procedure. It can be safely ignored here in
unit cell analysis as we are interested to find the reflection loss not the frequency shift.
The benefit of perturbation technique will become clear when the waveguide coupled
resonator is introduced in Section 3.2.1.2.
For the unit cell resonator of Fig. 3.2 (b), the unloaded quality factor (Q) being a ratio
of average energy stored to the energy lost per second then becomes [166]:
Ws L ωo τo
Q = ωo = ωo = (3.12)
Pd R 2
3.2.1.2 Waveguide Coupled Unit Cell Resonator Modeling Fig. 3.2 (c)
displays a resonator coupled to an external waveguide. In the absence of any excitation
signal (vs (t) = 0), the RLC resonator is terminated in the waveguide impedance Zg .
Now in addition to the energy being lost internally in a lossy resonator, the energy will
also escape into the waveguide. Therefore, the new mode amplitude ag (t) will have a
time variation of the form ej ωo t e− t/τo e− t/τe , and (3.8) is modified as:

d ar (t) d ag (t) 1 1
→ = j ωo ag (t) − + ag (t) (3.13)
dt dt τo τe
where τe is the additional decay rate due to an external waveguide. The time rate of
change of the energy then becomes:

d Ws 1 1
= −2 + Ws (3.14)
dt τo τe
The power escaping into the waveguide is Pe = 2Ws /τe and therefore, the resonator’s
external Qext is expressed by:
Ws τe
Qext = ωo = ωo (3.15)
Pe 2
When the excitation signal vs (t) is non zero, it sets up a wave s+ (t) = s+ ej ωt , where
s+ = |s+ | ej ψ and ψ is the phase; traveling towards the resonator. The power carried
by this incident wave is |s+ |2 . This wave serves as a drive for the waveguide coupled
resonator’s resonant mode amplitude ag (t). If κ represents the coefficient of coupling
between the resonator and this incident wave, then the time derivative of the resulting
mode amplitude is given by:

d ag (t) 1 1
= j ωo ag (t) − + ag (t) + κs+ (t) (3.16)
dt τo τe
Because excitation source has frequency ω, the response will be at the same frequency.
p
Therefore, the resulting mode amplitude comes out to be (3.17) [165], where κ = 2/τe
and its phase can be absorbed by the arbitrarily defined phase of mode amplitude
relative to the incident wave s+ (t):
κs+ (t)
ag (t) = (3.17)
1 1
j (ω − ωo ) + τo + τe
To find the reflection coefficient Γ = s− (t)/s+ (t) of the resonator excited by an incident
wave, one needs to find s− (t). Because the system is linear, the reflected wave will be a
sum of terms proportional to the incident wave and the mode amplitude. The reflected
wave and the resulting Γ are given as [165]:
s− (t) = −s+ (t) + κ ag (t) (3.18)
1 1 1 − 1 − 2 j (f − fo )
τe − τo − j (ω − ωo ) Qext Qo fo
Γ= = (3.19)
1 1 1 1 2 j (f − fo )
τe + τo + j (ω − ωo ) Qext + Qo + fo
At resonance we have:
1 1 1 1
τe − τo Qext − Qo
Γ |ω=ωo = 1 1 = 1 1 (3.20)
τe + τo Qext + Qo
The expressions for Γ are very important results in the characterization of a reflectarray
unit cell.
3.2.1.3 Application to Millimeter Wave Reflectarray Unit Cell: To

characterize the reflection properties of the unit cell of a reflectarray one need to
calculate the Qo and Qext . This section presents the design of a mm-wave unit cell
and its characterization.
1. Design of Reflecting Surface Geometry: A unit cell’s reflecting geometry can

take various shapes including rectangular, square, triangular, circular, rings, and
their complex combinations. The only difficulty faced during analysis presented
here is the calculation of radiated power from the selected shape required for the
calculation of radiation quality factor Qrad . At mm-waves the dominant shapes
in use are the simple rectangular or square geometries because these are more
tolerant to fabrication tolerances at these frequencies and are easy to fabricate
reliably. Here, the analysis proceeds with a simplified case of a rectangular
reflecting patch. For square patches the analysis is similar when width and length
of the patch are made equal.
For a unit cell consisting of a rectangular reflecting patch over a grounded

substrate, when the effect of fringing fields is considered, the effective width Weff
that leads to a good radiation efficiency is given by (3.21) [167, 168]. Similarly,
the effective length Leff to resonate at frequency f is expressed by (3.22) [35].
r
λo 2
Weff = + 2 ∆W = W + 2 ∆W (3.21)
2 r + 1
c
Leff = √ = L + 2 ∆L (3.22)
2 f eff
where λo is the free space wavelength, r is the substrate dielectric constant, eff is
the effective dielectric constant for the substrate at operating frequency as defined
by [169], W and L are physical width and length of a rectangular patch, ∆W
and ∆L are effective increase in width and length due to fringing fields as defined
by [170] and [171] respectively.
2. Conductor and Dielectric Losses: Internal losses in a unit cell are due to the
losses associated with dielectric and conductor (metal) properties. A combined
quality factor Qo takes into account both of these losses and is calculated using
a parallel combination of quality factors due to dielectric and conductor losses
modeled by Qd and Qc respectively.
1 1 1
= + (3.23)
Qo Qd Qc
The quality factor Qd relating to dielectric loss due to tanδ of the microstrip
substrate is defined by [172]. The conductor loss is due to finite metal conductivity
and surface roughness of the reflecting surface. Surface roughness is mainly

ignored at lower frequencies. At mm-wave frequencies, surface roughness becomes
very comparable to the skin depth, thereby increasing the internal losses in a
unit cell. The metal surface roughness Rrms ranges from 0.3 µm to 2.4 µm for
commercially available copper depositing techniques used in manufacturing of
printed circuit boards (PCB) [173]. Considering conductivity of copper, one finds
skin depth δs of the order 2 µm ≤ δs ≤ 0.2 µm in the frequency range of 1 GHz
≤ f ≤ 100 GHz. It is comparable to the metal surface roughness of commercially
available PCB substrates.
With an increase in the frequency the dielectric loss tangent increases. Similar
is true for the conductor surface roughness. At mm-waves conductor loss is
significantly high. Therefore, it is essential to consider the effects of surface
roughness in mm-wave designs. The effective conductivity of a rough surface σr
is modeled as [174, 175]:
σ
σr = " ( )#2 (3.24)
2 −1 Rrms 2
1 + tan 1.4
π δs
For thin microstrip substrates h λo one can find the rough conductor model
based quality factor Qc using (3.25) which is particularly important in relation
to higher frequencies in the range of mm-waves.
p
Qc = h π µ f o σr (3.25)
Accuracy of the model can be improved even further at higher frequencies by

using complex models for effective metal conductivity.
3. Radiation Quality Factor Qrad: Finding the radiation quality factor Qrad of
a microstrip based antenna necessitates having an insight of its electromagnetic
radiation behavior in terms of stored energy Ws and radiated power Prad . Qrad
is expressed as [176]:
Ws Ws
Qrad = ωo = 2 π fo (3.26)
Prad Prad
By virtue of the cavity model for patch antennas [35, 177], the electric field
distribution under the patch surface can be modeled by equivalent magnetic
currents at its radiating edges [177]. The presence of a ground plane doubles
the magnitude of magnetic currents due to image theory assuming a very
thin substrate. These radiating edge magnetic currents excite mode(s) in
the waveguide, the amplitude(s) of which can be found using mode matching
technique [166]. Here the simplified case of rectangular or square patch is
considered. Subject to the orientation, the dominant mode of a rectangular patch
is TM01 or its orthogonal. A square patch can simultaneously support both of
these modes, one for each linear polarization. Similarly, a rectangular waveguide
has TE10 as dominant mode, while a square waveguide can simultaneously support
TE10 and its orthogonal modes, where each mode represents a linear far field
polarization. Placing the unit cell containing reflecting patch geometry inside
the dominant mode waveguide; considering one mode at a time for the patch
(e.g. TM01 ) and a corresponding mode in the waveguide (TE10 ), the amplitude
of excited waveguide mode by the equivalent magnetic currents at radiating edge
is given by [162]:
+ 8 h Eo π Weff
A = sin (3.27)
πb 2a
where A+ is the TE10 mode amplitude, Eo is the electric field amplitude under
the patch, ‘a’ and ‘b’ are waveguide dimensions, Weff is the effective width of a
patch.
TE10 mode radiated power inside a rectangular waveguide by a rectangular patch

then becomes:
16 h2 a Eo2

ab + 2 2 π Weff
Prad = |A | = 2 sin (3.28)
4 Zg π b ZTE 2a
where ZTE is the wave impedance for TE mode in the waveguide given by:
ωµ
ZTE = q (3.29)
ω2 µ ε − ( π
a)
2
At resonance, the stored electric and magnetic energies are equal; therefore total
stored energy is given by [166]
Eo2 h Weff Leff

WsT = (3.30)
4
The resulting radiation quality factor Qrad is expressed by [178]:

fo π 3 Weff Leff b ZTE
Qrad = (3.31)
32 h sin2 ( π 2Waeff ) a
In case of a square patch, the treatment for the orthogonal mode (i.e. TM10 and
corresponding TE01 ) is similar by making the width equal to the length. However,
the mode amplitude for other patch shapes need to be calculated using equivalent
magnetic current density of patch and the magnetic field of the waveguide mode.
The expressions for other shapes can be derived following the treatment given
by [179]. At mm-waves; due to manufacturing inaccuracies, a simplified shape is
preferred however.
4. Surface Wave Effect: For thin substrates the surface wave effect is negligible.
However, for thick substrates its effect needs to be included as illustrated by
[180, 181]. At mm-waves h/λo becomes relatively greater as compared to lower
frequencies for the same substrate thickness. Therefore, the surface wave effect
is more pronounced at mm-wave frequencies. Surface waves are not a strong
function of the reflecting element geometrical shape and are mainly characterized
by substrate parameters [168,182]. It implies that the radiated wave efficiency can
be determined without the knowledge of the reflecting element’s exact geometry.
Therefore, one can replace the reflecting patch geometry by a simplified model
like an infinitesimal electric dipole current. A surface wave in itself is not a loss
in a unit cell rather an undesired radiation thereby lowering the radiated wave
efficiency η defined as:
Pd |rad
η= (3.32)
Pd |rad + Pd |sw
where Pd |rad is the radiated wave power and Pd |sw is the surface wave power due
to an electric dipole current on a grounded substrate. Radiated wave and surface
wave powers for such a simplified scenario were modeled by [182], only the final
results (3.34) and (3.35) respectively are retained here for completeness. The
following expressions are valid for substrate thickness values as defined by (3.33)
depending on the value of r :
√

1/(4 r − 1) for a low value of r ≈ 2.55

h /λo ≤ (3.33)
1/(8√ )

for a high value of r ≈ 12.8
r

1 2
Pd |rad =40 ko2 (ko h µr )2
1− + (3.34)
r µr 5 (r µr )2
p
ηo ko2 r (x2o − 1)3
Pd |sw = p (3.35)
8 r (1 + x1 ) + ko h (1 + 2r x1 ) x2o − 1
where
p
x2 − 1 αo α1 − 2r + r 2r − 2 αo α1 + αo2
x1 = o , xo = 1 +
r − x2o 2r − α12
√ √
tan (k h − 1) + ko h √r −1
√ √ o r cos2 (ko h r −1)
αo = r − 1 tan (ko h r − 1) , α1 = − √
r − 1
When the effect of surface waves in taken into account the new value of Qrad is
modified by η. Therefore, new value of Qrad denoted by Qrad |sw is given by (3.36),
which is the value one have to use in place of Qext in the reflection co-efficient
formulation in (3.19) [183].
Qrad |sw = η Qrad = Qext (3.36)
The surface wave quality factor can be found by:

η Pd |rad
Qsw = Qrad = Qrad (3.37)
1−η Pd |sw
5. Reflection Conditions for Reflectarray Unit Cell: The unit cell response is
a function of its dimensions, substrate parameters and the frequency. Examining
(3.19), four conditions can be inferred:
(a) When |ω−ωo | 1/Qext +1/Qo , then (3.19) becomes Γ(ω ωo ) = ej π = −1

which indicates that a far resonance excitation of unit cell results in perfect
reflection providing a reflection phase shift of 180◦ always, which is not
desired in a reflectarray.
(b) At resonance, the condition Qext > Qo corresponds to the under coupled
condition and from (3.20) the reflection coefficient becomes Γ(ωo ) =
|Γ(ωo )|ej π giving a reflection phase of 180◦ . An under coupled reflectarray
unit cell exhibits anomalous phase phenomenon known as phase misbehavior
as was observed in the past [161] using a circuit model. In a reflectarray unit
cell design it is necessary to avoid this condition to produce a reasonable
reflection phase swing.
(c) At resonance, the condition Qext = Qo is known as critical coupling and

Γ(ωo ) = 0 indicating there is no reflection from the unit cell. All the
incident energy is dissipated internally inside the resonator. This condition
is contrary to the reflectarray reflection based operation and therefore is not
useful. The absorbed energy is dissipated as heat and may cause thermal
issues in the system.
(d) At resonance, when Qext < Qo , is called over coupled condition and
Γ(ωo ) = |Γ(ωo )|ej 0 . This is the only useful condition for reflectarray design.
Therefore, the reflectarray unit cells are designed for over coupled condition
for a reliable operation.
6. Reflection Phase of Unit Cell: The reflection phase of a resonator can be

derived by taking the argument of (3.19) as [178]:
 
 
2 2(f −fo )

 

 
−1 Qext fo
Φres (f, fo ) = arg Γ(f, fo ) = tan − 2 2 2
− 2(ff−f o)
 1

− Q1o

 

 Qext o

(3.38)
Due to the fact that resonator (reflecting patch) is partially covering the area
of a unit cell (AUC ), therefore the reflection phase of unit cell ΦUC is a linear
combination of reflection phase due to the resonator Φres (f, fo ) and the phase
contribution due to the wave from the unit cell area not covered by the resonator.
The phase contribution from the uncovered area depends on the substrate
√
thickness (h) and wavelength in the substrate λsub (f ) = c/(f r )
AUC − Ares h
ΦUC (f, fo ) = Φres (f, fo ) − 2π (3.39)
AUC λsub (f )
where Ares is the area covered by the resonator.
7. Phase Derivative at Center Frequency: The phase derivative at resonant

(center) frequency is expressed as follows:
4Q2o Qext

∂ Φres (f, fo )
[∂Φ/∂f ] = − = (3.40)
∂f
f =fo fo (Q2o − Q2ext )
8. Unit Cell Bandwidth and Phase Swing: The unit cell bandwidth is the range
of frequencies over which the reflection phase differs by ±45◦ as compared to the
center frequency. A unit cell can benefit from multiple means of phase control, e.g.
phase change due to geometrical changes or a change is substrate properties. Here
a variable length patch approach of unit cell has been chosen. In a variable length
reflecting patch technique, if Lo is the resonant length at center frequency fo , δLo
represents a differential change in the length, and lengths Lo + δLo , Lo − δLo
corresponds to frequencies f1 and f2 as in (3.41) such that f1 < fo < f2 , then the
unit cell phase swing ∆Φ is defined as:
f1 [Lo + δLo ] = fo Lo , f2 [Lo − δLo ] = fo Lo (3.41)
∆Φ = ΦUC (fo , f2 ) − ΦUC (fo , f1 ) (3.42)
Unit cell bandwidth BW, then becomes:
π Q2o − Q2ext

1 π 1
BWUC = = (3.43)
fo 2 [∂Φ/∂f ] 8 Q2o Qext
9. Unit Cell Lumped Element Equivalent Model: A unit cell’s equivalent

lumped element model is derived using the analytical phase derivative at the
center frequency. The lumped equivalent model consists of an inductor in parallel
with a series RC branch as shown in Fig. 3.2(b). Due to the fact that loss
increases with frequency, the resistor is added in series with the capacitive branch.
As the resistor value is very small therefore, the L and C can be modeled by (3.44)
and (3.45) respectively [124].
Q2o − Q2ext

2 ηo ηo
L= 2
= (3.44)
π fo [∂Φ/∂f ] 2 π fo Q2o Qext
2
[∂Φ/∂f ] 1 Qo Qext
C= = (3.45)
8 π ηo 2 π fo ηo Q2o − Q2ext
where ηo is the free space intrinsic impedance. At resonance, where reflection
phase is 0◦ , the value of R is given by:
" 2 #2
1 − Γ(f = fo ) 16 ηo ηo Qext
R= 2 = 1− (3.46)
1 + Γ(f = fo ) fo2 [∂Φ/∂f ] Qo Qext Qo
3.3. Parametric Study of Reflectarray Unit Cells 72
Equations (3.19) to (3.46) can be programed for computer aided design to parameterize
the variable of interest to study its impacts on the resulting performance of a designed
unit cell. The extracted lumped circuit elements can be used for analysis at the array
level. The unit cell equivalent lumped model accounts for the effects of mutual coupling
based on infinite array due to periodic boundray conditions represented by waveguide
metallic walls used in analysis. For the numerical efficiency comparison of this analytical
methodology these equations were programmed in MatLabr . It took only 96 µs to
complete one iteration for a set of parameters for a single frequency on a Dell Optiplex
790 core i5-2400, quad core, CPU @ 3.1 GHz with 8 GB RAM. Time was recorded
using tic-toc stop watch in MatLabr . One can compare this with CST microwave
studio simulation which would take several minutes for the same set of parameters for
a single frequency calculation depending on the mesh settings (5 minutes in this case).
The real performance is seen when user have to optimize or parametrize a unit cell,
where each individual set of parameters would cost a separate EM simulation.
3.3 Parametric Study of Reflectarray Unit Cells
In this section a parametric analysis of mm-wave unit cells is presented at 60 GHz based
on the analytical model. Rogers RT duroid 5880 substrate with rolled copper was used
with these parameters: tanδ = 0.004, r =2.24, copper conductivity σ = 5.85 × 107
siemens/m, copper surface roughness Rrms = 0.3 µm, patch width W = 1.55 mm,
substrate thickness 1 to 20 mil, WR15 rectangular waveguide, and frequency = 60
GHz.
The radiated wave efficiency (η) as a function of substrate thickness for various values
of relative dielectric constants r is plotted in Fig. 3.3. In each case the substrate
thickness is limited to allow only the first order mode of surface waves. A low value
of r allows using relatively thick substrates while still maintaining a good value of
efficiency (η). In case of high values of r a significant amount of power is carried away
by surface waves. Therefore, it is important to select materials with low r values which
also improves the radiation efficiency in antennas.
Radiated wave efficiency (ηrad)

0.8
0.6
0.4 ǫr = 2.24
ǫr = 6.15
ǫr = 10.2
0.2 ǫr = 12.8
0
0 5 10 15 20
Substrate thickness h (mil)
Figure 3.3: Radiated wave efficiency = η as a function of the substrate thickness for
various values of relative dielectric constants. For each r case the substrate thickness
is limited to allow only the first order mode of surface waves.
200
Quality factor values
150
Qo
100 Qext
50
0
0 5 10 15 20
Substrate thickness (h) mil
Figure 3.4: Quality factors Qo and Qext as a function of the substrate thickness. When
Qext ≤ Qo is the over coupled region.
Quality factors Qo and Qext are plotted in Fig. 3.4 against the substrate thickness. For
very thin substrates the value of Qo is less than that of Qext . In this region the unit
cell absorbs almost all the incident power without any significant power reflected back.
This results in an excessive loss in the unit cell. The absorbed power is wasted as heat.
In the region where value of Qo is greater than that of Qext is the useful region for
reflectarray design. A thick substrate is preferred as it offers a smooth reflection phase
curve to attain relatively wider bandwidths. However, choosing too thick substrate
leads to a significant reduction in the achievable phase range.
The reflection coefficient’s magnitude (|Γ|) and the corresponding achievable phase
swing are plotted in Fig. 3.5 and 3.6 respectively against the substrate thickness.
Substrate thickness values less than 3 mil result in a significant loss in reflection
coefficient’s magnitude as can be seen from the Fig. 3.5 where three cases are plotted.
Γ(fring smooth) is for the case when only the fringing fields in a microstrip patch
are considered for the substrate with smooth copper metallic surface. Copper surface
roughness is considered in case of Γ(fring rough). Whereas the effect of surface waves is
added in the case of Γ(fring rough SW). The phase swing for a patch length variation
(∆L) of ±20% and ±100% is shown in Fig. 3.6 with and without the effect of surface
waves while fringing field and copper surface roughness was considered in both cases.
It can be observed from curves that an increase in the substrate thickness decreases the
achievable phase swing for a particular percentage value of ∆L. Although the phase
swing can be extended to some extent by varying the ∆L significantly, doing so makes
the patches highly off resonant therefore lowering the radiation efficiency when used in
a reflectarray. Here, a low r value (2.24) was used therefore, surface wave effects are
minimal. However, these effects are very pronounced for higher values of r .
Reflection coefficients magnitude (dB)

−5
Γ (fring smooth) dB
−10
Γ (fring rough) dB
Γ (fring rough SW) dB
−15
−20
0 2 4 6 8 10
Figure 3.5: Reflection coefficient as a function of the substrate thickness at 60 GHz.

Γ (fring smooth) considers the effect of fringing fields in microstrip and copper
conductivity without its surface roughness. In Γ (fring rough) the effect of conductor
surface roughness is also considered in addition to fringing fields. While in Γ (fring
rough SW) the effects of fringing field, conductor roughness and surface waves are
considered.
350
Phase swing ∆Φ (deg)
300
250
∆Φ (∆L = 0.2L with surface waves)

200 ∆Φ (∆L = 0.2L without surface waves)
∆Φ (∆L = L with surface waves)
∆Φ (∆L = L without surface waves)
150
2 4 6 8 10 12 14 16 18 20
Figure 3.6: Phase swing ∆Φ (deg) as a function of substrate thickness. Two cases are
shown where patch length was changed by 20% and 100%.
Fig. 3.7 displays the reflection coefficient’s magnitude as a function of frequency for 60
GHz resonant patches on different substrate thickness values with a fixed patch width
of 1.55 mm. As the substrate thickness increases, the unit cell loss reduces. For each
thickness value four cases of reflection coefficient magnitude are plotted. Γ (no fring,
smooth, no SW) represent an ideal case where there are no fringing fields generated in
the substrate, copper is smooth, and surface waves are absent. In Γ (fring, smooth, no
SW) the effect of fringing fields is included, while copper surface is still smooth and
surface waves are absent. There is a slight change (increase) in the reflection coefficient’s
magnitude meaning the unit cell is presenting more loss while reflecting the incident
EM field. When the effect of copper roughness is also included, it significantly changes
the reflection coefficient’s magnitude as represented by Γ (fring, rough, no SW). The
fourth curve for each case of substrate thickness is when the effect of surface waves
is also included in addition to fringing fields, and copper surface roughness. Surface
waves improve the reflection loss as is visible from the curves. This is due to the fact
that surface waves exhibit a constructive interference in case of an individual unit cell
therefore, reducing the loss in a unit cell. When surface waves are confined to the unit
cell, this reduction in loss can be exploited as an increase in gain at the reflectarray
level. All these cases are considered for the reflection phase of unit cell in Fig. 3.8.
For low substrate thickness values the effect of surface waves in negligible, however it
becomes noticeable with an increase in the substrate thickness.

20 mil
−0.5
10 mil
−1
−1.5
5 mil
−2
Γ (no fring, smooth, no SW)
Γ (fring, smooth, no SW)
−2.5
Γ (fring, rough, no SW)
Γ (fring, rough, SW)
−3
50 52 54 56 58 60 62 64 66 68 70
Frequency (f ) GHz
Figure 3.7: Reflection coefficient (Γ) magnitude versus frequency for various substrate
thickness values considering the effects of fringing fields, conductor roughness, and
surface waves.
ΦU C (no fring, smooth, no SW)

150 ΦU C (fring, smooth, no SW)
Reflection coefficient’s phase (deg)
ΦU C (fring, rough, no SW)

100 ΦU C (fring, rough, SW)
50
10 mil
0
20 mil
−50
5 mil
−100
−150
50 55 60 65 70
Frequency (f ) GHz
Figure 3.8: Reflection coefficient’s (Γ) phase versus frequency for various substrate
thickness values considering the effects of fringing fields, conductor roughness, and
surface waves.
At 60 GHz for three substrate thickness values i.e. 5, 10 and 20 mil; the reflection
coefficient’s magnitude as a function of patch’s length for a fixed value of patch width
= 1.55 mm is shown in Fig. 3.9. In these curves and curves presented from this point
onwards all the effects (fringing fields, rough copper, and surface waves) are considered.
Thick substrates present a better value of reflection loss. However, a reduced phase
swing is observed as is shown in Fig. 3.10. A reduced phase range in unit cells lowers
the gain of a reflectarray by restricting the available range of phases.
The effect of patch width on reflection magnitude and phase is shown in Fig. 3.11 and
3.12 respectively for 60 GHz resonant patches at various substrate thickness values.
For very narrow width patches, there exist the under coupled condition and unit cell
becomes very lossy. The reflection phase also changes with a change in patch width.
Higher is the substrate thicknes; greater is the change in phase observed.
For 60 GHz resonant patches on various substrate thickness values, the reflection
magnitude and phase are plotted as functions of relative dielectric constant in Fig.
3.13 and 3.14 respectively. It can be observed that there exist anomalies in reflection
phase for 20 mil thick substrate for r values between 5 and 10. Therefore, significant
care is required during the design process.
0
−0.5
−1
−1.5
−2
Γ magnitude (dB), (5 mil)
−2.5 Γ magnitude (dB), (10 mil)
−3
1 1.2 1.4 1.6 1.8 2 2.2
Patch length (L) (mm)
Figure 3.9: Reflection coefficient’s (Γ) magnitude versus change in patch length (L) for
various substrate thickness values.
180

ΦU C (5 mil)
135 ΦU C (10 mil)
90 ΦU C (20 mil)
45
−45
−90
−135
−180
1 1.2 1.4 1.6 1.8 2 2.2
Patch length (L) (mm)
Figure 3.10: Reflection coefficient’s (Γ) phase versus change in patch length (L) for
−0.5
−1
−1.5
−2
−2.5
−3
0 0.4 0.8 1.2 1.6 2 2.3
Patch width (W ) (mm)
Figure 3.11: Reflection coefficient’s (Γ) magnitude versus change in patch width (W )
for various substrate thickness values.

−10
−20
−30
ΦU C (5 mil)
ΦU C (10 mil)
−40 ΦU C (20 mil)
−50
0 0.4 0.8 1.2 1.6 2 2.3
Patch width (W ) (mm)
Figure 3.12: Reflection coefficient’s (Γ) phase versus change in patch width (W ) for
Reflection coefficient’s magnitude (dB)
−1
−2 Γ magnitude (dB), (5 mil)

−4
−5
−6
0 5 10 15 20
Relative dielectric constant ǫr
Figure 3.13: Reflection coefficient’s (Γ) magnitude versus change in substrate relative
dielectric constant(r ) for various substrate thickness values.
20
Reflection coefficient’s phase (deg) 0
−20
−40
−60
−80
ΦU C (5 mil)
−100 ΦU C (10 mil)
−120 ΦU C (20 mil)
−140
0 5 10 15 20
Relative dielectric constant ǫr
Figure 3.14: Reflection coefficient’s (Γ) phase versus change in substrate relative
dielectric constant(r ) for various substrate thickness values.
Loss tangent of a substrate have significant impact on the unit cell losses. The effect of
loss tangent on reflection loss and phase is shown in Fig. 3.15 and 3.16 respectively. For
very lossy substrates there exist a significant loss in a unit cell. Moreover, an increased
loss tangent can lead to an under coupled unit cell therefore, wasting the absorbed
incident power as heat. This is more visible for thin substrates. As the loss tangent is
a loss mechanism, it do not have any impact on the reflection phase.
Finally, the effect of copper conductivity is shown on the reflection loss in Fig. 3.17 for
a 60 GHz resonant patch on various substrate thickness values. Copper conductivity
is plotted as log10 to make the lower conductivity effects more noticeable. For thin
substrates with lower copper conductivity under coupling can exist. This also finds the
conductivity limits for printed unit cells by 3D printers which are gaining significant
attention in recent times.

0
−5
−10

−20
0 0.02 0.04 0.06 0.08 0.1
Substrate loss tangent tanδ
Figure 3.15: Reflection coefficient’s (Γ) magnitude versus change in substrate loss
tangent (tanδ) for various substrate thickness values.
−5
−10
−15
ΦU C (5 mil)
−20 ΦU C (10 mil)
ΦU C (20 mil)
−25
−30
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Substrate loss tangent tanδ
Figure 3.16: Reflection coefficient’s (Γ) phase versus change in substrate loss tangent
(tanδ) for various substrate thickness values.

−5
−10

−20
−25
5 5.5 6 6.5 7 7.5 8
log 10 (Metal conductivity σ) S/m
Figure 3.17: Reflection coefficient’s (Γ) magnitude versus change in metal conductivity
(σ) for various substrate thickness values. Metal conductivity is plotted as logarithm
to show the effect of lower conductivity values leading to sever losses in unit cell for
thin substrates.
Dual polarized unit cells are highly desired to accommodate multiple polarizations.
At mm-waves simplified basic shapes are more desirable as they are more tolerant to
manufacturing tolerances. Therefore, various results for a square patch are plotted here
by setting patch length equal to its width. In these calculations a square waveguide
with both dimensions equal to the wider dimension (a = 3.7592 mm) of WR15 were
used. Fig. 3.18 displays |Γ| versus variation in length at 60 GHz for various substrate
thickness values. From this graph one can choose the length for 60 GHz resonance
for a selected substrate thickness. For a patch on a thin substrate, we need a slightly
bigger length to make it resonant at 60 GHz and vice versa. Moreover, thin substrates
are relatively lossy when used in unit cells. In Fig. 3.19 the |Γ| is plotted for 60 GHz
resonant patches on various substrate thickness values. Similarly, reflection phase is
plotted for these 60 GHz resonant patches against frequency in Fig. 3.20. Only the
resonantor phase is plotted here, the actual unit cell reflection phase would have a
slightly reduced phase range in comparison to the results being displayed here. Fig.
3.21 presents an analytically derived lumped equivalent circuit model of a unit cell.

0
−0.5
−1
−1.5
−2
−3.5
−4
0.5 1 1.5 2 2.5
Square patch length (L) (mm)
Figure 3.18: Reflection coefficient’s (Γ) magnitude versus change in length (L) of a
square patch for various substrate thickness values.
−0.5
−1
−1.5
−2
−3
−3.5
−4
50 55 60 65 70
Frequency (f ) GHz
Figure 3.19: Reflection coefficient’s (Γ) magnitude versus frequency (f ) of a square

patch for various substrate thickness values.
200
Φres (5 mil)
Reflection coefficient’s phase (deg) 150 Φres (10 mil)
100
Φres (20 mil)
50
−50
−100
−150
−200
50 55 60 65 70
Frequency f (GHz)
Figure 3.20: Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch for
various substrate thickness values. Φres is plotted only without considering the effects
of reflections from un-occupied area of the unit cell.
Figure 3.21: An RLC equivalent circuit of a mm-wave unit cell on a 10 mil RO5880 at
59 GHz.
3.4. Electromagnetic Simulation of Unit Cells 86
3.4 Electromagnetic Simulation of Unit Cells
This section describes the electromagnetic simulation of a square unit cell on a 10

mil thick RO5880 substrate using two well known techniques. The first technique is
based on Floquet modes to model periodic boundary conditions under TEM excitation.
Whereas, the second technique is mainly based on the test setup for unit cells. It is
widely accepted to test unit cells’ performance by placing them inside the dominant
mode waveguide. Metallic walls of the waveguide enable infinite periodicity. For floquet
modes simulations, there is a flexibility to vary the wave incident angle, whereas for
waveguide based simulations the incident angle is decided by the waveguide used and
in mainly fixed at a particular frequency. There exist ways to modify the incident angle
in waveguide structures [137], however it is complicated and therefore not widely used.
Fig. 3.22 displays the floquet mode simulation setup for a square unit cell at V band in
CST. For an accurate simulation the reference plane is chosen as the ground plane facing
the reflecting patch as shown by a cross in Fig. 3.22 (c). Electric field configuration
is shown in Fig. 3.22 (d). In a floquet mode simulation under port setting a user
can specify the scan angles, Theta and Phi which are also the incident angles through
reciprocity. For various incident angles the results for reflection coefficient’s magnitude
and phase are shown in Fig. 3.24. A key observation here is by increasing the incident
angle lowers the patch’s resonant frequency as well as an increased reflection loss is
observed.
Fig. 3.23 displays the simulation of a square unit cell using waveguide ports. The unit
cell is placed inside the waveguide excited through a TE10 mode. The incident angle
of a TE10 wave inside a waveguide with broad dimension ’a’ is given by:

−1 λo −1 λo
Θincident = cos = cos (3.47)
λcutoff 2a
The wave incident angle in a waveguide is frequency dependent. For a V band WR-15
waveguide at 60 GHz the incident angle is 48◦ . A comparison of the unit cell’s reflection
performance using floquet ports with incident angle of 48◦ and metallic waveguide is
shown in Fig. 3.24. For the said incident angle a very good agreement of both reflection
magnitude and phase is observed.
(a) Unit cell on a grounded substrate (b) Front view of the unit cell
(c) Unit cell Floquet port assignment (d) Electric field and periodicity
Figure 3.22: Unit cell on a grounded substrate simulated in CST using Floquet ports.
(a) Unit cell inside a metallic waveguide (b) Front view
(c) Electric field configuration of the excitation waveguide port for unit cell in a metallic
waveguide. The reference plan is placed at the patch to make the simulation independent
of waveguide length. Waveguide is considered as PEC. The ground plane, patch and substrate
are considered lossy.
Figure 3.23: Unit cell on a grounded substrate simulated in CST using metallic
waveguide.
−0.5
CST UC in a Metallic Waveguide

CST UC with Floquet Ports 0 deg
−1 CST UC with Floquet Ports 20 deg
−1.5
50 52 54 56 58 60 62 64 66 68 70
Frequency (f ) GHz
(a) Reflection coefficient’s (Γ) magnitude versus frequency (f ) of a square patch on a 10 mil
thick substrate.
200
CST UC in a Metallic Waveguide
150 CST UC with Floquet Ports 0 deg

50
−50
−100
−150
−200
50 52 54 56 58 60 62 64 66 68 70
Frequency (f ) GHz
(b) Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch on a 10 mil thick
substrate.
Figure 3.24: A comparison of unit cell simulated using metallic ports; and floquet ports
with various wave incident angles. Incident angle at 60 GHz is around 48◦ in metallic
waveguide which matches very well with the floquet port simulation for the same angle.
3.5. Passive Unit Cells’ Fabrication and Measurements 90
3.5 Passive Unit Cells’ Fabrication and Measurements
The measurements of a unit cell using waveguide setup posses significant difficulties
where unit cell(s) diced to the waveguide dimensions are placed inside a dominant
mode waveguide. It is often the case when trenches are made in the waveguide shorting
plate to house the unit cell(s). A tiny unit cell’s placement and its correct orientation
in a waveguide at mm-waves becomes a headache where dimensions are too small.
The problem becomes even more severe when DC bias lines are required to be taken
outside the waveguide to provide required DC bias to the reconfiguration mechanism
implemented in unit cells.
To provide correct alignment, right orientation and to facilitate the DC biasing when
required, two structures were proposed which directly interface the waveguide adapter
and do not necessarily require a shorting plat at their rear side [184]. These structures
are shown in Fig. 3.25 (b and c). The only difference between these is the existence
of metalization layer on the substrate face containing reflection patch to provide a
better electrical contact. These unit cell structures can be made large to directly fit
the waveguide interface. By drilling the holes for waveguide alignement pins as well
as for the mounting screws a unit cell can be interfaced with the waveguide at the
correct calibration plane in the correct orientation to provide accurate measurements
at mm-waves.
Figure 3.25: Unit cell types, (a) to be tested by placing inside the waveguide, (b) unit
cell to be mounted to waveguide by screws, and (c) unit cell with extra copper on top
surface to be mounted to waveguide with screws.
(a) Front view of fabricated unit cells (b) Rear view of fabricated unit cells
Figure 3.26: Unit cell on a grounded substrate under test, front and rear views.
Figure 3.27: Waveguide test setup with a unit cell.
Fig. 3.26 shows the fabricated unit cells using the above three test schemes for their
performance comparison. A large number of unit cells were fabricated to characterize
the fabrication facility as well as to verify the test structures. The waveguide test setup
used in measurements is shown in Fig. 3.27. It consists of a coax to waveguide adapter,
a rectangular to square transition and a square waveguide which is finally interfaced
with the unit cell structure. The transparent sheet serves as a support structure for
fragile mm-wave unit cells and does not effect the measurements. Measurements of unit
cell shown in Fig. 3.25 (a) are shown in Fig. 3.28. It was verified during the test that
the presence and absence of the waveguide shorting plate do not effect the performance
for a well diced unit cell.
Figure 3.28: Fabricated unit cells under test.
The performance of a V band square unit cell is shown in Fig. 3.29 and 3.30.
Measurement results for three test setups are in an excellent agreement around the
resonance frequency ( 59 GHz) of unit cell. The predicted analytical and CST simulated
results are shown as well for a comparison. The results for reflection phase are the most
important parameter in reflectarray designs are in an excellent agreement for all the
cases. For reflection magnitude it was observed that the substrate behaves with less
loss than expected which is a desireable feature at mm-waves. In simulations a tanδ =
0.004 and metal surface roughness Rrms = 0.3 µm were used.
Table 3.1 compares the unit cell test techniques shown in Fig. 3.25. Here the unit cell
shown in Fig. 3.25 (a) is taken as a reference for the comparison purpose. There is a
negligible difference in resonance frequency and the reflection phase for the proposed
unit cell test structures. These unit cell structures are very useful in measurements of
active unit cells.
Table 3.1: A comparison of unit cell measurement techniques/structures.
Parameter UC (a) UC (b) UC (c)

Resonant Frequency (GHz) 59.23 59.35 59.04
Frequency deviation (GHz) 0 0.12 - 0.19
Phase deviation (deg) 0 6 -10
Reflection coefficients magnitude (dB) −1
−2
UC type A
UC type B
−3
UC type C
CST
−4 Analytical
−5
50 52 54 56 58 60 62 64 66
Frequency (f ) GHz
Figure 3.29: Reflection coefficient’s (Γ) magnitude versus frequency (f ) of a square

patch on a 10 mil thick substrate. A comparison of measured unit cells’ results, CST
prediction and analytically calculated results
UC type A
150
UC type B
UC type C
100
CST
Analytical
50
−50
−100
−150
50 52 54 56 58 60 62 64 66
Frequency (f ) GHz
Figure 3.30: Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch on
a 10 mil thick substrate. A comparison of measured unit cells’ results, CST prediction
and analytically calculated results

0
Measured reflection loss
−0.5
−1
−1.5
−2
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
Patch length (L) mm
Figure 3.31: Reflection coefficient’s (Γ) magnitude versus patch length (L) for a square
patch on a 10 mil thick substrate at 59 GHz.
150
Reflection coefficient’s phase (Deg)
100
Measured reflection phase
50
−50
−100
−150
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9

Patch length (L) mm
Figure 3.32: Reflection coefficient’s (Γ) phase versus patch length (L) for a square patch
on a 10 mil thick substrate at 59 GHz.
Finally, the measured reflection loss and reflection phase versus patch length for unit
cells of type (c) on a 10 mil RO5880 lossy substrate is shown in Fig. 3.31 and 3.32
respectively. As these unit cells were measured in an array of unit cells with variable
length with a continuous substrate therefore, one observes highest reflections at the
resonant frequency around 1.45 mm at approximately 59 GHz. At off resonance the

array of patches act as a partial sink which partially dissipate the incident energy and
only partial reflections are observed. Based on reflection phase curve a suitable length
can be selected to realize a desired reflection phase from the unit cell at the operating
frequency which is 59 GHz in this case.
3.6 Chapter Summary
In this chapter the most fault tolerant unit cell topology (square/rectangular) for
mm-waves was selected and characterized using a full-wave EM simulation software
and analytically developed formulations. A very fast detailed analysis program was
conceived and implemented to accurately characterize the unit cells. This program
considers the effects of surface waves, metal surface roughness, and finite metal
conductivity in addition to substrate loss tangent (tanδ). It provides a great insight
of the operation of mm-wave unit cells. During the design phase there is a great deal
of optimization required which takes significant time in the 3D EM simulator. By
having a fast analytical program greatly facilitated the characterization of unit cells at
mm-waves. The formulation derived was used in parametric studies of unit cells. It was
observed that for a carefully selected substrate with a low value of dielectric constant,
the effect of surface waves is negligible. However, a pronounced effect is observed for
higher dielectric constants and thick substrates. The developed analytical program was
extended to extract an equivalent RLC circuit model of the unit cell at mm-waves. EM
simulations of a unit cell using metallic waveguide and floquet ports were compared
for various incident angles. The floquet port simulations at an incident angle equal to
the waveguide incident angle were observed in a very close agreement with the metallic
waveguide based simulations.
A large number of passive unit cells were designed and fabricated to characterize their
performance at mm-waves. As a part of this task PCB fabrication facility was also
characterized at mm-waves. The measured results were compared with the results
from 3D EM solver simulations, and analytical analysis predictions. An excellent
agreement was observed in all the cases. Through a series of measurements the
reflection loss and reflection phase curves for a 10 mil thick RO5880 substrate at V
band were prepared which would lead to reflectarray developments in the later chapters.
During the reflection performance measurement of a unit cell the waveguide structure
is used to realize periodic boundary conditions. An accurate alignment/placement
of unit cell(s) in the waveguide structure and provision of DC bias for active unit
cells cause significant difficulty and often requires custom made waveguide parts. Two
alternate solutions which fix the issue of alignment/orientation/placement and ease
the provision of DC biasing to avoid custom designs of waveguide parts are realized.
The measured results of representative unit cells obtained through these new proposed
measurement structures were compared with the traditional waveguide test setup. A
strong agreement was observed for both proposed solutions. Such test structures greatly
simplify the metrology of passive and active unit cells.
Chapter 4
Characterization of Active
Reflectarray Unit Cells
Smart antennas are vital in next generation mm-wave communication systems capable
of extremely high data transfer rates. Currently, available smart antennas are mainly
very low gain solutions. Mainly, smart antennas rely on the beamforming principles
which benefit from phase shifters to control the radiations from array elements. At
mm-waves, phase shifters are very lossy. Similarly it is not feasible to build a large
number of RF chains. Therefore, one need alternate solutions to realize high gains and
wide angle beamsteering at mm-waves. Active reflectarrays are a potential alternate
to beamforming solutions at mm-waves to simultaneously achieve high gain and wide
angle beamsteering. This is the core advantage of reflectarrays which is absent in other
antenna technologies. To make a reflectarray smart, one need to make its comprising
unit cells active/reconfigurable to provide the desired amount of reflection phase. This
chapter is dedicated to realize a practically implementable novel solution for a mm-wave
reflectarray reconfigurable unit cell.
A brief classification of active phase control techniques was presented in section 2.8.2
including the characteristics of underlying control devices. A comprehensive discussion
is made in Appendix A on methods of reconfiguration including electrical/electronic,
optical, smart materials and element rotation. Similarly, in Appendix A
reconfigurable unit cells are explored in more details including polarization selection
97
4.1. PIN Diode Characterization 98
for unit cells, varactor and PIN diode based unit cells, polarization reconfigurable unit
cells, frequency tunable unit cells, and monolithic MEMS based unit cells.
Section 4.1 presents the characterization of an optimally selected technology (PIN

diodes) for mm-waves operation. PIN Diodes’ measurements and extraction of their
RLC parameters at mm-waves is described. This research used PIN diodes as series
switches. In section 4.2 the power consumption and insertion loss of a series PIN
diode switch are presented. The theoretical modeling results are compared with
the measurements data. A greener solution is realized by reducing the DC power
consumption in PIN diodes through an optimal trade-off with their insertion loss. In
section 4.3 the design and characterization of an active unit cell is presented. This unit
cell enables three phase states in each linear polarization where both polarizations can
operating at the same or different frequency. This active unit cell implicitly implements
low loss phase control mechanism directly at the RF plane of the mm-wave reflectarray.
The issue of cross polarization isolation/discrimination which needs significant attention
in active unit cells is resolved. It is the very first time the DC bias line is exploited to
control the XPD of an active unit cell. Finally, the fabrication and measurements of
the active unit cell capable of offering three selectable phase states are discussed.
4.1 PIN Diode Characterization
This section presents the characterization of PIN diodes at X (10 GHz) and V (60
TM
GHz) bands. Two MACOM PIN diodes MA4AGBLP912 (Al-GaAs beamlead)
and MA4GP905 (GaAs beamlead) were selected for this purpose. Both diodes are
characterized upto 30 GHz by the manufacturer. To use them at 60 GHz one need
to measure their equivalent circuit parameters. In this project PIN diodes were used
as series switches inside a unit cell with the reflecting patch. A PIN diode in series
configuration [185] can be represented by its equivalent circuits in ON and OFF states
as shown in Fig. 4.1.
C
R
L L
(a) ON state of PIN diode (b) OFF state of PIN diode
Figure 4.1: Equivalent circuits of a PIN diode in ON (forward bias) and OFF (reverse
bias) states
In ON state a PIN diode can be effectively represented by its forward resistance (R)
in series with an equivalent inductance (L). In OFF state the capacitive reactance of a
PIN diode’s equivalent capacitor is much larger than the forward resistance, therefore
resistance can be ignored. In OFF state the resistor can be effectively replaced by a
capacitor (C) in series with the inductor. An accurate measurement of these equivalent
R, L and C components of a PIN diode at higher frequencies is a challenging task.
In this work, these component values are extracted using the measured S (scattering)
parameters of above PIN diodes at X and V bands. Using insertion loss measurements of
a PIN diode series switch, the corresponding forward resistance values can be derived
at the known DC current. Similarly, using isolation measurements of a series PIN
diode switch, its capacitance value can be extracted. To measure inductance involves
the measurement of equivalent transmission impedance in a series configuration. The
measurement of R, L and C require the PIN diode to be mounted in a test structure. To
characterize the test structure itself is not an easy task at mm-waves. When these test
structures are measured containing the PIN diode, the mounting gap of a PIN diode
presents a capacitance which may lead to incorrect extraction of final L and C values if
not dealt carefully. The characterization of PIN diodes should be on the same substrate
in which it is intended to be used finally. It is also preferred to mount them in the
same configuration as they are to appear in the final circuit. With these constraints,
an issue of test equipment’s availability becomes a concern at mm-waves. For example,
the substrate used for reflectarray unit cells was RO5880 with dielectric constant of 2.2
which resulted in a test probe pitch size of 250 µm which are not available at mm-wave.
Normally, the probe station is used with GaAs substrates with dielectric constant of
12.9 which results in an available probe pitch size of 150 µm. For conformance with
the test equipment the test structure is required to be modified which further limits
the choices and makes these measurements difficult.
4.1.1 A Description of the Measurement Setup
Fig. 4.2 displays the measurement structure for characterization of PIN diodes at V
TM
band. Two MACOM PIN diodes MA4AGBLP912 and MA4GP905 were selected for
their characterization. These diodes are compact to facilitate their accommodation in a
unit cell and are available in beam lead packages to ease the mounting on PCBs. Both
of these diodes have low values of ON state inductance and OFF state capacitance
which is a required feature at mm-waves for fast switching. Fig. 4.2 (a) and (b) are
grounded co-planar microstrip structures for measurements using a probe station. The
DC bias is provided through a pair of λ/4 thin and thick line chokes for its isolation
from the RF circuitry. To make the probe station prongs land on the RO5880 PCB
with a low dielectric constant (2.2) the ends of the co-planar lines were made tapered
to fit the 150 µm probe pitch. In Fig. 4.2 (b) inter-digital capacitors (finger brakes:
FB) were included to block the DC from reaching the test probes. The top co-planar
line in each structure is a short circuit line replacing the PIN diode. The middle line
is an open circuit in place of the PIN diode having a gap width equal to the gap for
PIN diode mounting in unit cells. A PIN diode is mounted in the lowest co-planar line.
The co-planar line containing the PIN diode was replicated multiple times to mount
multiple PIN diodes of each type. Two PCBs of each type as shown in Fig. 4.4 (a-b)
were fabricated to provide multiple measurements. Similar structures for PIN diode
characterization at X-band were prepared using microstrip lines and SMA connectors
mounted at the ends as shown in Fig. 4.3. and Fig. 4.4 (d-e).
(a) Test structure w/o FB (b) Test structure with FB
Figure 4.2: V band PIN diodes test structures with and without DC isolation capacitors
(FB: finger break).
(a) Test structure w/o FB (b) Test structure with FB
Figure 4.3: X band PIN diodes test structures with and without DC isolation capacitors.
The measurement setups at V and X band are shown in Fig. 4.4 (c) and (f)
respectively. For V band measurements the mm-wave probe station at NPL was
used with DC feeding probes. Whereas, at X band simple soldering jigs were used
to hold the DC bias wires in contact with the DC pads on the PIN diode circuit
PCBs. The DC biasing of the measurement setup consists of a PIN diode mounted in
the co-planar/microstrip circuitry in series with the current limiting resistor and an
ampere meter, connected across the DC supply. A selectable resistor network was used
to vary and limit the amount of DC current through the diode. A port of a vector
network analyzer (VNA) connects at each end of the co-planar line through 150 µm
pitch waveguide probes at V band. While for X band the VNA probes directly connect
to SMA connectors at the each end. At both bands calibration lines (thru, reflect and
a phase shift line: TRL) were fabricated as well (not shown here). At V band VNA
calibration was done using on wafer calibration kit while at X band calibration was
done using standard short-open-load-thru (SOLT) kit.
(a) V band with FB (b) V band without FB (c) V band test setup
(d) X band with FB (e) X band without FB (f) X band test setup
Figure 4.4: V and X band PIN diode fabricated PCBs for PIN diode characterization
and their corresponding test setups.
4.1.2 PIN Diode Parameters Extraction
PIN diode models in ON and OFF state were considered as series combination of RL
and LC circuits respectively. Therefore there were effectively three component values
to be extracted from the measured scattering parameters. In this work, insertion loss
measurements were used to extract the value of diode forward resistance. Transmission
scattering parameters S21 in ON and OFF states were used to find L and C values
respectively. A comparison of short circuited path (top line in Fig. 4.2 and Fig. 4.3)
with the path containing diode (lowest circuit in Fig. 4.2 and Fig. 4.3) gives the value
of insertion loss due to diode in ON state. Similarly, a comparison of top short circuited
line with the middle open circuited line provides the estimate of the capacitance due
to diode mounting gap which need to be de-embedded from the measurements. To find
OFF state capacitance S parameters of short circuited line and the line containing diode
in OFF state were compared and C value was found. The mounting gap capacitance is
considered in parallel with the diode capacitance, therefore it was subtracted to find the
final value of diode capacitance. To find the ON state inductance S parameters of ON
state short circuited line were compared with the ON state diode containing line with
mounting gap capacitance de-embedded. The measured S parameters were imported
in ADS software and calculations were performed using equations in data-sets. For V
band the measurement setup was 50 Ω while the test circuits were 86 Ω, therefore
an impedance transformation was done on S parameters before LRC calculations.
Equations 4.1 and 4.2 [186] were used to extract the parameters.

IL/20
Rs = 2 Zo 10 −1 (4.1)
1 − S21
ZT = 2 Zo (4.2)
S21
The extracted values of L and C for both diodes at V and X band are shown in Fig.
4.5 and 4.6 respectively. Forward resistance at V band is shown by Fig. 4.7 while for X
band it is discussed in section 4.2. The values shown are average extracted values for
multiple measurements of the diodes mounted in different locations on the same PCB
as well as on another PCB of the same type.
55
ON state inductance (pH)

50 MA4AGBLP912
MA4GP905
45
40
35
30
25
58 58.5 59 59.5 60 60.5
Frequency (f ) GHz
(a) ON state inductance values for MACOM PIN diodes MA4AGBLP912 and MA4GP905.
32 MA4AGBLP912
OFF state capacitance (fF)
MA4GP905
31
30
29
28
58 58.5 59 59.5 60 60.5

Frequency (f ) GHz
(b) OFF state capacitance values for MACOM PIN diodes MA4AGBLP912 and MA4GP905.
Figure 4.5: Characterization of MACOM PIN diodes MA4AGBLP912 and MA4GP905

at V band.
300
MA4AGBLP912
250 MA4GP905
ON state inductance (pH)
200
150
100
50
0
9 9.5 10 10.5
Frequency (f ) GHz
(a) ON state inductance values for MACOM PIN diodes MA4AGBLP912 and MA4GP905.
200
MA4AGBLP912
MA4GP905
OFF state capacitance (fF)
150
100
50
0
9 9.5 10 10.5
Frequency (f ) GHz
(b) OFF state capacitance values for MACOM PIN diodes MA4AGBLP912 and MA4GP905.
Figure 4.6: Characterization of MACOM PIN diodes MA4AGBLP912 and MA4GP905

at X band.
4.2. PIN Diode Power Consumption and Insertion Loss 106
14
MA4AGBLP912
Forward resistance (Ω) 12 Average value
10
4
58 58.5 59 59.5 60
Frequency (f ) GHz
Figure 4.7: Forward resistance under 5 mA forward current for MACOM PIN diode
MA4AGBLP912 at V band.
4.2 PIN Diode Power Consumption and Insertion Loss
It is a fact that higher DC drives result in a lower insertion loss in PIN diode switches.
How significant is the gain when these switches are operated at higher DC powers?
This section presents the DC power consumption and insertion loss in microwave PIN
diode switches to resolve this delusion. At various drive levels, the insertion loss of a
series PIN diode switch was measured in an X-band prototype. It was found that there
is a very marginal gain in driving these switches at higher DC powers. Findings from
these measurements would result is a significant power saving in next generation high
gain smart antennas where DC power from thousands of these elements can be saved
therefore, leading to an environment friendly solution.
4.2.1 Series SPST PIN Diode Switch
A series SPST PIN diode switch is shown in Fig. 4.8. The series capacitors act as
DC blocks while passing the RF signal. DC feed inductors provide a DC path while
act as RF chokes to stop the RF signal from reaching the DC power source. In a
microstrip realization of these components; capacitors can be modeled as gaps, finger
breaks or interdigital capacitors, whereas RF chokes can be effectively represented by a

combination of high and low impedances. The insertion loss (IL) and power dissipation
of a series PIN diode SPST switch depends on its forward resistance Rs , whereas its
achievable isolation (Iso) is a function of OFF state capacitance Ct [187,188]. Insertion
loss and isolation of a PIN diode series switch are given as:
IL = 20 log (1 + Rs / 2 Z0 ) dB (4.3)
Iso = 10 log (1 + 1/(4 π f Ct Z0 )2 ) dB (4.4)
Figure 4.8: A PIN diode series switch
4.2.2 Insertion Loss of a PIN Diode Switch
As shown in Table 4.1 a series diode has the best DC power consumption and the
worst insertion loss characteristics. Due to this insertion loss, it has worst RF power
consumption. If this insertion loss is well understood and minimized; the series switch
would be a suitable candidate for integration at RF plane of smart antennas which
consume lower DC power. This section explores the insertion loss in terms of Rs for
a series SPST PIN diode switch. An Al-GaAs beam lead PIN diode package from
MACOM ‘MA4AGBLP912’ was used for analysis here. The forward resistance (Rs )
of a PIN diode is a function of forward bias current (If ) [185]. For an impedance
matched PIN diode at frequencies greater than the transit time of intrinsic region (i.e.
f > 1300/W 2 (f in MHz and W in microns) Rs can be calculated as:
Table 4.1: A comparison of pin diode switches
Parameter Series Shunt Series-Shunt

Insertion loss Worst Moderate Best
Isolation Worst Moderate Best
VSWR Moderate Worst Best
RF incident power Worst Best Moderate
RF power dissipation Worst Best Moderate
DC power consumption Best Moderate Worst
Switching speed Worst Best Moderate
Rs = 0.1 + W 2 / (µe + µp ) If τ (4.5)
where W = width of intrinsic region, If = forward bias current, τ = minority carriers

life time, µe and µp are electron and hole mobilities respectively. Here, 0.1 Ω resistance
is due to the chip package and leads. For the selected PIN diode with W = 3 µm,
µe = 8600 cm2 / vs, µp = 400 cm2 / vs, τ = 5 - 10 ns the graph of Rs versus If is
shown in Fig. 4.9. At microwaves / mm-waves for well manufactured PIN diodes,
the value of Rs should be low and in a the similar range as shown in Fig. 4.9. At
this moment one should not get confused with the quoted typical value of Rs in data
sheets as normally it is calculated through insertion loss measurements with possibly
phase only de-embedding, which makes it to appear higher. Another reason is the
impedance matching of the test setup which is usually good for the manufacturers test
setup. Due to If dependence of Rs ; the mathematics to calculate If as a function of
voltage, becomes quite involved. Therefore, diode junction voltage were calculated in
terms of If as:
Vd = ln (1 + If / I0 ) n k T / q + If Rs (4.6)
where I0 = 1×10−14 A, n = 2, k = boltzman constant, T = room temperature 300

K, q = electron charge. The value of If was varied from 10 µA to 40 mA (maximum
diode current) and the corresponding values of Rs for τ = 5 ns were used. A graph of
current versus junction voltage is shown in Fig. 4.10. Although, here the diode voltage
variable was made a dependent parameter to make calculation easier; it is still shown
as abscissa to be consistent with the literature. Here two curves with and without the
drop due to Rs are shown. The drop due to Rs even for allowable maximum current is
not huge as indicated by the closeness of these curves. If these two curves are closely
following each other, one should expect low insertion loss because these curves only
differ from one another by the voltage drop across Rs . The insertion loss of a series
matched diode with circuit characteristic impedance Z0 = 50 Ω is given by (4.3).
2.5
τ = 5 ns
2
Forward resistance Rs (Ω)
τ = 10 ns
1.5
0.5
0
0 5 10 15 20 25 30 35 40
Forward bias current If (mA)
Figure 4.9: Forward resistance (Rs ) as a function of forward bias current (If ).
A plot of insertion loss is shown in Fig. 4.11 using current from 1 mA to 40 mA and
corresponding values of Rs for τ = 5 ns. Looking at the curves slope after 5 mA,
there is only a very slight change in the insertion loss even when the current reaches its
maximum allowed value of 40 mA. There is a change of only 0.0303 dB in insertion loss
when current increases from 5 mA to 40 mA. It means for well-constructed microwave
/ mm-wave PIN diodes with low value of Rs one would expect a minimal insertion
loss. Secondly, when impedance matching is perfect one would achieve this minimal
insertion loss. This also highlights the importance of impedance matching in PIN diode
switching circuits. Although, it was concluded after discussions with the manufacturer
of this PIN diode that Rs is higher than what is calculated through intrinsic region’s
width. Nonetheless, it provides a good starting point for a design. On the other hand
manufacturer specified Rs in data sheet is a typical value or an upper bound which a
user may not experience.
40
(mA)
30
f
Drop due to R included

Forward bias current I
s
Without R drop
s
20
10
0
1 1.1 1.2 1.3 1.4 1.5 1.6
Diode forward voltage drop Vd (V)
Figure 4.10: PIN diode current and voltage drop in forward bias.
0.2
Insertion loss IL (dB)
0.15
0.1
0.05
0
0 5 10 15 20 25 30 35 40
Figure 4.11: Insertion loss as a function of forward bias current.
Insertion loss not only attenuates RF signal, it also results in heat generation through
power dissipation inside a PIN diode. The power rating of a PIN diode is the amount of
power when dissipated would increase PIN diode junction temperature to its maximum
allowable value (175◦ C in this case) from the ambient temperature. How much power
a PIN diode is consuming is found by:
2
Pd = Irf Rs + Idc Vdc (4.7)
Here, Irf is the RF current, Idc is the forward bias DC current (If = Irf + Idc ) and
Rs is the value of forward resistance at chosen Idc . For small signal situations (where
Idc ≈ If ) Irf is usually small, however exists in the circuit purposefully. The second
part in this equation is the DC power consumption which should be minimized to avoid
excessive power consumption in large arrays to prevent thermal issues. By reducing
the amount of Idc beyond a certain limit would result in an increased value of Rs which
would eventually increase the insertion loss and RF power consumption. It is required
to avoid this situation to have lower values of insertion loss through a switch. For this
purpose one would need to have a set of curves to decide where to operate / bias the
PIN diode.
4.2.3 PIN Diode Series Switch Measurements
For measurement purposes multiple sets of series PIN diode SPST switches were
fabricated using 0.787 mm thick RO5880 substrate. One of these sets is shown in
Fig. 4.12 (a). In each set there are six sub circuits. Each sub circuit consists of 50
Ω matched transmission lines at both ends connected to SMA connectors. Interdigital
capacitors are used as DC blocks on either side of centrally mounted PIN diode. A
series combination of λg /4 (λg = substrate wavelength at operating frequency) high and
low impedance transmission lines realize DC feeds to isolate RF from DC power source.
The top first sub-circuit has a shorted transmission line in place of the PIN diode. By
acting as a through measurement this provides a fairly good estimate of RF loss in
sub-circuit without a PIN diode. Therefore, one does not have to necessarily rely on
phase only de-embedding. From top the second sub-circuit has a gap in place of a PIN
diode. PIN diodes are mounted in sub-circuits 3 to 6 which are not visible anyway due
to their microscopic physical appearance. Fig. 4.12 (b) shows the measurement setup
for one of the sets where a sub-circuit is under test. Current meter, current limiting
resistors (a set of) and the PIN diode are all in series connection, finally connected to
a DC power supply. To land DC bias lines at pads of DC feeds soldering jigs were used
during X band measurements.
Figure 4.12: Fabricated PIN diode switch circuit PCB, and measurement setup.
In Fig. 4.13, the insertion loss measurement results are shown. Here, the insertion loss
was normalized with the loss of short circuited first sub-circuit. These measurements
were made by varying the voltage which eventually varies the current flowing through
the PIN diode when current limiting resistor is fixed. For curves with current ≥ 26
mA, current limiting resistor value was 240 Ω, while for rest of the curves a 357 (=330
+27) Ω resistor was chosen. All the resistors used were commercially available gold
(±5%) tolerance band. At center frequency of 10.33 GHz, the insertion loss is less than
0.5 dB. Forward resistance (Rs ) derived from measured insertion loss (in Fig. 4.13) is
shown in Fig. 4.14 at 10.33 GHz.
0
2.25 V 2 mA
2.5 V 2.6 mA
−0.2
2.75 V 3.2 mA
3 V 3.8 mA
Insertion loss I L (dB)
−0.4 3.5 V 4.8 mA
4 V 6.6 mA
4.5 V 7.3 mA
−0.6 5 V 9.5 mA
6 V 12 mA
−0.8 7 V 15 mA
8 V 18 mA
9 V 21 mA
−1 10 V 23 mA
8 V 26 mA
9 V 30 mA
−1.2
10 V 35 mA
Short circuit
10.2 10.25 10.3 10.35 10.4 10.45 10.5

Frequency (GHz)
Figure 4.13: Measured insertion loss of a series PIN diode switch with variable DC
voltage and fixed values of the current limiting resistor.
0.6 6
Forward resistance Rs (Ω)

Insertion loss I (dB)
Forward resistance R s (Ω)

L
(dB)
0.4 4
L
Insertion loss I
0.2 2
0 0
0 5 10 15 20 25 30 35
Forward bias current I f (mA)
Figure 4.14: Measured insertion loss and corresponding forward resistance at 10.33
GHz with variable DC voltage and fixed values of the current limiting resistor.
It is worth mentioning here that the value of Rs is only few ohms for a considerable
range of If . Simultaneously, diode junction voltages are unknown precisely over the
range of If . When these are coupled with high tolerance current limiting resistors, the
precise calculation of Rs by ohm law becomes invalid as there are too many unknown
variables without precision. The value of Rs gets dominated by the tolerance of current
limited resistors. Therefore, the easiest way to find Rs is through the insertion loss
measurements.
Fig. 4.15 displays the insertion loss results of another fabricated set. In these
measurements, the DC supply voltage was fixed at 5 V and current limiting resistors
were selected to produce the shown forward bias current values. This demonstrates the
effects of comparatively bigger step changes in forward bias current on insertion loss.
Fig. 4.16 displays the insertion loss and corresponding forward resistance for this case.
Very comparable normalized insertion loss results to Fig. 4.13 were found through
these measurements for corresponding forward bias current values in either case.
−0.2
−0.4
−0.6
5 V 5 mA
−0.8 5 V 10 mA
5 V 15 mA
−1 5 V 19.5 mA
5 V 24 mA
−1.2
Short circuit
10.25 10.3 10.35 10.4 10.45

Frequency (GHz)
Figure 4.15: Measured insertion loss of a series PIN diode switch with fixed DC voltage
and selectable variable values of the current limiting resistor.
0.3 3

Forward resistance R (Ω)
s
Forward resistance R s (Ω)

0.2 2
0.1 1
4 6 8 10 12 14 16 18 20 22 24
Figure 4.16: Measured insertion loss and corresponding forward resistance at 10.33 GHz
with fixed DC voltage and selectable variable values of the current limiting resistor.
Forward voltage drop of PIN diode considering the effect of measured Rs (as calculated
from measured If ) was calculated using (4.6) and shown in Fig. 4.17. These calculations
used the results from Fig. 4.13 and 4.14. There is considerable change in diode voltage
drop when the effect of Rs is accounted. DC power consumption versus insertion loss
of the PIN diode as shown in Fig. 4.18 was calculated using forward voltage drop
with measured values of Rs and current values from Fig. 4.17. It can be ascertained
from Fig. 4.18 that for small signal RF / Microwave / mm-wave PIN diode switches a
considerable amount of DC power can be saved at the cost of not a huge insertion loss.
To illustrate the saving in power consumption, consider a smart antenna with 25,000
elements each embedding a single PIN diode switch operating at 30 mA (to achieve an
insertion loss of 0.1 dB). The power drawn from DC source and dissipated as heat by
diodes would be approximately (30 mA × 1.5 V) × 25,000 / 2 = 562.5 W. On the
other hand if diodes are operated at 5mA (with an insertion loss of 0.3 dB), the power
consumption would reduce to (5 mA × 1.40 V)× 25,000 / 2 = 87.5 W only.
35
30 With voltage drop due to measured R

s
Without voltage drop of Rs
With voltage drop due to calculated Rs
25
20
15
10
0
1.34 1.36 1.38 1.4 1.42 1.44 1.46 1.48 1.5 1.52
Diode forward voltage drop Vd (V)
Figure 4.17: PIN diode current and voltage drop in forward bias calculated based on
corresponding values of Rs from the measurements.
0.5
0.4 Diode power consumption P (mW)

d
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Diode power consumption Pd (mW)
Figure 4.18: DC power consumption and insertion loss of a series PIN diode switch.
4.3. Active Unit Cell Characterization 117
This investigation presented the measurement of insertion loss for a PIN diode series
SPST switch at X-band. Based on V band measurements (not presented here in full
detail) a similar argument can be made for V band. It was found that forward resistance
calculations based on intrinsic region width provide a good starting point for PIN diode
switch design. However, one needs to precisely characterize the insertion loss through
measurements. Through various calculations based on measured data the DC power
consumption in a PIN diode series switch was found. It was concluded that at a
tolerable insertion loss, considerable amount of DC power can be saved. When such
power saving is made in large smart antenna arrays containing thousands of elements at
mm-waves, it would have certain benefits including realization of large smart antennas
for small satellite platforms where electrical power is at economical, and in terrestrial
backhauls where cooling costs a significant amount of energy. Similarly, this would be
a significant step towards the realization of greener smart antennas for small platforms.
4.3 Active Unit Cell Characterization
4.3.1 Active Unit Cell Design
At mm-waves the unit cell’s physical size becomes tiny and it can not accommodate
multiple electronic devices. The DC biasing of multiple electronic components in a
small space present a serious challenge and often results in degrading the polarization
purity. A continuous reflection phase control in excess of 360◦ further complicates
the array design. Normally, varactor diodes are used at lower frequencies to achieve a
continuous phase control. Presently, these diodes are not available at mm-waves. In
case a designer is ambitious to use these low frequency diodes at mm-waves there are
mainly two problems in addition to the provision of variable DC control bias. These
are their size and achievable capacitance range at mm-waves. Their size do not let
them get integrated in a tiny unit cell. If the integration aspect is overlooked, their
capacitance range is very limited which limits the achievable reflection phase range at
mm-waves. Therefore, a varactor diode based continuous reflection phase solution is
not optimum at mm-waves.
To conceive an implementable solution at mm-waves, the unit cell design needs

significant simplifications. Therefore, a discrete reflection phase control is more
appealing at mm-waves. The discretization of the reflection phase introduces phase
quantization error which results in a gain reduction, higher sidelobes and increased
pointing error in an array [144–146]. It has been demonstrated that sidelobes and
pointing error can be optimized for a large array by randomizing the fewer available
phase states [189]. Too coarse quantization severely degrades gain performance.
Therefore, an optimum solution is desirable at mm-waves. One bit (two states) phase
quantized array implementations have been demonstrated by [142, 143, 190]. This
research presents a unit cell design which can achieve three phase states and results
in an overall degradation of 1.6 dB in gain as compared to 4 dB for one bit phase
control designs. Any three chosen phase states can be designed out of a 360◦ reflection
phase range. This unit cell design is very flexible in terms of dual polarization and
dual frequency band operations. This architecture enables simultaneous scanning of
both far field linear polarizations. These linearly polarized mm-wave antenna beams
can have same or different operating frequencies, where one polarization beam can
be used as receive and the other as transmit; an excellent practical case for satellite
communication.
The unit cell design controls four switching PIN diodes by a single DC line, thereby
significantly reducing the complexity arising due to DC biasing of multiple electronic
elements. It is more practical to mm-waves although the lower frequency bands can
benefit from the design as well. This architecture integrates the phase shifter circuitry
in a mm-wave unit cell implicitly, thereby reducing the losses due to separate additional
phase shifters and thus eradicating their associated problems. Such an implementation
is directly controlling the mm-wave signal at the RF plane of a reflectarray antenna to
minimize the issues arising due to a large number of antenna elements at mm-waves. To
use PIN diodes in such a configuration to realize three phase states for each polarization
and frequency band by using only one DC bias line is the prime outcome of this PhD
research.
The design is shown in Fig. 4.19 where two PIN diodes for each linear polarization are
connected in reverse order and a single DC line that can assume three voltage levels
(+5, 0, -5 V) at array controller are shown. Diodes D1X and D2X are for one linear
polarization whereas D1Y and D2Y control the orthogonal polarization. The pair of
diodes associated with each linear polarization can be either OFF, or one of these is
ON to provide a particular selected phase state. The state of a diode is controlled by
its DC bias. This arrangement of PIN diodes and DC bias voltages therefore produces
three distinct selectable reflection phase states of the unit cell to implement an implicitly
integrated phase shifter for each linear polarization.
Figure 4.19: Active unit cell using PIN diodes.
An important aspect of the design is PIN diode directions w.r.t. their DC biasing which
make the operation in three phase states possible. A common DC bias line for all these
four diodes is used to provide DC power. All the reflectarray unit cells share a common
DC return line (ground). A DC bias controller can be used to control the operation of
PIN diodes switching states. DC bias controller can select one of the three available
voltage levels for a DC bias line. These available voltage levels are 0V (ground), +5V,
and -5V. One could have thought of the complexity introduced at DC bias controller
due to inclusion of negative voltage; however, it greatly minimizes the complexity in RF
design of a mm-wave reflectarray at the expense of relatively manageable DC biasing
control circuitry at DC bias controller which in itself is not an RF unit and operates
at DC levels through control commands issue by the array controller computer.
When DC bias line of a unit cell is connected to ground (0V); all the PIN diodes are
OFF. It is the first state for reflection phase. When a DC bias line is connected to
+5V, diodes D1X and D1Y are forward biased whereas other two diodes are reverse
biased. PIN diode mm-wave equivalent circuits in ON and OFF state are different.
When respective equivalent circuits replace the PIN diodes, a different reflection phase
is achieved as compared to the first state. This forms the second state of the reflection
phase. Phasing lines (L1X and L1Y ) can adjust the required reflection phase. These
lines for horizontal and vertical polarizations can have different lengths. An additional
degree of freedom for frequency and phase selection is obtained when unit cell employs
a different design for each polarization. When a DC bias line is connected to -5V, the
D2X and D2Y are forward biased whereas other two diodes are reverse biased. This
represents the third phase state. Again the lengths of phasing lines (L2X and L2Y ) can
be adjusted as desired. By the procedure described above, one can achieve effectively
three phase states with a single DC bias line per unit cell to simultaneously control two
orthogonal linear polarizations .
These active unit cells provide a flexible design to independently configure each
polarization at a different frequency of operation during the design process. The
reflection patch dimensions and phasing lines (stub lengths: L1X , L2X , L1Y , and L2Y )
can be adjusted to synthesize the required phase states at the desired frequency of
operation. Diodes’ placement, reflecting patch geometry, and shape of stubs is not
fixed in the way as shown. They can take any form and are acceptable as long as they
fulfill the purpose of reflecting the desired polarization field at the desired frequency.
The function of ∆X and ∆Y offsets in DC bias location is discussed in section 4.3.3.
4.3.2 Active Unit Cell Simulations: Using PIN Diodes
This section presents an example design and its simulation for an active unit cell
discussed in the previous section. The design starts by selecting the reflection phases
required from a unit cell. How this selection effects the performance shall be discussed
in the reflectarray design. Here one can start by choosing any three phase states in the
range of 0◦ to 360◦ . Out of many, one combination of three phase states is {0◦ , 90◦ ,
and 180◦ }. These states are related to unit cell’s calculated continuous phase shift by:

π π

0, − ≤ ∆ΦU C % 2π < State 1
4 4




π π 3π

∆ΦU C = , ≤ ∆ΦU C % 2π < State 2 (4.8)
Q 
 2 4 4
5π π



π, −
 ≤ ∆ΦU C % 2π < − State 3
4 4

where ∆ΦU C is the quantized reflection phase shift introduced by unit cell, ∆ΦU C is
Q
the non-quantized required reflection phase from the unit cell found from reflectarray
design assuming continuous phase shift is available, and % represents the modulo
(remainder) operator.
A square reflecting patch design on a 5 mil thick RO5880 substrate was selected as the
basic design to be made reconfigurable. A simple geometrical shape was selected to
account for the manufacturing tolerances. ON and OFF state models of the selected
PIN diode were integrated using lumped elements in the unit cell in CST Microwave
Studio simulations. Fig. 4.20 displays various views of the designed active unit cell
containing four PIN diodes to realize dual polarized symmetrical behavior at mm-waves.
The unit cell offers three selected phase states in each linear polarization. Via holes are
shared by neighboring unit cells when the unit cell is placed in an array, therefore only
a half portion is shown. Shared via holes is another feature of this unit cell design. It
shrinks inter element spacing; a highly desirable feature to achieve wide angle grating
free beam scans. It also reduces the number of via holes required.
(a) Front view with lumped element diodes (b) Isometric view
(c) Rear view showing RF-DC isolation (d) Rear view: removed substrates
(e) Front vie:, patch, substrates removed (f) Side view, showing layer stack up
Figure 4.20: Various views of a mm-wave active unit cell using PIN diodes for
reconfiguration to implement a three state integrated RF plane implicit phase shifter
on a 2.7 mm square lattice. Via holes are shared by neighboring unit cells when placed
in an array, therefore only half portion is shown.
Here a unit cell consists of a three layer PCB design. Its top layer contains a square
patch, phasing lines, and PIN diodes. The DC bias for PIN diodes in provided through
a single bias line. A DC bias line is taken to the bottom layer using a plated through
hole in the central region of the patch. A DC pad (circular ring here) and quarter
wave stubs are placed in the bottom layer. DC pad serves the purpose of a contact
to provide DC bias to PIN diodes, whereas quarter wave stubs provide the necessary
RF-DC isolation. Stub and DC pad can take any reasonable shape and are not fixed.
Middle layer in the design is for ground. There is a hole in the central region to let the
DC plated through hole via pass through it without an electrical contact. The other
four via holes are also plated vias which are connected to the ground layer. These
vias were made plated through holes to allow easy manufacturing by avoiding blind
vias. Both substrates here are 5 mil RO5880. However, the lower substrate can be a
different/thicker to allow the strength without effecting the design. While selecting the
thickness of lower substrate one need to be within the manufacturer specified limits of
drill hole diameter to its depth for drilling the vias. The minimum drill-able size of
a via hole, its required pad and its positional accuracy are stringent constraints from
manufacturers which often negatively impact the design at mm-waves. To bond both
substartes togather Rogers bondply RO2929 was used.
Full wave electromagnetic (FW-EM) simulations were performed to optimize the design
using CST microwave studio. Optimization of reflection loss, phase response, and
polarization purity of a reconfigurable unit cell in its all operating states present a
serious challenge. Numerous simulations were performed to ultimately converge on an
optimum design which achieves lower reflection loss, closely agreed phase states, and
high polarization purity in all three states. However, the solution is not unique as there
are multiple degrees of freedom that can be exploited as per requirements.
State 1 represents the case when all diodes are OFF. State 2 is for the case when
D1X , D1Y are ON, and D2X , D2Y are OFF. Whereas State 3 indicates the situation
when D1X , D1Y are OFF, and D2X , D2Y are ON. Due to the symmetric design of the
unit cell, the behavior in both polarizations is similar. Therefore, results for only one
polarization are included here. Fig. 4.21 displays the reflection loss magnitude for
these states. This is the loss due to PIN diodes inclusion only. Losses incurred due to
substrate loss tangent, finite conductivity, and roughness of reflecting metallic surface
are additional to the losses presented here. The reflection phase response of the unit
cell is shown in Fig. 4.22 in relation to operational states. A static phase offset of
around 100◦ is observed which is due to Floquet ports excitation modeling in CST
simulations. However, the relative phase of three states can be observed from markers
in the phase response curves.
To realize a desired reflection response in active unit cells is a challenging task which
would be more clear when the cross polarization optimization is discussed in the next
section.
Figure 4.21: Reflection loss magnitude response of an active unit cell using PIN diodes.
Figure 4.22: Reflection phase response of an active unit cell using PIN diodes.
4.3.3 Cross Polarization Isolation in a Dynamic Environment
Polarization purity is one of the major design concerns in microstrip printed antennas.
Reconfigurable antennas have multiple operating states. Reconfiguration of various
states share many common features of the unit cell structure therefore, these operational
states are not completely independent of each other. To achieve high polarization
purity in all these states present a significant design and optimization challenge where
reaching an acceptable cross polar field’s level in one state may lead to disturbing the
cross polarization in rest of the states. This section addresses the issue of improving the
polarization purity of a millimeter wave (mm-wave) reconfigurable unit cell intended
for a smart reflectarray. A design topology is presented where a unit cell can have
three operational states for each linear polarization operating at the same or different
frequency bands. A single DC line is used to bias four switching devices for simultaneous
dual polarization and dual frequency operation. A novel technique to control the
magnitude of cross polar fields is presented. DC biasing usually deteriorates the
performance. High polarization purity is achieved in all the three states of this
multi-state reconfigurable unit cell by exploiting the DC bias line for the very first
time.
Polarization purity is a serious concern in dual polarized reflectarrays, particularly

for the cases of high power transmission. Pencil beam reflectarrays can achieve
good polarization purity at boresight, whereas contoured beam reflectarrays present
a significant challenge; where the cross polarization level is required to be reduced over
a wider satellite footprint [147]. In a good reflectarray design, the polarization purity
issue is addressed at unit cell level as well as at reflectarray level by manipulating the
electromagnetic (EM) fields and associated currents. In order to obtain quasi-zero levels
of cross polarized fields in a unit cell, one needs to consider its polarization reflection
parameters constituted by incident and reflected waves corresponding to transverse
electric (TE) and transverse magnetic (TM) fields in an assumed periodic environment.
The polarization reflection matrix (P RM ) of a unit cell; having a practical electrical
size, in a periodic environment is:

T Er T Er
 
   T Ei T M =0 T Mi T Ei =0 
i
P11 P12  
P RM = = (4.9)
   
 
P21 P22 
 T Mr

T Mr



T Ei T Mi =0 T Mi T Ei =0

where subscripts i and r represent incident and reflected waves respectively. In a unit
cell which is not intentionally designed for polarization conversion, the polarization of
incident and reflected fields should be preserved with cross coupling terms (P12 and
P21 ) being zero. However, in a practical unit cell these cross coupling terms are present
and issue gets worse when a unit cell is made reconfigurable.
At unit cell level, a symmetric structure where possible along electric (E) and magnetic
(H) planes is helpful in canceling the cross polarized field components. In order
to control cross polarized fields, the geometry based parallelogram and trapezoidal
transformations were proposed by [147] at unit cell level. In reflectarrays, the cross
polarized fields’ cancellation at boresight can be achieved by dividing the array aperture
into four quadrants and arranging the unit cells in each quadrant to exhibit mirror
symmetry [148]. However, cross polarized fields near boresight still remains noticeable.
By applying mirror symmetry at sub-array level, the localized cancellation of cross
polarized currents was demonstrated by [149] by the application of mirror symmetry
to every four unit cells. Doing so significantly reduced the cross polarized field level
throughout the radiation pattern of a passive reflectarray.
The above techniques are mainly applicable to unit cells which are passive and
non-reconfigurable. When a unit cell is reconfigured to produce a different phase
response than the originally set phase, the transformation techniques being fixed in
nature do not perform well. Similarly this is true for mirror symmetry as there are
high chances for it to get violated during reconfiguration of a unit cell’s reflection phase.
Non-availability of reconfiguration devices and restricted geometrical features of unit
cells at mm-waves impose significant difficulties in design and performance optimization
of unit cells. Therefore, mm-wave reconfigurable unit cells need development of further
techniques to improve the polarization purity.
(a) Both diodes OFF, Vdc = 0V
(b) D1 ON, Vdc = 5V
(c) D2 ON, Vdc = - 5V.
Figure 4.23: Co and cross polarization reflection magnitudes when reconfiguration takes
place. Only single polarization results are shown due to symmetry.
A square patch can achieve good cross polarization isolation when used without
reconfiguration. When diodes and phasing lines are connected with square patch the
cross polarization isolation becomes worse. Fig. 4.23 displays the behavior of co and
cross polar field’s reflection magnitudes in the three state integrated phased shifter unit
cell. It can be observed that unit cell exhibit an increased loss in co-polar reflected field.
The loss in cross polar field becomes either comparable or even less than the co polar
reflected field. In either case the polarization purity is lost. Polarization purity is not
only the issue of this unit cell design, it was observed for other unit cell design too. All
the reconfigurable unit cell structures tried during design phase failed the polarization
purity check. It took a significant amount of efforts to figure out the solution to improve
polarization purity of reconfigurable unit cells.
The current distribution of passive/active patches with stub lengths was studied
to understand the behavior for various switching states. An imbalanced structure
generates higher cross polarized fields. A common variable that could lead to the control
in all switching states was pointed out and it was the DC bias line. This was the last
resort that was exploited after trying various reconfigurable unit cell topologies. In the
search of a better cross polar isolation in all the reconfiguration states, the DC bias line
was offset from the reflecting patch center. In this unit cell design because stub lengths
on either side of the patch e.g. (L1X and L2X ) and diode state models (ON/OFF)
were different resulting in an asymmetric structure which lead to the generation of
cross polar radiating modes. By offsetting the DC bias point by an amount ∆X and
∆Y , these undesired modes get reasonably suppressed for both linear polarizations.
This leads to a good cross polar isolation. The goal is to find optimum values of ∆X
and ∆Y which lead to an acceptable cross polar isolation in all the desired reflection
phase states. It was found during optimization that the final values are not the best
for individual states. There is a trade-off among the reflection phase states which leads
to an acceptable best compromise in all the cases.
Fig. 4.24 displays parameteric analysis for one of the reflection phase states (when
D2X , D2Y ON). The analysis for other two states is performed in the similar way and
is not presented here. The reflection coefficient (magnitude & phase) in the desired
polarization and its cross polar isolation are monitored for all the states simultaneously.
(a) Co polar reflection coefficient’s magnitude for various positions of ∆X and ∆Y in mm
(b) Co polar reflection coefficient’s phase for various positions of ∆X and ∆Y in mm
(c) Crooss polar reflection coefficient’s magnitude for various positions of ∆X and ∆Y in mm
Figure 4.24: Co and cross polarization reflection coefficient behaviour for various values
of offsets ∆X and ∆Y in mm: D2X , D2Y ON. Only single polarization results are shown
due to symmetry.
Figure 4.25: Reflected co and cross polarized field magnitude response of an active unit
cell using PIN diodes after optimization.
For three states there would be nine total simulations for each value of ∆X and ∆Y . It
was a cumbersome situation to perform the parametric analysis and find the acceptable
offset. The acceptable offset should produce a reasonably low loss, the desired phase
states, and a good cross polar isolation in all the selected states. Once the analysis
is complete the resulting cross polar isolation performace is shown in Fig. 4.25. The
reflection coefficient (magnitude & phase) performance presented in Fig. 4.21 and 4.22
is the optimized performance. One could have thought to perform the optimization in
the EM simulator instead of parametric study, however, the convergence would have
taken much longer than the time required for parametric study without a guarantee to
get an optimum solution.
This method of achieving symmetric structure for cross polar isolation improvement
is generic and is applicable to other unit cell configurations which have some form of
common control in all the reconfiguration states.
4.3.4 Fabrication and Measurements
The unit cell basic structure used for manufacturing purpose is shown in Fig. 4.26.
A lattice size equal to the waveguide (WR15 square) inner dimensions was used in
simulations. Two via hole pads were changed to square pads where the holes are offset
from the pad center to make it manufacturing process compatible. The layer stack
up was also changed due to manufacturing reasons. Here, the top substrate is the 5
mil thick RO5880 with a ground plane. The second substrate was removed. Roger’s
bondply RO2929 with 102 µm thickness was used as the second substrate for printing
the RF-DC isolation stub on the bottom surface. A through hole in the ground layer
is offset as desired by XPD improvement and DC bias provision. It was simulated in
CST microwave studio using ON-OFF state models of PIN diodes and the resulting
performance is shown in Fig. 4.27 to 4.29. The modified unit cell design was optimized
to achieve the required three phase states while providing a low value for XPD. XPD
values can be further optimized, however the present value was considered good enough
to proceed for manufacturing.
The active unit cell’s structure as shown in Fig. 4.26 was embedded in the test structure.
This test structure was prepared for mounting the active unit cell at the end of a square
WR15 waveguide for making the measurement possible and provision of DC bias easy.
As a result, the unit cell’s full structure used for manufacturing is shown in Fig. 4.30.
Various views of the full structure are shown with zoomed views. To bring the electrical
ground at the top surface of unit cell, plated through hole vias were used as shown in
the Fig. 4.30 (d & f). This full structure was again simulated in CST microwave studio
using waveguide structure (waveguide ports). Fig. 4.30 (c & f) displays the resulting
mesh for the structure.
(a) Front view with lumped element diodes (b) Isometric view, substrate and bondply
(c) Rear view showing RF-DC isolation (d) A view showing hole in ground
Figure 4.26: Various views of a mm-wave active unit cell for manufacturing. Unit Cell
simulated in CST using Floquet ports.
0
−0.2
−0.4
−0.6
Diodes OFF
−0.8 D1XY ON
−1 D2XY ON
−1.2
−1.4
−1.6
−1.8
58 58.5 59 59.5 60 60.5 61 61.5 62
Frequency (f ) GHz
Figure 4.27: Reflection coefficient’s magnitude for the active unit cell, simulated in
CST for manufacturing.
−50

Diodes OFF
−100 D1XY ON
D2XY ON
−150
−200
−250
−300
−350
−400
58 58.5 59 59.5 60 60.5 61 61.5 62
Frequency (f ) GHz
Figure 4.28: Reflection coefficient’s phase for the active unit cell, simulated in CST for
manufacturing.
0
Cross polar discrimination (XPD) (dB)
Diodes OFF
−5 D1XY ON
D2XY ON
−10
−15
−20
−25
−30
−35
−40
58 58.5 59 59.5 60 60.5 61 61.5 62
Frequency (f ) GHz
Figure 4.29: Reflection coefficient’s XPD for the active unit cell, simulated in CST for
manufacturing. XPD can be further optimized for OFF state.
(a) Front view (b) Rear view (c) Front meshed view
(d) Front zoomed view (e) Isometric zoomed view (f) Zoomed mesh view
Figure 4.30: Various views of a mm-wave active unit cell for manufacturing. Unit Cell
simulated in CST using waveguide ports. The unit cell structure is made big to be
compatible with the waveguide test setup.
(a) Incident field for mode 1 (b) Surface currents for mode 1
(c) Incident field for mode 2 (d) Surface currents for mode 2
Figure 4.31: Incident electric field for two modes (TE10 and its orthogonal) and unit cell
surface currents for D1X and D1Y ON. Red color represents maximum while minimum
is represented by blue.
Fig. 4.31 (a & b) display the electric field for mode 1 (horizontal polarization) and
resulting surface currents in the unit cell when D1X and D1Y are in ON state. Various
colors for surface currents represent the relative intensity of surface currents: red is
the highest while blue is the lowest. From Fig. 4.31 (b) it can be observed that for a
horizontally polarized incident electric field, the horizontal parts of the unit cell exhibit
significantly higher values of current density in comparison to the vertical parts. These
surface currents produce far field of the antenna. Therefore, one can observe a good
cross polarization discrimination from these surface current values. Fig. 4.31 (c & d)
display the incident electric field for mode 2 (vertical polarization) and its resulting
current density at the unit cell’s surface. A similar trend of good XPD can be observed
here as well.
The active unit cell’s full structure was simulated for lossless and lossy cases. In the loss
less case all the structure was loss less except for the losses due to PIN diodes itself. In
the lossy case, the top substrate RO5880 was considered as lossy (tan δ = 0.004) while
all the conductor surfaces were considered as PEC and loss free. Simulated results are
shown in Fig. 4.32. Loss less case provides an indication of the losses one should expect
when PIN diodes are used in the unit cell. These losses due to PIN diodes are less than
2 dB in magnitude. Due to losses in substrate there is an increase in the overall loss of
an active unit cell. In this case the overall loss of 4 dB was observed. Therefore, one
should expect a loss of around 2 dB due to lossy substrate at V band. In both cases the
stated loss is a reasonable value considering the RF plane integration of reconfiguration
mechanism. A similar trend in XPD was observed in both cases. Therefore, only one
the case is shown in Fig. 4.33. Due to additional test structure it was observed that
when D1X and D1Y are ON, the XPD value gets deteriorated. There is a polarization
conversion phenomenon too around 57 GHz. However, it is out of the frequency band
of interest therefore was accepted as it is.
Multiple active unit cells were fabricated and PIN diodes were mounted on them. The
whole process took much longer than expected. Mainly it was due to multiple design
changes to make it compatible with the manufacturing process. Although, the unit
cell structure was made quite simple in design, it was a three layer structure which
presented significant challenges in manufacturing. Finally, PIN diodes were mounted
and the resulting active unit cell pictures are shown in Fig. 4.34. These diodes are too
small for naked eye to see, therefore microscopic pictures are included.

0

0
−1 −2
−2 −4
Diodes OFF
D1XY ON Diodes OFF
−3 D2XY ON −6 D1XY ON
D2XY ON
−4 −8
−10
−5 56 57 58 59 60 61 62
56 57 58 59 60 61 62
Frequency (f ) GHz Frequency (f ) GHz
(a) Reflection coefficient’s magnitude: loss (b) Reflection coefficient’s magnitude: lossy
less
100 100
Diodes OFF Diodes OFF

D1XY ON D1XY ON
D2XY ON D2XY ON
0 0
−100 −100
−200 −200
−300 −300
−400 −400
56 57 58 59 60 61 62 56 57 58 59 60 61 62
Frequency (f ) GHz Frequency (f ) GHz
(c) Reflection coefficient’s phase: loss less (d) Reflection coefficient’s phase: lossy
Figure 4.32: Reflection coefficient’s magnitude and phase for a loss less and lossy full
structure of the unit cell. For lossy case, the substrate (tan δ = 0.004) is lossy while
metallic surfaces are considered as PEC.
Cross polar discrimination (XPD) (dB)
0
Diodes OFF
D1XY ON
−5 D2XY ON
−10
−15
−20
−25
−30
56 57 58 59 60 61 62
Frequency (f ) GHz
Figure 4.33: Cross polarization discrimination (XPD) for the active unit cell’s full
structure. A similar trend was observed for lossy and loss less cases.
(a) Fabricated unit cell (b) Zoomed photo (c) Zoomed photo
Figure 4.34: Fabricated unit cell photos (full structure and zoomed by microscope).
Fig. 4.35 shows the measured phase response of an active unit cell in passive state
prior to PIN diodes mounting. Its resonant frequency has shifted to 54.3 GHz and its
frequency response is narrower than the expected. Although the response was narrower
than expected, the main interest was to find if this active unit cell can produce three
phase states.
0
Reflection phase
−50
−100
−150
−200
−250
−300
−350
−400
54 54.2 54.4 54.6 54.8 55
Frequency (f ) GHz
Figure 4.35: Reflection coefficient’s phase for the active unit cell in passive state; prior
to PIN diodes’ mounting
Fig. 4.36 displays the measurements of active unit cells inside a waveguide test setup.
A unit cell is interfaced at the end of a square WR15 waveguide using screws and an
acrylic back support plate. This support plate has a hole in its center to let the DC
probe reach the unit cell’s rear side. The DC bias is provided through a DC probe. A
selectable resistor network with various resistance values in series with PIN diodes was
used to control the amount of current. An ampere meter was used to provide current
readings. DC return ground was achieved through the top metallic surface of a unit
cell. The middle ground plane of a unit cell was transferred to its top surface using a
large number of plated through hole vias around the unit cell’s periphery.
(a) Measurement setup (b) DC probe feed
Figure 4.36: Measurement setup for active unit cells. The DC probe feeds the DC
voltage to unit cell. Current limiting resistors, current meter, and diodes are all in
series connected to the DC power supply.
Finally, the measured phase and magnitude response of active unit cell’s reflection
coefficient are shown in Fig. 4.37 to 4.39. For three DC bias voltages, the unit cell can
produce three phase states. These three phase states can be clearly observed from the
magnitude response where three distinct resonant frequencies were achieved through
DC bias control. These unit cells were designed for three relative phase states (0◦ ,
90◦ , and 180◦ ), demonstration of DC bias control, and low loss. The required accuracy
for mm-waves was maximum 25 µm in simulations and the achievable accuracy by
fabrication process was 125 µm. On top of it, the via hole positioning accuracy was
very poor which was a sensitive point of the design. Via holes for diodes D1X and D1Y
are very sensitive to their position in terms of phase response where design can not
tolerate more than 25 µm inaccuracy. The goal here was to use a low cost fabrication
facility to see how close one can get to the goals even at mm-waves. This then would
act as a pilot project for further research activities in mm-waves.
Although, exact required phases in three states are not achieved however, through
this activity it is verified that there are three distinct states of the unit cell operation
controlled by a single DC bias line. The average loss in unit cells including all the losses
(due to substrate, copper metalization, PIN diodes, etc.) is around 3.2 dB. This is the
amount of gain reduction from its ideal value that one could expect at the reflectarray
level. This project ended in one time fabrication process only. Through repeated
fabrications the unit cell design can be modified to conform to the tolerances of a low
frequency facility or preferably a better fabrication facility is required for mm-wave
unit cells. This substrate thickness resulted in a severely narrow band response. It
was demonstrated in Chapter 3, that for a 10 mil thick substrate a better bandwidth
can be achieved. However, these activities did not take place in a chronological order.
Similarly, the fabrication facilities for both cases were different.
200
150 0V
5V
100 −5V
50
−50
−100
−150
−200
54 54.2 54.4 54.6 54.8 55
Frequency (f ) GHz
Figure 4.37: Reflection coefficient’s phase for the active unit cell with PIN diodes.
100

0V
X: 54.47 5V
Y: 47.09
50 −5V
X: 54.47
Y: −0.02211
0
X: 54.47
−50 Y: −32.16
−100
54.4 54.45 54.5 54.55
Frequency (f ) GHz
Figure 4.38: Reflection coefficient’s phase for the active unit cell with PIN diodes: a
zoomed view
Reflection coefficient’s Magnitude (dB)
−1
−2
0V
5V
−3 −5V
−4
−5
−6
54 54.2 54.4 54.6 54.8 55
Frequency (f ) GHz
Figure 4.39: Reflection coefficient’s magnitude for the active unit cell with PIN diodes.
4.4 Chapter Summary
This chapter presented the characterization of the selected reconfiguration technology

and the resulting active unit cell. Through a critical review of existing reconfiguration
technologies in Chapter 2 and Appendix A, PIN diodes were selected for mm-waves
operation. Two selected PIN diodes were characterized at X band (10 GHz) and V band
(60 GHz) using measurements of the insertion loss. To gain confidence in the insertion
loss measurement structure, PIN diodes were characterized at X band. Similarly, the
characterization was performed at mm-waves using a grounded co-planar waveguide
measurement structure. Good confidence was established in the measured data at
both bands. The RLC component values in PIN diodes’ ON-OFF states in a series
switch topology were extracted through these measurements. Various measurements
were performed to characterize the behavior of PIN diodes with voltage and current
variations at X band and V band with an objective of using them at low power. A
significant insight was found to economize their power consumption. It was observed
that these PIN diodes can be operated at low DC drive without significantly effecting
their insertion loss. Therefore, the smart reflectarray consisting of active unit cells
containing these PIN diodes would consume a considerably less amount of DC power.
A novel design of reconfigurable/active unit cells was conceived while considering

the implementation complexity, performance, availability and integration of elctronic
control devices (PIN diodes here). A unit cell offering three phase states, where phase
control is implemented directly at the RF plane of a reflectarray antenna, was realized.
Although it is extremely difficult to integrate multiple reconfiguration devices in a single
tiny unit cell due to the comparable sizes, this research managed to integrate four PIN
diodes in a single unit cell. The proposed reconfigurable unit cell exploited the behavior
of PIN diodes w.r.t. its biasing states. While having multiple reconfiguration devices
enable multiple phase states in a unit cell, this also give rise to two more fundamental
problems in addition to their physical integration/placement. These are the provision
of DC biasing for these devices and maintaining low cross polarization for all the phase
states.
As a part of this research an optimum solution was realized which use only one DC
bias line per unit cell to bias four PIN diodes to implement a three state phase
shifter. Further this single DC bias line was exploited to optimize the XPD of these
reconfigurable unit cells in all of its three operating phase states. This proposed
topology is capable of achieving any selected three phase states in the full range of
0◦ to 360◦ while preserving low cross polarization level of the reflected field. These
active unit cells can be single or dual polarized. The frequency of operation of both
polarized beams can be same or each polarization beam can have an independent
frequency. These reconfigurable unit cells were designed, analyzed, and were fabricated
through an external source. The performance of these unit cells was measured and
a significant insight to realize three phase states for each unit cell of a large phase
quantized reflectarray was achieved. It was observed that the low cost fabrication
facility needs characterization before it can reasonably produce mm-wave unit cells or
a better fabrication facility is required at mm-waves. It was observed to use a thicker
substrate at least 10 mil for a reasonable bandwidth response. It was shown through
simulations and measurements that these proposed unit cells are extensively less lossy
even at V band which favors their integration at mm-waves.
These reconfigurable unit cells when eventually replace the passive unit cells in a
reflectarray, the reflectarray becomes smart. One need to include the DC switching
procedure for PIN diodes in unit cells based on their location in the reflectarray to
produce the required phase shift from that location. It is discussed further in the
Chapter 5, and 6. It is worth mentioning here that the performance of a three state
phase quantized smart reflectarray based on these proposed novel unit cells would be
only 1.6 dB down in comparison to the ideal continuous phase shift smart reflectarray
(as shown in Chapter 5 and 6). The resulting smart reflectarray would be able to
steer its beam in the range of θ = ±55◦ (extendable to ±78◦ ) and in any φ direction
in both linear polarizations at the two similar or different frequencies. These unit
cells enable capability to produce dual linearly polarized reconfigurable contoured
beams. By exploiting these unit cells multiple pencil beams can be achieved. Although
the contoured and multiple beams operations are not implemented in this research,
however the proposed unit cells are fully capable to provide the functionality for their
implementation. All of these functions are highly desirable in modern satellite antennas.
Chapter 5
Reflectarray Design and Analysis
A reflectarray mainly consists of a feeding source antenna and a reflecting aperture.

Usually the feeding source antenna is a horn antenna and the reflecting aperture
consists of a large number of small antenna elements called unit cells. In a majority
of reflectarray implementations these unit cells are realized through printed microstrip
technology. The presence of a large number of unit cells in the reflectarray aperture
enables very versatile functionality by providing an excellent candidate platform to
implement phase only synthesis techniques for radiation pattern control. On one
hand it provides multiple functions, while on the other hand it complicates the design
and analysis. Reflectarrays operate on the principle of controlled reflection phase
from a large number of unit cells. Therefore, one need to be very careful in their
design to achieve a desired performance. A continuous phase control is required over
the reflectarray aperture which is impractical at mm-waves. This chapter presented
the realization of mm-waves reflectarrays, their performance estimation, degradation
in performance due to simplifications to phase quantization effects, and DC power
consumption of mm-wave smart reflectarrays.
The design and analysis of a reflectarray is presented in section 5.1 where deriving the
design specifications from system level requirements is discussed. A detailed analysis
of reflectarray geometry is presented. A program based on numerical integration to
calculate the aperture efficiency of an arbitrarily shaped reflectarray is developed.
Similarly, a parametric analysis of reflectarray efficiency is performed to study the best
144
5.1. Design and Analysis of Reflectarrays 145
combinations of reflectarray variables to provide a better efficiency value. Radiation

pattern prediction and its various constituting parameters are examined in the later part
of this section. It is not feasible to implement a continuous phase control at mm-waves.
The effects on the reflectarray performance (i.e. pointing accuracy, sidelobes, gain etc.)
when a phase discritization takes place are investigated in section 5.2. An optimum
phase quantization for mm-wave reflectarray implementation is achieved. An efficient
algorithm is devised in section 5.3 which can be used to build the simulation model of
a large reflectarray. This proved itself a very useful tool in building simulation models
for multiple iterations.
Once the order of phase quantization is selected, one need to establish which phase
states would lead to better performance? Investigation, in section 5.4, present the
analysis of a three state phase quantized reflectarray for phase states selection and
their scattering around the chosen value due to fabrication and other tolerances. The
resulting effects on performance due to phase state scattering are revealed in this
section. The DC power consumption of large reflectarrays is presented in section
5.5. Based on PIN diode power consumption versus insertion loss measurements in
section 4.2, it is established analytically that the DC power consumption of large smart
reflectarrays is not an issue. It is found that smart reflectarrays can be realized at
very low DC power levels (few watts). Similarly, a trade-off of DC power consumption
with achievable gain, polarization diversity, and number of phase discritization states
is presented.
5.1 Design and Analysis of Reflectarrays
Reflectarray design is quite similar to reflector antenna designs in many aspects. The
only difference is the reflecting surface. Similar to reflector antennas, the design is
application specific. Certain applications may pose more stringent constraints on the
design in comparisons to others. As an example, a design for its installation on a
satellite has accommodation constraints due to launch fairing. Similarly, in terrestrial
applications, aesthetics may limit the design options. Being application specific, certain
design parameters would be known as inputs and/or constraints. In the case of a
satellite, the gain is known from the EIRP specification of a satellite footprint. At the
same time, the maximum antenna diameter that can be accommodated (in case of a
non-folding design) in the launch fairing is known. Maximum allowable dimensions also
put an upper bound on the focal length. During the system engineering of a satellite,
it is reasonably known where to install the antenna e.g. on earth deck or a side wall of
the satellite. This informs the antenna designer to certain degree which configuration
(e.g. prime focus, offset or dual) of the antenna would be the best. For terrestrial
applications, maximum allowable volume due to aesthetics would provide the upper
bounds on the gain thereby limiting the performance/range of a communication link.
In this research, the design would be limited to a single reflector case only, in a prime
focus or an offset configuration. The gain of an antenna is related to its aperture area
by [35]:
4πA
G = ηa (5.1)
λ2
where G = gain, ηa = aperture efficiency, A = physical aperture area, and λ = operating
wavelength. The aperture area of a circular antenna with diameter D is Ac = π D2 /4
whereas for a rectangular aperture with length (L) and width (W ) it is given by Ar =
L × W . In an ideal case when ηa = 1, the physical aperture directly translates into
the gain. For usual reflector antennas the aperture efficiency is normally in the range
of ηa = 0.5 − 0.8. For reflectarray antennas the actual aperture efficiency, reported in
literature, is quite low mainly ηa < 0.5 [25, 191]. Once ηa is found, one can find the
reflectarray gain. The procedure to find ηa is discussed in section 5.1.2.
The following sections cover the design of reflectarrays, their efficiency prediction, and
radiation pattern analysis.
5.1.1 Reflectarray Physics
A simplified co-ordinate system used to analyze a reflectarray is displayed in Fig. 5.1

where a Y offset reflectarray is shown. Here the origin of the co-ordinate system O
is located at the reflectarray aperture center. Axes x and y are set on the aperture
plane. A feed horn with its phase center located at F illuminates the reflectarray
aperture. It is focused towards point Po on the reflectarray aperture. Its height
along z-axis is H and projection on the aperture plane is given as F 0 . In an offset

fed reflectarray, the offset angle is θo . The center of an mnth unit cell is located at
point Pmn , shown as point P in the figure. A reflectarray aperture is usually flat and
two dimensional. Therefore, points Po and Pmn would have the same z-coordinate.
However, in case of conformal reflectarrays the aperture may not be fully flat and
Z-coordinate of different locations on the aperture may differ. Similarly, for a metal
only reflectarray implementation (where phase shift is achieved by controlling the depth
of metallic ditches) the z-coordinate may vary from unit cell to unit cell. The distance
from feed to Po and Pmn is given by ro and rmn respectively. The distance between Po
and Pmn is given by smn . Radiation patterns of feed and unit cell are both modeled
as cosine q with different values of the parameter q. Radiation pattern angles for
feed and unit cell are represented by θf and θe . The reflectarray focuses its main
beam in ûo (θb , φb ) direction, whereas û is the general observation direction (θ, φ) in
spherical coordinates. Important parameters of a reflectarray are given in Table 5.1.
The formulation presented in the table accommodates single reflector configurations;
prime focus, offset and reflectarrays in which points Po and Pmn differ in z-coordinates.
x´ y ´
Z
F ߠ݂
ߠ݁
rmn Unit cell
z´
P
ro smn Y
H ߠ‫݋‬ Po
dmn Z ‫ݑ‬
ෝ
ෝ‫݋‬
‫ݑ‬
O
Y
Reflectarray
aperture X
F´
X
Figure 5.1: Co-ordinate system for reflectarray analysis.

Table 5.1: Reflectarray formulation definitions
Parameter Formulation
Center of co-ord. system O = (0, 0, 0)

(0, 0, zf ) = (0, 0, H), prime focus





Feed location ‘F’ (xf , yf , zf ) = (0, yf , zf ) = (0, −H tan θo , H), Y offset




(xf , 0, zf ) = (−H tan θo , 0, H), X offset

Feed beam point (FBP) Po = (xo , yo , zo )

Location of mnth unit cell Pmn = (xmn , ymn , zmn )
Origin to unit cell vector d~ = xmn x̂ + ymn ŷ + zmn ẑ
−−→
Position vector from feed ~ro = F P o = (xo − xf ) x̂ + (yo − yf ) ŷ + (zo − zf ) ẑ
to FBP = xo x̂ + (yo + H tan θo ) ŷ + (zo − H) ẑ, for Y offset
p
Feed to FBP distance ro = |F Po | = (xo − xf )2 + (yo − yf )2 + (zo − zf )2
Feed to FBP unit vector r̂o = ~ro /ro
−−→
Position vector from feed ~rmn = F P = (xmn − xf ) x̂ + (ymn − yf ) ŷ + (zmn − zf ) ẑ
to unit cell
p
Feed to unit cell distance rmn = |F P | = (xmn − xf )2 + (ymn − yf )2 + (zmn − zf )2
Unit vector from feed to r̂mn = ~rmn /rmn
unit cell
p
FBP to unit cell distance smn = (xmn − xo )2 + (ymn − yo )2 + (zmn − zo )2
zf − zmn
Unit cell pattern cos(θe ) =
rmn
parameter
ro2 + rmn
2 − s2
mn
Feed pattern parameter cos(θf ) = , (using law of cosines)
2ro rmn
Direction of main beam (θb , φb )
Unit vector of main beam ûo = sin(θb ) cos(φb ) x̂ + sin(θb ) sin(φb ) ŷ + cos(θb ) ẑ
direction
Direction of observation (θ , φ)
Unit vector of observation û = sin(θ) cos(φ) x̂ + sin(θ) sin(φ) ŷ + cos(θ) ẑ
direction
5.1.2 Aperture Efficiency of a Reflectarray
In reflectarrays the overall aperture efficiency is contributed by various factors including

spill over efficiency (ηs ), illumination efficiency ηi , feed mismatch loss, ohmic losses
in feed, losses in unit cells, and polarization mismatch loss. Losses due to feeding
source can be improved using a well matched feed horn with high conductivity coatings
where possible. Losses in unit cells can be minimized using a better design of unit
cells by exploiting the available degrees of freedom. The parametric studies, presented
in Chapter 3, are a useful tool to identify and address such losses. Polarizations at
both ends of a communication link should be aligned to minimize polarization losses.
However, the polarization loss in consideration here is due to depolarization of the
desired field owing to design issues. During an antenna design the cross polarization
discrimination (XPD) of an antenna should be high to reduce the amount of power
being lost in the opposite undesired polarization. In reflector antennas, there are no
unit cells therefore, unit cell losses are absent and only an ohmic loss of few tenths
of a dB is present. Similarly, due to the absence of unit cells, the depolarization
loss in reflectors is almost zero. However, in reflectarrays the unit cells contribute to
co-to-cross field coupling which causes a loss in the reflected co-polarized field. Losses
in unit cells are presumably considered as the main reason for lowering the overall
efficiency of a reflectarray. A good unit cell design is the solution to avoid these losses
which was investigated and achieved in Chapter 3 and 4 for passive and active unit
cells respectively.
Here, only the spill over efficiency ηs and illumination efficiency ηi shall be investigated
for reflectarrays. Although, other factors stated above do contribute to the overall
efficiency however, they are dealt with separately. The overall aperture efficiency is
given as:
ηa = ηs ηi ηothers (5.2)
where all the factors other than illumination and spillover are grouped together in a
single term called other efficiencies ηothers . The purpose of this analysis is to present a
picture of effects on the performance in terms of ηs and ηi for various design choices.
5.1.2.1 Feed Horn and Unit Cell Radiation Pattern Models Radiation
patterns of feed horn and unit cell are modeled as cosq . The source feed horn is
located at ‘F’ while a unit cell is located at point ‘P’. The normalized power pattern of
a feed horn in its own (source) co-ordinate system is given by:
 2 qf

 cos(θf )
 (0 ≤ θf ≤ π/2)
Uf (θ, φ) = (5.3)

0

elsewhere
Similarly, the normalized power pattern of a unit cell in its co-ordinate system is gievn
by:  2 qe

 cos(θe )
 (0 ≤ θe ≤ π/2)
Ue (θ, φ) = (5.4)

0

elsewhere
In Fig. 5.1 at point ‘P’, the angle constituted by lines F Po and F P is θf whereas, θe is
the angle between the line FP and the direction of local normal at point ‘P’ (unit cell’s
central point). For a planar aperture this local normal direction would be the same as
that of the aperture normal. Usually the source feed horn is conical or pyramidal with
a certain gain or directivity value. In reflectarray designs a value in the range of 10 to
15 dBi is very common. From power pattern, the feed directivity is calculated as:
1
Df = 2π π/2
= 2 (2 qf + 1) (5.5)
1/(4 π) Uf (θ, φ) sin θ dθ dφ
0 0
For various values of feed parameter qf , the feed horn radiation pattern and directivity
values are displayed in Fig. 5.2. For higher values of qf , the directivity increases and a
narrower radiation pattern is observed. However, the directivity curve starts to sturate
for qf values higher than 10. The feed horn size becomes unnecessarily large for higher
values of the parameter qf and is therefore, avoided.
The value of unit cell parameter qe is usually low in the range of 1 to 2. Using the
radiation pattern and directivity of a unit cell the value of qe can be found. A unit cell
can be simulated like a simple patch antenna to get this value.
0 15
qf = 1
Normalized power level (dB)

qf = 3
−1 qf = 5
Directivity (dBi)
qf = 7 10
−2 qf = 9
−3
5
−4
−5 0
−60 −40 −20 0 20 40 60 0 2 4 6 8
Angle θ (deg) Value of parameter q f
(a) Normalized radiation pattern (b) Directivity for various values of qf
Figure 5.2: Feed horn radiation pattern and directivty for various values of parameter
qf .
0.9
Z ηs Numerical
0.8 ηs Analytical
Values of spillover efficiency ηs
F 0.7
S
0.6
α
0.5
Ωs
H 0.4
0.3
0.2
O D/2
0.1
Y
0
Reflectarray D/2 0 10 20 30
aperture X Half cone angle of feed α (deg)
(a) A prime focus circular aperture reflectarray. (b) ηs computed analytically and numerically.
Figure 5.3: A prime focus circular aperture reflectarray and its spillover efficiency
calculation (F = 70 mm, D = 94.5 mm, qf = 4, qe = 1.6, α = 1 to 34 degree).
5.1.2.2 Spillover Efficiency For a reflectarray the spillover efficiency (ηs ) is defined
as the percentage of the radiated power from the feed horn which is intercepted by the
reflectarray aperture. From the illustration in Fig. 5.3, a feed horn; located at the
reflectarray focal point, radiates in a full sphere ‘S’. However, a reflectarray aperture
only intercepts a part of feed radiations through a solid angle ‘Ωs ’ of the full sphere ‘S’.
In this situation, the spillover efficiency is given as [192]:

P~ (~r) · d~s P~ (~r) · d~s
ηs =
Ωs
=
A
(5.6)
P~ (~r) · d~s P~ (~r) · d~s
S S
Here, the numerator represents the power intercepted by the reflectarray aperture while
the denominator is the total power radiated by a feed horn. P~ (~r) is the Poynting
vector associated with the feed horn. Both integrals in (5.6) are fluxes of the feed horn
Poynting vector through certain surface areas. The denominator integral can be found
numerically or analytically over the entire spherical surface around a feed horn. An
evaluation of the numerator is done only on a portion ‘Ωs ’ of the full sphere. When
the numerator is calculated through the ‘Ωs ’ portion of a sphere surrounding the feed
horn, the approach is known as the direct approach.
To find boundaries of ‘Ωs ’ in terms of feed horn’s spherical co-ordinates presents a

potential challenge when the effective aperture of a reflectarray is not circular in shape.
Such cases are very common e.g. an offset fed reflectarray, and a reflectarray with
rectangular aperture. Therefore, it is necessary to find a simpler solution for the
effective evaluation of ‘Ωs ’. Due to the fact that a reflectarray aperture and ‘Ωs ’
share the same solid angle of the sphere w.r.t. feed horn therefore, calculation of
the intercepted power can be performed on the reflectarray aperture instead of over
the solid angle ‘Ωs ’. This is known as the indirect approach as presented in the later
part of (5.6). This indirect approach is more practical and computation friendly due to
the calculation of its numerator integral in a coordinate system which accommodates
the various shapes of a reflectarray aperture boundaries. Because the numerator is
computed in the reflectarray coordinate system, it provides a flexibility to position the
feed horn arbitrarily.
From the normalized power pattern of the feed horn its associated Poynting vector is
given by:
(cos θf )2qf π
P~ (~r) = r̂ (0 ≤ θf ≤ ) (5.7)
r2 2
Therefore, in the sphere ‘S’ centered at the feed horn with coordinate variable θf and
φf one can find the denominator of (5.6) as:

2π π/2
(cos θf )2qf

P~ (~r) · d~s = r̂ · r 2
sin θ f r̂ dθ f dφ f (5.8)
0 0 r2
S
2π
= (5.9)
2 qf + 1
For a center fed circular aperture reflectarray with half cone angle α as shown in Fig.
5.3, the numerator of (5.6) can be calculated analytically as:
2π α
(cos θf )2qf

~
P (~r) · d~s = 2
r̂ · r sin θf r̂ dθf dφf (5.10)
0 0 r2
Ωs

2π 2 qf +1
= 1 − (cos α) (5.11)
2 qf + 1
Therefore, the spillover efficiency ηs for this case using the direct approach becomes
[193]:
ηs = 1 − (cos α)2 qf +1 (5.12)
To numerically compute the integrals in (5.6), one needs to express the Poynting
vector in terms of the reflectarray variables listed in Table 5.1 for an arbitrary point
Pmn (xmn , ymn , zmn ) on the reflectarray aperture as:
2qf
2 − s2
ro2 + rmn

(xmn − xf ) x̂ + (ymn − yf ) ŷ + (zmn − zf ) ẑ
P~ (~rmn ) = mn
3
2 ro rmn rmn
(5.13)
where the variables used here are shown in Fig. 5.1 and also listed in Table 5.1. As the
differential surface area would have −z surface normal direction, and a circle is traced
by the ρ2 = x2 + y 2 . Therefore, integral in numerator of (5.6) for indirect method
becomes:

P~ (~r) · d~s = P~ (~rmn ) · (−ẑ) dx dy (5.14)
A A
√
D/2
(D/2)2 −x2 2qf
ro2 + rmn
2 − s2

(zf − zmn ) mn
= 3
dx dy (5.15)
√ rmn 2 ro rmn
−D/2 − D/2)2 −x2
Therefore, the spillover efficiency ηs for a center fed circular aperture reflectarray finally
becomes:

P~ (~r) · d~s √
D/2
(D/2)2 −x2 2qf
ro2 + rmn
2 − s2

2 qf + 1 (zf − zmn )
ηs = A
= 3
mn
dx dy
2π rmn 2 ro rmn
P~ (~r) · d~s −D/2 −
√
D/2)2 −x2
S
(5.16)

= F D, θo , qf , xf , yz , zf , xo , yo , zo , zmn (5.17)
A comparison of both techniques is given in Fig. 5.3 (b) for a circular reflectarray,
where an exact match can be seen. A rectangular aperture center fed reflectarray with
dimensions Lx and Ly has ηs as:
x /2
L y /2
L 2qf
ro2 + rmn
2 − s2

2 qf + 1 (zf − zmn ) mn
ηsRect = 3
dx dy (5.18)
2π rmn 2 ro rmn
−Lx /2 −Ly /2
5.1.2.3 Illumination Efficiency From the analogy of reflector antennas the

illumination efficiency for reflectarrays can be defined as [192]:
2

I(x, y) dA
1 A
ηill =
Aaperture
2

(5.19)
I(x, y) dA

A
where ηill is illumination or taper efficiency, Aaperture is the aperture area, and I(x, y)
is the field’s amplitude distribution of a certain polarization on the aperture. The
aperture field of a reflectarray depends on the radiation patterns of its feed horn and
the unit cells. When the field from feed horn propagates towards the reflectarray
aperture containing the unit cells it has 1/rmn dependence. By making use of the
pattern parameters of feed and unit cell as defined in Table 5.1, the I(x, y) in its
normalized form becomes [194]:
(cos θf )qf
I(x, y) = (cos θe )qe (5.20)
rmn
qf
zf − zmn qe
2 2 − s2

1 ro + rmn mn
= (5.21)
rmn 2 ro rmn rmn
Therefore, for a center fed circular aperture reflectarray the illumination efficiency is
calculated by:
√

D/2
(D/2)2 −x2 qf qe 2
ro2 + rmn
2 − s2

1 mn zf − zmn
dx dy
√ rmn 2 ro rmn rmn
4 −D/2 − D/2)2 −x2
ηill = √ (5.22)
πD2
D/2
(D/2)2 −x2 qf qe 2
ro2 + rmn
2 − s2

1 mn zf − zmn
dx dy
√ rmn 2 ro rmn rmn
−D/2 − D/2)2 −x2

= F D, θo , qf , qe , xf , yz , zf , xo , yo , zo , zmn (5.23)
For a center fed rectangular aperture the limits of integral are changed accordingly as
in (5.18) and the aperture area is given by Lx × Ly .
Due to numerical integration over the reflectarray aperture, an arbitrary aperture shape
can be accommodated as long as it is expressible in terms of co-ordinate system variables
for the calculation of spillover and illumination efficiency.
5.1.2.4 Parametric Study of Reflectarray Efficiency This section investigates

the parametric study of reflectarray efficiency w.r.t. various constituting variables of
a rectangular/square reflectarray. The constituting variables of a reflectarray are its
aperture dimensions (and shape) (diameter D or length L, width W ), focal length
(F ), feed offset angle (θo ), feed pointing over the aperture (P (xo , yo , zo )), directivity
and radiation pattern of the feed horn (in terms of qf ), and finally the directivity and
radiation pattern of a unit cell (in terms of qe ). In this study when not varied; these
parameters have values of D = L = W = 94.5 mm, F = 70 mm , θo = 0◦ (center fed
case), feed pointing (0,0,0), Qf = 4, Qe = 1.6. The choice of these parameters was made
based on the available feed horn, and ease of PCB fabrication. A choice of L = 94.5
mm permits putting 6 reflectarrays on an A4 sheet. By fixing L and qf an optimum
value of F can be found. The variation of efficiency ηill × ηs is investigated for various
combinations below. Mainly, ηill × ηs is plotted instead of individual illimination and
spillover efficiencies.
1 1
Predicted values for efficiencies

0.8 ηill × ηs 0.8 ηill × ηs
ηill ηill
ηs ηs
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 1 2 3 4 0 100 200 300 400
Focal length / Diamter (F/D) Focal length F (mm)
(a) ηill × ηs versus F/D (b) ηill × ηs versus F
Figure 5.4: Illumination and spillover efficiency values versus F/D for a fixed diameter,
(D = 94.5 mm, qf = 4, qe = 1.6, focal length (F ) 10 to 400 mm , offset angle =0 (center
fed case), feed pointing (0,0,0) ).
In Fig. 5.4 the computed values of spillover, illumination efficiencies, and their product
are shown as a function of F/D ratio for a fixed D of 94.5 mm. It can be observed
that for the selected set of parameters the maximum of the product of these efficiencies
occur at an F/D value of 0.74. For a fixed value of D, at lower values of the focal
length, there is less energy being spilled over which maximizes the spillover efficiency.
However, the aperture is not illuminated efficiently i.e, unit cells located close to outer
boundaries have a significantly lower magnitude of illumination from the feed horn
(higher taper) which results in a lower illumination efficiency. In this case, as the F
increases the spillover efficiency decreases and illumination efficiency increases. The
product of spillover and illumination efficiencies attain a maximum value around F/D
= 0.74 in this case. A further increase in F do not increase the efficiency, however may
be useful for lowering the value of XPD in certain cases by sacrificing the efficiency
slightly.
Fig. 5.5 provides efficiency contours by varying the value of feed horn radiation pattern
by means of qf . The color bars in this section represent the values of efficiency. Two
cases are studied where F or D is varied and resulting values of ηs × ηill are plotted.
For a fixed D when a feed horn with a higher directivity and narrow beamwdith i.e. a
higher value of qf is used, one need to increase the focal length to attain a better value
of the efficiency. When F is fixed, one needs to use a higher value of qf for smaller
diameters for a better value of efficiency.
200 0.194
Focal length, F (mm)
140 0.235
Side length, D (mm)

0.6 0.6
0.419
120 0.5 0.5
0.444 150
100 0.4
0.4
80 100
0.652 0.3 0.3
60 0.532
0.548 0.644 0.2
0.339 0.2 50 0.306
40
0.1
2 4 6 8 10 2 4 6 8 10
qf qf
(a) ηs × ηill for F versus qf . (D = 94.5 mm) (b) ηs × ηill for D versus qf . (F = 70 mm)
Figure 5.5: Efficiency ηs × ηill as a function of array side length, focal length and qf
where color bars represent the value of efficiency.
The effects of unit cell and feed horn parameters qe and qf over efficiency are shown
in Fig. 5.6. Spillover efficiency is independent of qe while a lower value of qe results
in better illumination efficiency. The efficiency product decreases for higher values of
qe . Usually qe has a value around 1 therefore, it do not effect efficiency much. A
good choice of feed horn is an important design concern in reflectarrays. It is mainly
based on the values of F and D which result in a edge taper of around -10 dB. To
achieve a higher value of qf one needs to have a larger aperture of the feed horn which
increases the blockage in boresight reflectarrays and should be avoided in good designs.
With a selection of F = 70 mm and D = 94.5 mm it can be observed from Fig. 5.6
that an optimum value of qf is around 4 to 5. However, one needs to see what is
available commercially if one needs to avoid a custom made feed horn. In this research
a commercially available feed horn from A-Info (LB-15-10-A) was used. It has a gain
of around 12 dBi at 60 GHz therefore providing a value of qf = 4.
0.95 1

ηill × ηs
ηill 0.9
0.9 ηs
0.8
0.85
0.7
0.8
0.6
ηill × ηs
0.75 0.5 ηill
ηs
0.7 0.4
0.5 1 1.5 2 0 2 4 6 8 10
Unit cell pattern parameter (q e ) Feed horn parameter (q f )
(a) ηill × ηs versus qe (b) ηill × ηs versus qf
Figure 5.6: Illumination and spillover efficiency values versus unit cell and feed cosine
function powers (qe , qf ), (D = 94.5 mm, qf = 4, qe = 1.6, focal length (F ) 70 mm ,
offset angle =0 (center fed case), feed pointing (0,0,0) ).
0.9
0.7
0.85
0.6
0.8
0.75 0.5 ηill × ηs for θo = 0 ◦

ηill × ηs for θo = 25 ◦
0.7
ηill × ηs 0.4
ηill × ηs for θo = 40 ◦
0.65
ηill
0.6 ηs
0.3
0.55
0 10 20 30 40
2 4 6 8 10
Offset angle, θo (deg) Feed horn parameter (q f )
(a) Efficiency versus offset angle, qf = 4. (b) Efficiency for offset angles versus qf
Figure 5.7: Efficiency ηs × ηill versus feed cosine power qf and offset angles, (D = 94.5
mm, F = 70 mm, qe = 1.6 ).
Fig. 5.7 displays the effect of the feed horn’s offset angle over efficiency. With fixed
values of F , D, and qf ; the efficiency starts to degrade with an increase in the offset
angle. For a fixed offset, a higher value of qf is required to achieve the maximum
efficiency.
5 0.698 0.704 5 0.666

0.708 0.671
0.709 0.675 0.68
0.706
yo (mm)
yo (mm)
0.704 0.68 0.675
0 0.702 0 0.684
0.7 0.67
0.707 0.698 0.665
0.701 0.696
−5 −5
−5 0 5 −5 0 5
x o (mm) x o (mm)
(a) Offset angle 0◦ (b) Offset angle 20◦
0.549
yo from array center (mm)

0.75 0
2 0.551 0.558
Eficiency (ηs × ηill )

0.554 0.556 0.7 −2
0
yo (mm)
0.554
−2 0.559 0.65 −4
0.552
−4 0.557 0.55
0.6 ηill × ηs −6
−6 0.548 yo
−5 0 5 0.55 −8
0 10 20 30 40
x o (mm) Offset angle, θo (deg)
(c) Offset angle 40◦ (d) Efficiency and offset from array center
Figure 5.8: Offset angle versus efficiency predictions for various offset angles. The feed
beam point (xo , yo ) is adjusted to obtain the maximum efficiency for the configuration.
It is found that feed beam point should be offset by an amount yo from the center of
a reflectarray fed in offset configuration to achieve the maximum efficiency. (D = 94.5
mm, qf = 4, qe = 1.6, focal length (F ) 70 mm). The color bars represent the value of
efficiency.
In a center fed reflectarray, the taper angle (half cone angle) of a feed horn is equal in
either side of the aperture center for a given plane. However, in offset fed reflectarrays
this is not the case as the distance between father edge of aperture from feed horn
is more than that of the closer aperture edge. In this case to properly illuminate the
aperture, one need to point the feed horn slightly off aperture center. This case is shown
in Fig. 5.8 where the feed beam point (xo , yo , zo = 0) and offset angle is analyzed for
efficiency. For a center fed case, the feed is pointed at the aperture center to achieve
maximum efficiency as shown in Fig. 5.8 (a). With an increase in offset angle the
feed beam point is lowered as shown in Fig. 5.8 (b, c). Another factor observed is the
reduction of efficiency value with an increase in offset angle. It is to note here that the
feed blockage effect is not considered while evaluating efficiency. Fig. 5.8 (d) shows the
required shift in feed beam point relative to aperture center to achieve the indicated
value of efficiency for a given offset angle.
In Fig. 5.9 the efficiency performance of a reflectarray is analyzed for the feed location
for various values of qf parameter. It indicates the effect on efficiency when feed horn is
off the reflectarray focal point. Maximum efficiency in each case is achieved when feed
horn’s phase center is aligned with the focal point of the array. However it is not an
easy task as different frequencies have different phase center for feed horn. Therefore, a
certain value of inefficiency would remain in place. A higher value of qf requires larger
focal lengths to maintain the edge taper for a better efficiency as evident from Fig. 5.9
(b).
0.589
100 0.575
0.6020.63 0.644 0.674
0.68 120 0.7
90 0.671 0.73
0.66
zf (mm)
zf (mm)
0.699
80 0.64 100 0.65
70 0.62
80 0.711 0.6
0.685 0.6 0.692
60 0.657 0.58 0.655 0.599
0.616 0.618 0.58 0.562 0.55
50 60
−40 −20 0 −40 −20 0
yf (mm) yf (mm)
(a) ηs × ηill for qf = 4 (b) ηs × ηill for qf = 8
Figure 5.9: Efficiency ηs × ηill versus feed location for qf = 4 and 8, (D = 94.5 mm,
qe = 1.6), where color bars represent the value of efficiency.
Finally the plots of efficiency for variation in qe and qf are shown for center fed and
offset cases in Fig. 5.10. A lower value of qe results in a better efficiency however, it is
bounded by the achievable value from a unit cell.
From these figures it is evident that the reflectarray efficiency is dependent on several
factors, therefore a designer need to make best use of the available degrees of freedom
while meeting constraints to achieve a good trade-off.

0.628
2 2
0.7 0.503
0.561 0.65
1.5 0.65 1.5 0.6
0.721 0.691
0.6 0.55
1 0.615 1
qe
qe
0.668 0.5
0.55
0.5 0.508 0.5 0.441
0.566 0.45
0.5
0.4
0 0
2 4 6 8 10 2 4 6 8 10
qf qf
(a) ηs × ηill for center fed. (b) ηs × ηill for offset fed θo = 25◦
Figure 5.10: Efficiency ηs × ηill versus feed and unit cell cosine powers qf & qe . (D =
94.5 mm, F = 70 mm), where color bars represent the value of efficiency.
5.1.3 Radiation Pattern of a Reflectarray
To investigate the radiation pattern characteristics of a reflectarray there exist multiple

techniques. The most commonly used techniques are: array theory and aperture
field method. In recent years due to availability of high end computing machines 3D
EM simulations are made possible. Very recently, TICRA (Denmark) have developed
sophisticated analytical tools specially for reflectarrays [29, 30, 151, 151, 195]. Accuracy
varies with the choice of method. More accurate methods require a significant amount of
time to generate a result. Except for the array theory technique, all other methods are
computationally expensive. However, this method is limited in predicting the co-polar
field only. It can predict beam pointing direction, main lobe’s beamwidth, and the
general pattern shape. It does not provide any information about cross polar fields.
Because, feed horns used for reflectarrays, have good cross polarization characteristics,
therefore it is almost certain that one would get a good cross polarization isolation
when unit cells are well designed. To realize a very accurate analytical technique or to
compare the performance of existing techniques is not in the scope of this research.
Because for a smart reflectarray antenna to steer its radiation pattern towards the
intended direction, it need a potentially fast algorithm to get an estimate of where
the main beam would be pointed when certain phase configuration of unit cells is
implemented. It would be equally valid for tracking an intended signal by steering
the beam by phase reconfiguration. The purpose of this research is to quickly predict
the radiation pattern for a certain phase configuration of unit cells in a reflectarray to
estimate where the main beam should be pointed. Therefore, only the array theory
technique is discussed further.
Here the radiation pattern for a reflectarray configuration shown in Fig. 5.1 is calculated
based on array theory. In general a reflectarray’ has Nx × Ny unit cells arranged in a
certain pattern to direct the main beam to a desired direction. The radiation pattern
of such an array is found as [194]:
Nx XNy
X ~
E(θ, φ) = A(θ, φ) I(m, n) e j k (dmn · û) (5.24)
m=1 n=1
where d~mn is the mnth unit cell’s position vector, û is the observation direction in
reflectarray’s spherical co-ordinate system, d~mn · û = xmn sin θ cos φ + ymn sin θ sin φ +
zmn cos θ, A(θ, φ) = (cos θ)qe is the unit cell’s radiation pattern, and I(m, n) is the unit
cell’s excitation. The phase of unit cell’s excitation I(m, n) contains two parts. One
of these is the spatial delay due to Euclidian distance between the feed horn and the
mnth unit cell. This delay is represented by −k rmn , where k = 2π/λ. The second
contributor is the phase shift introduced by the mnth unit cell to direct the main beam
in a certain direction ûo . The required phase shift (∆φmn ) from each mnth unit cell in
the reflectarray configuration of Fig. 5.1 is given as [18]:
∆φmn = k (rmn − d~mn · ûo ) mod 2π (5.25)
where mod is the remainder operator. Therefore, the unit cell’s excitation function
becomes [194]:
qf
cos θf (m, n) qe
I(m, n) = cos θe (m, n) e j (−k rmn + ∆φmn ) (5.26)
rmn
qf
zf − zmn qe j (−k rmn + ∆φmn )
2 2 − s2

1 ro + rmn mn
= e (5.27)
rmn 2 ro rmn rmn
This excitation function takes into account the feed horn’s radiation pattern reaching
the unit cell’s location and receiving pattern of the unit cell located at that particular
location. It is to note for a practical case, due to tapering of the feed horn radiation
pattern, the reflectarray aperture is not illuminated uniformly which is considered in

the above function.
Finally, using these values in (5.24) results in:

qf qe
Nx X
X Ny cos θf (m, n) cos θe (m, n)
E(θ, φ) = (cos θ)qe ×
rmn
m=1 n=1
e j (−k rmn + ∆φmn ) × e j k (xmn sin θ cos φ + ymn sin θ sin φ + zmn cos θ) (5.28)
This is the reflectarray radiated field in scalar form in spherical co-ordinates (θ, φ)
when the main beam is pointed in a direction specified by (θb , φb ) as determined by
the ∆φmn phase distribution over its aperture.
In the following text, contributing factors in the reflectarray radiation pattern analysis
are discussed. For this purpose a 35 × 35 element center fed square reflectarray using
a square 2.7 mm unit cell’s lattice at 60 GHz are used. A side length (L = W = D)
= 94.5 mm, qf = 4, and qe = 1.6 are used in this example design. Pictures presented
in this section also services the purpose of a reflectarray design diagnosis. Various
calculations were performed based on the Table 5.1 and Fig. 5.1. In Fig. 5.11 the
required phase distribution over the reflectarray aperture to direct the main beam in
(θ = 0, φ = 0) direction is shown. The colors represent the required phase at the center
of a unit cell’s location, and the color bar represents the phase in degrees.
This phase is to be realized by the respective unit cell at that particular location. Here
the phase shift is shown only sub 360◦ by substracting any complete multiples of 360◦ .
The muliples of 360◦ phase shift is not practically implementable in reflectarrays at
mm-waves due to physical constraints. As can be observed a continuous phase shift is
required over the reflectarray aperture to direct the main beam in a desired direction.
Only the relative phases of unit cells are important in a reflectarray design to direct
the main beam in a certain direction. A common phase offset for all unit cells can be
adjusted by adjusting the focal length slightly. As it is a center fed configuration, the
phase of unit cells near the aperture center are zero as one should expect by taking
out any common phase offset. As the distance from the aperture center increases, the
required phase shift undergoes multiple 360◦ variations before reflectarray boundaries
are encountered. Therefore, unit cells at particular locations with the same values of
phase are replica of each other.
350
5
300
10
Unit cells in y-direction
250
15
200
20
150
25
100
30 50
35
5 10 15 20 25 30 35
Unit cells in x-direction
Figure 5.11: Continuous phases distribution over reflectarray aperture for beam
direction in (θ, φ) = (0, 0). The colors represent the required phase at the center
of a unit cell’s location, and the color bar represents the phase in degrees.
Fig. 5.12 show the radiation patterns of a unit cell and the feed horn for qe = 1.6 and
qf = 4. Higher the value of these parameters, more focused is the radiation pattern
yielding a narrower beam.
In Fig. 5.13 the distance of each unit cell from the aperture center dmn is shown.
The colors represent the distance of the center of a unit cell from the reflectarray
aperture center, and the color bar represents this distance in mm. Unit cells located
near aperture boundaries have a greater distance as compared to the unit cells located
in the vicinity of the aperture center.

0 0
−5 −5
−10 −10
−15 −15
−20 −20
−50 0 50 −50 0 50
Angle θ (deg) Angle θ (deg)
(a) Unit cell pattern qe = 1.6. (b) Feed horn pattern qf = 4
Figure 5.12: Power patterns for unit cell and feed horn, where a higher value of the
parameter q results in a narrower beamwidth.
60
5
50
10
40
15
30
20
25 20
30 10
35 0
5 10 15 20 25 30 35
Figure 5.13: Distance of each unit cell from the reflectarray center, dmn (mm), where
colors represent the distance of the center of a unit cell from the reflectarray aperture
center, and the color bar represents this distance in mm.
The distance of each unit cell from the phase center rmn is shown in Fig. 5.14. The
colors represent the distance of the center of a unit cell from the feed horn phase center
(reflectarray focal point), and the color bar represents this distance in mm. These
distances play an important role in the calculations of cos θe and the field incident
angle on each unit cell from the feed horn.
95
90
10
15 85
20
80
25
75
30
35 70
5 10 15 20 25 30 35
Figure 5.14: Distance of each unit cell from the feed horn’s phase center, rmn (mm).
The calculated values of θe are shown in Fig. 5.15. The colors represent the angle at the
unit cell center at each location. The color bar represents angle in degrees. This is the
angle in radiation pattern of the unit cell which contributes to the reception/reflection
of the incident signal. It is to note that the performance of a unit cell depends on
this angle. The resonant frequency (hence the reflected phase) slightly shifts to a lower
value due to an increase in the angle of incidence on a unit cell.
Radiated field from the feed horn in incident on a unit cell with an angle of incidence
shown in Fig. 5.16. The colors represent the angle at the unit cell center at each
location. The color bar represents angle in degrees. This is the angle in feed horn
radiation pattern which contributes to the illumination of the particular unit cell.
40
5
35
10
Unit cells in y-direction 30
15 25
20
20
15
25
10
30
5
35 0
5 10 15 20 25 30 35
Figure 5.15: Unit cells’ angle from cos θe (deg), where colors represent the angle at the
unit cell center at each location, and the color bar represents angle in degrees.
40
5
35
10
30
15 25
20
20
15
25
10
30
5
35 0
5 10 15 20 25 30 35
Figure 5.16: Incident angle for unit cells from feed horn (deg).
A feed horn due to its qf factor illuminates a reflectarray aperture with an amplitude
taper. In normalized form, the taper is 0 dB for aperture center and it reduces as the
distance from aperture center increases in a center fed design. The feed tapering over
the aperture is shown in Fig. 5.17, where an edge taper around -10 dB can be observed.
The colors represent the incident field taper at the center of a particular unit cell with
respect to the unit cell located in the reflectarray center (where the taper is 0 dB). The
color bar represents the illumination field taper in dB. Similarly the phase of incident
signal from the feed horn at reflectarray aperture is shown in Fig. 5.18. The colors
represent the angle at the unit cell center at each location. The color bar represents
angle in degrees.
−1
5
−2
10 −3
−4
15
−5
20 −6
−7
25
−8
30 −9
−10
35
5 10 15 20 25 30 35
Figure 5.17: Illumination tapering over reflectarray aperture by feed horn (dB).
150
5
100
10
50
15
0
20
−50
25
−100
30
−150
35
5 10 15 20 25 30 35
Figure 5.18: Phase of the incident field on unit cells from the feed horn(deg).
The normalized radiation pattern of the reflectarray assuming a continuous phase

from unit cells is shown in Fig. 5.19. It is observed that by setting the required
phases at unit cell locations a main beam can be synthesized in a reflectarray. It can
also be observed that a reflectarray has superb sidelobe performance as compared to
reflector antennas where the first sidelobe is usually 12 dB down the main lobe. It
is a significant advantage to comply with ITU-R radiation pattern regulations while
keeping the aperture small. The beam width of a reflectarray radiation pattern depends
on its gain as is the case in reflector antennas. Although, the array method used for
radiation pattern prediction do not provide any information of cross polarized field
levels, the XPD of a reflectarray is mainly a function of its unit cells and the feeding
horn. It is important to realize low cross polarization field levels (a high value of XPD)
in unit cells and to select a feeding horn with excellent XPD performance. When unit
cells and feed horn are chosen judiciously, one can achieve good values of XPD from a
reflectarray. Due to a large number of unit cells present in a reflectarray, it provides
an excellent candidate to implement phase only synthesis of the radiation pattern to

acquire the desired features in the resulting radiation pattern. It includes generation of
multiple beams and contoured beams. However, the purpose of this research is limited
to implement single beam reflectarrays and to prepare necessary analytical tools for
further research in this area.
0
−10
Normalized radiation pattern (dB)
−20
−30
−40
−50
−60
−80 −60 −40 −20 0 20 40 60 80
Angle θ (deg)
Figure 5.19: Normalized radiation pattern of a 35 × 35 element reflectarray.
This section presented the design of reflectarrays under the system engineering
constraints. A comprehensive discussion of reflectarays and their related analytics
was made. Various efficiency terms were discussed where a numerical procedure to
calculate illumination and spillover efficiencies of an arbitrary reflectarray aperture
was investigated. This section also presented the parametric analysis of these efficiency
terms w.r.t. reflectarray parameters. Finally, the radiation pattern of reflectarrays was
examined and various contributing factors were explained. All the presented details
in this section belong to continuous phase reflectarrays. It is not practical to achieve
continuous phase variations at mm-waves. What happens when a continuous phase
variation from unit cells of a reflectarray is not achievable? The next section presents
an analysis of discretized phase reflectarray’s performance, to answer this question.
5.2. Phase Discretization in Reflectarrays 171
5.2 Phase Discretization in Reflectarrays
Reflectarray implementations using a continuous phase control or a higher number of

phase quantization states are not practically feasible [146] at mm-waves. Due to a
large number of elements in a reflectarray, the reconfiguration mechanism in each unit
cell needs drastic simplifications [196]. This section presents the effects on antenna
performance resulting from such simplifications. The performance of various number
of phase states per unit cell is compared with the ideal continuous phase case for a
center fed 20 λ square reflectarray with 15 λ focal length at 60 GHz. A square unit cell
lattice with 0.54 λ inter unit cell spacing was used. Effects of the phase quantization
on radiation pattern, directivity loss, scan range, and bandwidth are presented here
through Matlab
R
simulations and 3D EM simulations in CST Microwave Studio.
Table 5.2 lists the loss in directivity for various phase states in comparison to a
continuous phase shift [197]. An eight state phase quantization practically results
in negligible directivity loss. Therefore, it is taken as a reference to normalize the
directivity values for CST simulations. Here it can be seen that directivity loss increases
when implementing coarse phase quantization; however doing so simplifies the DC
biasing and real estate issues as it requires less number of discrete electronic elements.
The difference in Matlab
R
calculated and CST simulated values is due to the fact
that reflection phase was calculated using (1); however, CST simulations consider
a more practical implementation of small scattering of reflection phase around the
exact desired value due to array behavior and incident angle dependence. Moreover,
judiciously selected reflecting patch dimensions have shown a decrease in loss. A two
state implementation would result in an excessive loss in directivity which may not
be practically feasible for certain situations. However, by just increasing the phase
states to three considerably reduces the loss in directivity to a practically acceptable
solution. It is possible to achieve three phase states while still maintaining the DC
biasing complexity of a two state phase shifter at antenna RF plane as was achieved
in Chapter 4, where a three state phase shifter was implemented for phase qunatized
reflectarray unit cells.
Table 5.2: Performance of phase quantized reflectarrays
Phase Directivity Directivity Bandwidth Max. scan

Quantization loss Matlab loss CST (3 dB) range
Continuous phase 0 dB — — ± 76.8◦
Eight phase states 0.22 dB — 8.8 GHz ± 76.8◦
Four phase states 0.99 dB 0.7 - 0.22 = 9.0 GHz ± 76.8◦
0.92 dB
Three phase states 1.66 dB 0.9 - 0.22 = 10 GHz ± 76.5◦
1.12 dB
Two phase states 3.92 dB 2.7 - 0.22 = 10.2 GHz ± 76.7◦
2.92 dB
Table 5.2 also lists the bandwidth and maximum scan range as simulated in CST and
Matlab
R
respectively. The grating free scan range was 55◦ due to inter-element spacing,
however the beam can be scanned to a maximum of approximately 77◦ at the cost of
reduced gain due to the generation of grating lobes. A similar maximum scan range
was observed through CST full wave simulations of this study. Through another study
using 0.5 λ spaced unit cells in reflectarrays simulated in CST a similar maximum
scan range was found. Reflectarrays suffer scan blindness above this range [196] and
no further improvement was observed. To find a solution to scan blindness above this
angle is not in the scope of this research.
The bandwidth improves as the phase quantization become coarse due to the fact that it
is more tolerant to phase changes because of larger phase steps. Table 5.2 provides the 3
dB bandwidth from CST simulated results of this study. A 16% percentage bandwidth
was observed using 10 mil thick RO5880 substarte at 60 GHz. The bandwidth response,
generated through CST simulations, is shown in Fig. 5.20 where all the arrays were
primarily designed at 60 GHz. Due to the selection of unit cell patches to produce the
desired reflection phases, a shift in the reflectarray frequency bandwidth was observed;
the coarser the quantization, higher is the shift in the operating band. However, this is
not a drawback; the reflectarray can be made to operate at the desired band by slightly
changing the patch dimensions. These curves are normalized using the directivity
value of the 8 states phase quantized reflectarray. It is worth reminding the reader that
directivity of a continuous phase reflectarray differs only by 0.22 dB than that of an 8
states phase quantized reflectarray.
0
−1
−2
Loss in directivity (dB)
−3
−4
−5 2 states
3 states
−6 4 states
8 states
−7
−8
−9
−10
52 54 56 58 60 62 64 66 68 70
Frequency (GHz)
Figure 5.20: Directivity loss in quantized reflectarrays with various number of phase
states per unit cell simulated in CST Microwave Studio.
When the radiated beam is scanned away from the boresight, it results in a directivity
loss. Matlab
R
code was used to generate the scan loss curves as shown in Fig. 5.21 for
various beam pointing angles. A consistent degradation of directivity can be observed
for each curve shown including continuous and phase quantized reflectarray. In Fig.
5.22, the directivity loss for quantized phase reflectarrays relative to the continuous
phase reflectarray at various beam pointing angles; known as differential scan loss, is
shown. A fairly constant differential scan loss in each curve indicates the beam scanning
loss is nearly independent of the scan angle for each phase quantization case. Therefore,
at any scan angle the loss in directivity for any phase quantized case in comparison to
the continuous phase case is constant. In other words the roll off in directivity is the
same for all cases w.r.t. the scan angle. For instance, the loss is directivity due to two
states phase quantized array is around 4 dB at boresight w.r.t. continuous phase case.
This loss has the same 4 dB value at any scan angle in comparison to the continuous
phase case.
As a reflectarray contains a significantly high number of reflecting unit cells, therefore,

the beam pointing error; when phase quantization takes place, is not significantly high
over the entire scan range in comparison to a continuous phase reflectarray. It is
shown in Fig. 5.23; where the achieved pointing angles are plotted with respect to
the desired pointing angles. The achievable beam pointing, even for the worst case
of phase quantization (two states) is within 0.1◦ of the value that a continuous phase
reflectarray can achieve. Therefore, the beam pointing angle is independent of the
number of phase states in large arrays. The achieved pointing angle curves saturate
near 77◦ in all the cases including continuous phase reflectarray and the beam pointing
error increases without any significant further improvement in achieved pointing angle.
This is due to scan blindness in reflectarrays.
−2
−4
Loss in directivity (dB)
−6
−8
Continuous phase
−10
8 states
4 states
−12 3 states
2 states
−14
−16
0 10 20 30 40 50 60 70 80
Scan angle, θ (deg)
Figure 5.21: Directivity loss in reflectarrays with scan angle simulated in Matlab.
Directivity for each case is normalized with the directivity of continuous phase case.
−0.5
Normalized loss in directivity (dB) −1
−1.5
−2
Continuous phase
−2.5 8 states
4 states
−3 3 states
2 states
−3.5
−4
−4.5
0 10 20 30 40 50 60 70 80
Scan angle, θ (deg)
Figure 5.22: Normalized directivity loss in reflectarrays with scan angle simulated in
Matlab. Directivity for each case is normalized with the directivity of continuous phase
case at each scan angle.
80
Achieved beam pointing angle, θa (deg)
70 54.8
60
54.6
50
54.4
40 54.9 55 55.1
Continuous phase
30 8 states
4 states
20 3 states
2 states
10
0
0 10 20 30 40 50 60 70 80 90
Desired beam pointing angle, θd (deg)
Figure 5.23: Desired beam pointing angle versus achieved beam pointing angle in
reflecarrays, simulated in Matlab.
0
0 −15
−5
Normalized radiation pattern (dB) −2 −20
−10
−4 −25
−15
−6 −30
−20 −2 0 2 5 10 15 20
−25 2 states
3 states
−30
4 states
−35 8 states
−40
−45
−50
−150 −100 −50 0 50 100 150
Angle, θ (deg)
Figure 5.24: Normalized radiation pattern of phase quantized reflectarrays, simulated

in CST Microwave Studio.
Fig. 5.24 displays co-polar radiation patterns of phase quantized reflectarrays generated
through CST simulations. Eight states phase quantized reflectarray directivity was used
to normalize the directivity of other phase quantized reflectarrays. There is a loss in
directivity at boresight due to phase quantization. The first sidelobe level in all the
cases is around 20 dB below the boresight peak value. The far out sidelobes are well
below the boresight directivity for most of the practical purposes.
The cross-polar radiation response is not significantly affected due to phase quantization
as was observed in CST simulations. In all cases with a feed horn having around 40 dB
cross polarization discrimination, better than 35 dB cross polarization discrimination
at reflectarray level was achieved.
This section presented a comparison of phase quantized reflectarrays w.r.t. continuous

phase reflectarray at 60 GHz. With phase quantization, a loss in boresight directivity
is observed which increases as the phase quantization becomes more coarse. A severely
coarse phase quantization of two phase states produces the highest calculated loss of
3.9 dB. However, with a suitable choice of reflecting patches this can be reduced to less
5.3. Reflectarray EM Simulation Model Building 177
than 3 dB, as shown in CST simulations. A three states phase quantized reflectarray
exhibits the most desirable performance tradeoff at mm-wave bands due to significantly
less complex DC biasing network required for its implementation. The higher phase
quantization states result in a better directivity; however may not be optimum when
implementation complexity is considered at mm-wave bands where real estate is a
potential issue. A smooth and consistent directivity degradation due to beam scanning
was observed for all the cases of phase quantization. The beam pointing error is minimal
and a reliable beam scanning can be achieved in all the cases of phase quantization.
In radiation pattern the first sidelobe level is around 20 dB down in all the cases. The
cross polarization discrimination was observed not being a strong function of the phase
quantization.
5.3 Reflectarray EM Simulation Model Building
The availability of high end computing resources and EM simulation packages has
enabled the simulation of large structures. To accurately predict the performance
of a reflectarray it is required to simulate it in a suitable electromagnetic (EM)
simulation software e.g. CST Microwave Studio, HFSS, or FEKO to name a few.
Such simulations require building a reflectarray structure in the chosen simulation
package. An electrically large reflectarray consists of thousands of unit cells. The design
parameters of a unit cell are selected based on reflection phase curves e.g. length and
width of patch in a variable length patch approach. To prepare a simulation model for
a large reflectarray is very cumbersome as it takes exhaustive efforts, keen attention
to details and significant amount of time to build such a simulation model. Moreover
it can be prone to human errors in forming the geometry. When several iterations of
modeling are required it worsens the issue even further. There is a need to automate
this activity to reduce the laborious effort for antenna designers. To date there is no
algorithm available at least in public domain to address this particular issue.
This section presents a time efficient automated modeling of reflectarray structure

by interfacing Matlab
R
and CST Microwave Studio. The presented algorithm is
very generic, time efficient and makes the modeling easy with least intervention of
the designer [198]. This generic algorithm can also be adopted to build complex
shapes other than reflectarrays. Depending on the users choice the algorithm can
be implemented in other programming softwares and simulation packages. Here the
required reflected phase angle calculations were performed in Matlab
R
to direct the
radiated beam in a certain direction. CST was used to simulate the phase curves of
unit cells with a variable length approach. Once the required set of dimensions is
decided, they are programmed in Matlab
R
to generate the structure of a reflectarray.
Several passive arrays were designed using this modeling technique and was found to
be approximately 35 times more efficient than manual modeling [198]. This technique
is particularly useful where several design iterations are required.
Using (5.25) one can find the continuous phases required from unit cells in a reflectarray.
Once the continuous phases for unit cell central locations in the reflectarray are
calculated, it is logical to discretize them into a set of phases that can be implemented
in a reflectarray design. The reflectarrays can benefit from implementing a discrete
set of reflection phases and normally an eight states phase quantization results in a
practically negligible loss in gain. When the set of discrete phases is selected, the
required reflecting patch parameters/dimensions in a unit cell to produce it are found.
For this purpose a unit cell is simulated in an EM simulation software to generate a set
of phase curves by parametrizing the variables of unit cell e.g. patch length in a printed
microstrip variable length based unit cell. From these simulation curves generated by
EM software or by some other means a set of dimensions is selected corresponding
to each phase in the set of required discretized phases. When the set of dimensions
is known, a reflectarray can be constructed in EM simulation software for analyzing
its RF performance. The construction of reflectarray from the known set of unit cell
dimensions is a laborious process. An automation procedure for the construction of
a reflectarray from the known set of unit cell dimensions using Matlab
R
and CST is
given below.
Matlab
R
is very common among engineers and similarly it is true for CST Microwave
Studio for RF and Microwave designers. Both provide an interface for visual basic (VB)
scripts [2, 3]. Their interfacing is the solution to design automation. Fig. 5.25 displays
the basic steps involved in the generation of a reflectarray structure. A Matlab
R
program calculates the required phases and the user selects the range of phase values
which should correspond to a certain discrete phase. The unit cell parameters required
to generate that particular discrete phase are fed to an EM simulator instance (CST
project) via VB script in Matlab
R
for building a unit cell. This procedure is repeated
for each location of the unit cells in a reflectarray.
Reflectarray and unit cell Unit cell reflection phase

dimensions curves
Calculate discrete reflection

phase from a unit cell location
in the reflectarray
Choose the dimensions Set of chosen dimensions for

corresponding to reflection discrete reflection phases for
phase from the discrete set implementation
VB script to build the model

Repeat for each location of
for unit cell in reflectarray EM
unit cell in the reflectarray
simulation software
Complete the remaining tasks

in EM software and simulate
Figure 5.25: Reflectarray construction algorithm.
This algorithm was implemented for various phase quantization states using variable
size square patches realized through the generation of bricks (a 3D shape) in CST.
The interface is syntax sensitive therefore needs a careful implementation. A designer
can add as many phase discretization steps as deemed necessary without changing the
construction time significantly. This is due to the fact that simple Matlab
R
commands
run much faster as compared to its interface with CST. The implemented script stores
editable history of objects which is a very desirable feature. Similarly, it arranges the
unit cells in rows and columns as in visible geometry for easy access and modification
at a later stage in a large set of unit cells. The implementation can be easily applied
to circular or other shaped reflectarrays by not calling the interface procedure for unit
cells that are out of the diameter range. For other shapes user can use “History List”
in CST for a particular object to understand the script required to build shapes of
interest.
A sample output from the script is shown in Fig. 5.26 for the structure of a four states
phase quantized reflectarray. The grey patches are the unit cell patches to produce the
required reflection phases from the particular locations. Once a geometry in CST has
been generated using script, the designer needs only to select other simple parameters
such as frequency, units and material specifications while also the feed horn model
needs to be formed too. These steps can also be automated. It takes about a minute to
manually build and set the required parameter values for each unit cell. For comparison
our model took around 40 minutes on a normal desktop computer to build a reflectarray
containing 37 × 37 unit cells which is drastically less as compared to the time taken if
done manually.
Figure 5.26: A sample of a four states phase quantized reflectarray constructed through
algorithm.
This section presented an efficient automated algorithm for building a software EM

5.4. Unit Cells’ Phase State Selection in Phase Discretized Reflectarrays 181
simulation model of electrically large reflectarrays. For elaboration CST and Matlab
R
were interfaced using visual basic scripts. The algorithm can be adapted to other EM
simulators which support such scripting interfaces e.g. HFSS. By having such a small
piece of script enables building very large structures automatically which would have
taken days to accomplish otherwise.
5.4 Unit Cells’ Phase State Selection in Phase Discretized

Reflectarrays
In a large mm-wave reflectarray the accommodation of multiple control devices

and their DC biasing complicates the situation and results in a performance
degradation. An implementation of a three states reflection phase quantization offers
a potential solution to realize practically acceptable optimum performance. Such an
implementation greatly simplifies the issues arising from DC biasing of multiple control
elements in a large array. When phase states are chosen judiciously it results in only
1.6 dB reduction in directivity when compared with the performance of continuous
reflection phase reflectarrays. Although, this reduction may appear to be a lot however,
it become acceptable when one considers the performance degradation due to the
accommodation and DC biasing of multiple control elements to implement a multi-state
reflection phase control. In general, implementing a coarse phase quantization results
in wider bandwidths because the selected phase states remain well maintained for a
greater extent of the required reflection phase as compared to a fine phase quantization
implementation.
This section presents two aspects of the phase selection in a three states phase quantized
reflectarray. These phase states can be any three phase states in the range of 0◦ to 360◦ .
First question is which choice of these phase states is optimum? Second question is
what happens when the chosen phase states experience a scattering around the chosen
ideal state due to manufacturing tolerances or a change in the angle of incidence on
a unit cell? This section provides answers to these two questions by presenting the
simulated and measured data.
Table 5.3: Effect of phase state selection in a three states phase quantized reflectarray.
Unit cells’ phase states Directivity loss w.r.t. Directivity loss w.r.t.
in the reflectarray continuous phases: continuous phases:
37 × 37 elements on a 35 × 35 elements on a
2.7 mm lattice, F = 75 2.7 mm lattice, F = 70
mm mm
Continuous 0 dB 0 dB
0◦ , 120◦ , 240◦ 1.66 dB 1.70 dB
0◦ , 90◦ , 225◦ 1.99 dB 2.05 dB
0◦ , 90◦ , 180◦ 2.54 dB 2.62 dB
−1
−2
−3
−4
−5
Coninuous phase
−6
Discrete phase (0 ◦ , 120 ◦ , 240 ◦ )
Discrete phase (0 ◦ , 90 ◦ , 180 ◦ )
−7 Discrete phase (0 ◦ , 90 ◦ , 225 ◦ )
−8
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2
Angle θ (deg)
Figure 5.27: Normalized radiation pattern of a three states phase quantized reflectarray
with various phase state selections.
In Table 5.3 Matlab

R
simulated data is presented for two, three states phase quantized,
reflectarrays having slightly different sizes and focal lengths with a fixed unit cell
lattice dimensions at 60 GHz. Three different phase state selections for each case
were investigated. For each reflectarray the difference in directivity loss (columns 2
and 3) for each corresponding case of the phase state selection (each row) is not huge.
On the other hand, each phase state selection case (each row) for three states phase
quantization experience a noticeable difference in the directivity loss. The minimum
loss for both arrays was observed when phase states are equally distributed in the range
of 0◦ to 360◦ . Therefore, a uniformly distributed phase states produce a minimum loss
and are an optimum choice for the phase quantized reflectarray. Fig. 5.27 presents the
data for 37 × 37 elements reflectarray discussed in Table 5.3 using Matlab
R
analysis.
The scattering of phase states around the chosen states can be understood from Fig.
5.28 where three phase states (0◦ , 120◦ and 240◦ ) are chosen states to implement a three
states phase shifter. In Fig. 5.28 shaded areas around each chosen phase state represent
scattering of the phase states. The actual phase state can lie anywhere in the shaded
area around its chosen phase state. These phase states are achieved through control of
various parameters of unit cells in a reflectarray. Each unit cell has its own scattering
of phase based on the fabrication process and its location in the reflectarray aperture.
A scattering of phase states’ can happen in passive as well as active implementations
of reflectarrays. In passive implementations there are mainly two factors causing this
situation which are fabrication process and the angle of incidence. Similarly, in active
unit cells, an additional disturbance is caused by a change of control device parameters
over a frequency range. In the following text fabrication process related issues are
discussed in a little more detail.
90°
120°
180° 0°
240°
270°
Figure 5.28: Phase state scattering in phase discretized reflectarrays.

Here, a variable size patch technique is explored, however scattering behavior in other
techniques would be present too. It is the frequency of operation which decides
acceptable fabrication tolerances. One can not ignore those tolerances at 60 GHz
which can be easily ignored at X band. The fabrication tolerances are dependent on
the chosen fabrication process. For an instance in a liquid acid based etching process
the screens used for photolithogrphic process are subject to the printer accuracy. It
was observed during this research that the printing accuracy in horizontal and vertical
axis was slightly different. Similarly, the unit cells in the middle portion were more
accurate during this printing process. During PCB etching process on unit cells (and
reflectarrays) the etching accuracy is highly dependent on the acid concentration and
the time a sample remains dipped in it. It was also observed that the relative location of
a unit cell in the reflectarray aperture also plays a role during etching. Unit cells located
near boundaries were etched more as compared to the unit cells near the aperture center.
It took several iteration to calibrate the PCB fabrication facility at 60 GHz to get the
samples right. It is quite likely that unit cells of the same reflectarray suffer different
phase scattering due to fabrication process. Similarly, the unit cells’ reflection phase is
dependent on the incidence angle of the field which consequently depends on where a
unit cell is located in the reflectarray aperture.
There are high chances of observing scattering of the phase states in a reflectarray
design. A scattering of reflection phases around the chosen original phase states can
cause performance degradation. Therefore, even for a coarsely quantized reflection
phase, maintaining the chosen phase states is very important in a good reflectarray
design. This can be achieved through considering the effect of incident angle and
selecting a low tolerance fabrication facility. However, cost would be an impeding
factor. Here, the effects of such scattering of phase states were investigated through
simulations, fabrication, and measurements of reflectarrays as discussed in the following
paragraphs.
The effect of such a scattering of phase has been emulated here by implementing two,
three states, phase quantized passive reflectarrays. Both of these reflectarrays were 37×
37 elements on a 2.7 mm lattice, with a focal length of 75 mm at 60 GHz [199]. A large
number of unit cells were fabricated using a low cost low frequency fabrication facility.
The reflecting patches in each of these unit cells were having different dimensions by
introducing an incremental step. Some dimensions were repeated to see the fabrication
facility repeatability. The reflection phase response of all these unit cells was measured
through the waveguide test setup. After analyzing the reflection phase curves for unit
cells two different sets of unit cells were selected.
Table 5.4: Effect of phase state scattering around the chosen phase states in a three
states phase quantized reflectarray.
Three states phase Phase state 1: Phase state 2: Phase state 3:

quantized reflectarray Unit Cell 1 Unit Cell 2 Unit Cell 3
Reflectarray 1 (set 1) L1 L2 L3
Reflectarray 2 (set 2) L1 + 40 µm L2 - 5 µm L3 - 40 µm
Each set consists of three unit cells to produce three required reflection phases in a
three states phase quantized reflectarray. The dimensions of unit cells in both sets
corresponding to a particular phase state were chosen slightly different to see the
effect of phase scattering on the resultant reflectarray performance. Table 5.4 lists
parameters of unit cells in each set. L2 represents the required patch length (width also
of square patch) for the phase state near resonance, while L1 and L3 represent lengths
corresponding to chosen phase states above and below the resonance respectively. For
set 2 the differential change in L2 is far less than that for L1 and L3 due to the high
sensitivity of reflection phase near the resonance. A larger change in L2 will cause a
shift in the operating frequency. Therefore, it is slightly changed in this exercise. Fig.
5.29 displays the resulting simulated phases for each case. Here, floquet ports normal
incidence was considered for unit cell analysis. It was observed through simulations
that a 5 µm change in resonant length does no shift the resonant reflection phase too
much as is evident for L2 case.
Reflection phase (deg) L2

L1 - 40 µm
L1
L1 + 40 µm
L2 - 5 µm
L2 + 5 µm
L3 - 40 µm
L3
L3 + 40 µm
Frequency (GHz)
Figure 5.29: Phase state scattering in phase discretized reflectarrays simulated in CST.
0
0 −17.5
−5
−0.2
– COMPUTER SIMULATION TECHNOLOGY | www.cst.com −18

CST−10
−0.4
−15 −0.6 −18.5
−0.5 0 0.5
−20
Reflectarray 1 3 4 5 6
−25 Reflectarray 2
−30
−35
−40
−45
−50
−150 −100 −50 0 50 100 150
Angle, θ (deg)
Figure 5.30: Normalized radiation pattern of two, three states, phase quantized
reflectarrays with phase scattering simulated in CST.
Based on the selected set of unit cells in each case two reflectarrays were constructed in
CST Microwave Studio using the custom build codes, discussed in Section 5.3. Fig. 5.30
displays the simulated radiation pattern of both arrays in CST Microwave Studio at 60
GHz. It is observed that both reflectarrays can maintain low sidelobes as indicated by
an inset. The directivity of reflectarray made of unit cells of set 2 (called reflectarray
2) was 0.14 dB less than that of the reflectarray 1. Therefore, it can be concluded that
at a fixed frequency both reflectarrays can produce a good radiation pattern with low
sidelobes and a small difference in directivity.
−1
−2
Normalized directivity (dBi)
−3
−4
−5
Reflectarray 1 simulated @ FL1
−6 Reflectarray 2 simulated @ FL1
Reflectarray 1 measured @ FL2
−7
Reflectarray 2 measured @ FL2
−8
−9
−10
55 56 57 58 59 60 61 62 63 64 65
Frequency (GHz)
Figure 5.31: Normalized directivities of two, three states, phase quantized reflectarrays
with phase scattering. @ FLx = at focal length x.
The simulated directivity versus frequency (bandwidth) response at a fixed focal length
(FL1) for both reflectarrays is shown in Fig. 5.31, where it can be observed that
reflectarray 1 has more than 2 GHz wider bandwidth as compared to reflectarray 2. This
is due to the fact that phase response of chosen unit cells in set 2 could not maintain the
required phase states over a wider frequency range. The measured bandwidth response
is also shown in this figure. To obtain maximum gain and to account for the unknown
phase center of the feed horn, the focal length was optimized during measurements. The
optimum focal length (FL2) used in measurements was slightly different than FL1 used
in simulations. Due to a change in the focal length, upper and lower side responses get
flipped around the center frequency. However, a clear trend of bandwidth reduction
is visible. Therefore, it can be concluded that the scattering of phase around the
5.5. DC Power Consumption in Large Reflectarrays 188
selected nominal value reduced the operating bandwidth in this case, which highlights
the importance of maintaining the phase states near their nominal values. It also
provides an insight into the performance when one uses low cost fabrication facility at
mm-waves frequencies where phase state scattering happens. It can be concluded that
when a very wide bandwidth is not required phase scattering can be accommodated
and reflectarrays can be produced using a low cost fabrication facility.
5.5 DC Power Consumption in Large Reflectarrays
The complexity and power consumption of smart antennas impeded their widespread
application until recently. This section reveals the power requirements of phase
quantized reflectarray smart antennas. Data presented in this section is based on
the PIN diode DC power consumption measurements in Chapter 4. At two frequency
bands the power consumption results are presented. At X band the forward voltage
required to produce a 5 mA forward current is taken as 1.41 V while for V band it is 1.5
V. The values of forward resistance at X and V bands were 3 Ω and 8 Ω respectively. A
PIN diode operating at X band would consume 7.05 mW DC power. While a V band
PIN diode would have 7.5 mW of DC power consumption. The power consumption
calculations are performed under the assumption that in a phase discretized reflectarray
all the phase states are equally likely. This in fact is the case in a real reflectarray. The
occurrences of each phase state were counted in a three states reflectarray design and it
was found that the resulting three sums were separated by a close margin. Therefore,
one can use this assumption with a fair degree of confidence. With this confidence, in a
two states phase discretized reflectarray the number of unit cells configured to produce
phase state 1 would be equal to the number of unit cells to produce the phase state 2.
Data presented here assumes a good unit cell design where for one polarization; two
states phase quantization is implemented by integrating only one PIN diode and three
states phase quantization is implemented using only two PIN diodes. The later case
was achieved in this research for dual polarization where each polarization is associated
with two PIN diodes.
Table 5.5: DC power consumption of reflectarrays at X band.
Reflectarray 2 states SP 3 states SP 2 states DP 3 states DP

size (elements) (W) (W) (W) (W)
30 × 30 3.17 8.46 6.34 16.92
35 × 35 4.32 11.51 8.63 23.03
37 × 37 4.82 12.86 9.65 25.73
40 × 40 5.64 15.04 11.28 30.08
50 × 50 8.8 23.5 17.625 47
60 × 60 12.69 33.84 25.38 67.68
Table 5.6: DC power consumption of reflectarrays at V band.
Reflectarray 2 states SP 3 states SP 2 states DP 3 states DP

size (elements) (W) (W) (W) (W)
30 × 30 3.37 9 6.75 18
35 × 35 4.59 12.25 9.18 24.5
37 × 37 5.13 13.69 10.26 27.38
40 × 40 6 16 12 32
50 × 50 9.375 25 18.75 50
60 × 60 13.5 36 27 72
In a good unit cell design for a smart reflectarray as was the case in this research one
phase state can be configured at no DC power consumption. Therefore, in a two states
phase discretized reflectarray, only one half of the total unit cells will be consuming
DC power. Similarly, in a three states phase quantization only two third of the unit
cells would be consuming DC power. The discussion is capped to the implementation
of two and three states phase quantization cases as higher order quantizations are not
feasible for mm-waves. The calculations are presented for single polarization (SP) as
well as dual polarization (DP) implementations of unit cells in a smart reflectarray.
Tables 5.5 and 5.6 present the data for various reflectarray sizes. The reflectarray size
is considered as the number of unit cells in both directions of the aperture. The power
consumption of V band is slightly higher due to a higher diode forward voltage. Fig.
5.32 presents the data for electrically very large smart reflectarrays where the number
of unit cells can be as large as 40,000. In Table 5.7 the gain values (with continuous
phase) considering various efficiencies are presented to provide an idea of the array size
and performance. Here a 2.7 mm square lattice is used to calculate the reflectarray
aperture size. To get gain values for two and three states phase quantized reflectarrays
one need to subtract the gain loss due to phase qunatization from the presented gain
values in this table.
Table 5.7: Expected gain of reflectarrays at V band.
Reflectarray Size (mm) Gain (dBi) Gain (dBi) Gain (dBi)

size (elements) η = 0.4 η = 0.5 η = 0.6
30 × 30 81 31.20 32.17 32.96
35 × 35 94.5 32.54 33.51 34.30
37 × 37 99.9 33.02 33.99 34.78
40 × 40 108 33.70 34.67 35.46
50 × 50 135 35.64 36.60 37.40
60 × 60 162 37.22 38.19 38.98
100 × 100 270 41.66 42.63 43.42
150 × 150 405 45.18 46.15 46.94
200 × 200 540 47.68 48.65 49.44
A key message from this data is: when the required gain is less than 35 dBi, the power
consumption is not a huge problem which is an excellent result. When the required gain
increases further one need to re-consider the single or dual polarization and two & three
states of phase quantization. Dual polarized unit cells integrate double the number of
PIN diodes therefore, double is the power consumption as compared to a single polarized
unit cell. Shifting to a single polarized case for higher gains directly reduces the amount
of DC power consumption to one half in implementations discussed here which is very
noticeable for three states implementation of a 40,000 elements reflectarray.
800
30 2 states, single pol
700
2 states, dual pol
DC power consumption (W)
20 3 states, sinlge pol
3 states, dual pol
600
10
500
0
0 0.5 1 1.5 2
400
300
200
100
0
0 5 10 15 20 25 30 35 40
Number of unit cells in the reflectarray ×1000
(a) DC power consumption of X band reflectarrays.
800
40 2 states, single pol
700 2 states, dual pol
30
3 states, sinlge pol
DC power consumption (W)
600 20 3 states, dual pol
10
500
0
0 0.5 1 1.5 2
400
300
200
100
0
0 5 10 15 20 25 30 35 40
Number of unit cells in the reflectarray ×1000
(b) DC power consumption of V band reflectarrays.
Figure 5.32: DC power consumption of large phase quantized reflectarrays at X and V

bands using PIN diode test data. Results for two and three states phase quantization
for single and dual polarized unit cells are shown.
Shifting to a single polarized case of two states phase quantization in such an

implementation would reduce the gain approximately by 2.26 dB whereas the DC power
consumption is lowered by 650 W which is a huge reduction. However, when gain is of
prime importance one would need extra unit cells to compensate this gain reduction of
2.26 dB which will result in an increase in power consumption proportionally. In this
case to compensate the gain loss one would need approximately 27,600 more unit cells
which consume an additional 103.5 W power. Therefore, the total power consumption
of such a huge two states single polarized reflectarray of 67,600 elements would be 253.5
W. Using the above discussion one can trade off the functionality (single/dual polarized)
with the power consumption against an implementation of the phase quantization.
This section presented the DC power consumption of large reflectarray smart antennas.
It was established based on measurements of the control devices that one can achieve a
low power consuming high gain smart antennas even at mm-waves. Achievable gains,
number of phase quantization states, polarizations can be optimized and traded-off
against the DC power consumption to realize a suitable mm-wave smart reflectarray
solution for a given application.
5.6 Chapter Summary
This chapter presented design of reflectarrays with a thorough analysis of the spillover
and illumination efficiencies using an analytical integration technique. Such an
analytical technique perform integration over the reflectarray aperture instead of a
partial sphere around the feed horn. Therefore, it is possible to predict the efficiency
values for any shape of the reflectarray aperture. A parametric study of illumination
and taper efficiencies was presented against various design variables of a reflectarray.
The radiation pattern of reflectarrays was calculated based on array theory to enable
fast prediction of main beam and its direction which is important for smart antennas.
A diagnostics of reflectarray design was presented where expected results from a given
set of design parameters are provided which proves very handy while troubleshooting
a reflectarray design. To predict the radiation characteristics of a reflectarray, it is
necessary to simulate it in a full-wave EM solver package. Due to presence of a large
number of elements, to prepare a simulation model for such EM simulations present

a significant challenge. Through this research a tool was prepared to integrate EM
solver with the analytical software package to automatically build such a simulation
model with least inputs from a designer. This tool enabled a fast simulation model
preparation for EM simulations which would take days to build manually. This tool
enabled multiple iteration of simulation model building which would have not been
possible without it.
At mm-waves implementation of continuous phase reflectarrays is not feasible due to

complexity and limited real estate. Therefore, one need to have a discretized set of
phases. This research investigated the resulting performance due to various number of
phase quantization states. As a result a performance matrix was realized to predict
the performance of phase quantized large reflectarrays. When the number of phase
quantization states are selected in a design, a designer needs to select the actual
phases corresponding to the phase states of a selected phase quantizer. This research
provided an answer to this problem by finding through analysis and simulations that a
uniformly distributed phase states result in a minimal loss of directivity. Finally, the
power consumption of large smart reflectarrays was investigated. The trade-off among
reflectarray size (gain and number of unit cells), number of phase quantization states,
functionality (single or dual polarization) and DC power consumption was presented.
It was established through measurements and analytical analysis that large smart
reflectarrays consume a considerably less amount of DC power which favours their
integration in next generation communication systems.
Chapter 6
Reflectarray Measurements
This chapter presents the design, fabrication and measurements of passive

demonstrators of phase qunatized reflectarrays at V band. These reflectarrays were
designed and simulated in Matlab
R
using array based technique (discussed in section
5.1.3) to access their initial performance. To build EM simulation models of these
passive demonstrators for CST 3D EM simulations, a fast Matlab
R
based algorithm
(discussed in section 5.3) was used. Twelve passive demonstrators were fabricated to
validate the design of 2, 3, 4, and 8 states phase quantized reflectarrays (discussed in
section 5.2) at various beam pointing angles. Phase quantized reflectarrays having
phase quantization of 2, 4, and 8 states were designed to point their beams at
0◦ and 55◦ . Whereas, 3 states phase quantized reflectarrays were designed for 0◦ ,
15◦ , 30◦ , 45◦ and 55◦ . These designs would prove the performance attainable at
designed pointing angles. Therefore, these designs would result in an assessment of
the main lobe scan performance and radiation pattern performance. Similarly the
cross polarization discrimination (XPD) performance (discussed in section 2.12) with
scanning would be demonstrated through these designs. These reflectarrays were
measured using a roll/azimuth positioner in the anechoic chamber. Various performance
parameters including gain, beamwidth, bandwidth, pointing angle, side lobes, XPD
etc. were estimated through 3D EM simulations and are compared with the measured
results. The 3D radiation pattern was also measured for a few of the reflectarrays
and is discussed here. Finally, the chapter provides guidelines to implement active
194
6.1. Design of Reflectarray Passive Demonstrators 195
reflectarrays.
Section 6.1 discuss the design and fabrication of these passive demonstrators. In section
6.2 the measurement setup at the chamber is described along with the measurement
structure used for reflectarrays. Section 6.3 explains the measurements of passive
demonstrator reflectarrays where their measured performance parameters are discussed
and compared with simulated results. Section 6.4 discusses how to implement the
active reflectarrays using the designs presented in this thesis. Finally, this chapter is
summarized in section 6.5.
6.1 Design of Reflectarray Passive Demonstrators
Millimeter wave active reflectarrays implemented using discrete phase quantization

targeted in this research consists of thousands of unit cells each containing 4 PIN
diodes. Fabrication of such a structure, a large number of PIN diodes and their
mounting was found to be too expensive to be affordable by an educational research
project. As a result free PCB samples were requested from Rogers to implement passive
demonstrators at fixed radiation beam pointing angles. How the passive demonstrators
differ from their active counterparts? This section answers this question and covers the
design of such passive demonstrators.
In passive demonstrators a variable patch phase control technique was used. In this
technique, to produce a desired discrete reflection phase, the dimensions of patches
were controlled. To implement a 3 states phase quantization this resulted in a selection
of three different sizes of patches; one for each phase state. In a 3 states phase
quantized active reflectarray all the unit cells are physically similar. However, their
DC (direct current) biasing decides the reflection phase they produce from their
location. Therefore, the main differences among passive and their corresponding
active implementations are: physical geometric difference and application of DC power.
Passive demonstrators have no control components, therefore no DC power is supplied.
However, active unit cells need DC power to function. On the performance side
one should expect a higher loss in active unit cells in comparison to their passive
counterparts as discussed in Chapters 3 and 4. Another difference in performance
is the resulting cross polarization in an active implementation. Various techniques

identified in Chapter 2 can be used to overcome this issue.
Table 6.1 lists the characteristics of fixed passive demonstrator reflectarrays. To verfiy
the performance of phase quantized reflectarrays 12 passive demonstrators were designs
and fabricated. In this table, ‘No. of states indicate the number of states used in phase
quantization, ‘Spacing’ refers to the inter-element spacing, beam pointing angle is listed
as ‘Angle’, and the size of reflectarrays is given by ‘Elements’. Mainly the focus was
to verify the performance of 3 states phase quantized reflectarrays. Therefore, multiple
reflectarrays with 3 states phase quantization at different beam pointing angles were
fabricated. To compare the performance of various number of phase quantization states
at the maximum grating free beam pointing angle (as permissible through inter-element
spacing), 2, 4, and 8 states phase quantized reflectarrays were fabricated as well. The
phase discretization for each case is represented by (6.1) to (6.4) where a relation
between number of states and corresponding phases is formulated.
Table 6.1: Passive demonstrators of reflectarrays at fixed beam pointing angles. (F =

70 mm, L = W = 94.5 mm, qe = 1.6, qf = 4)
Array no. No. of states Spacing Angle Elements

1 3 2.7 mm 0◦ 35 × 35
2 3 2.7 mm 15◦ 35 × 35
3 3 2.7 mm 30◦ 35 × 35
4 3 2.7 mm 45◦ 35 × 35
5 3 2.7 mm 55◦ 35 × 35
6 3 2.0 mm 90◦ 47 × 47
7 8 2.7 mm 55◦ 35 × 35
8 8 2.7 mm 0◦ 35 × 35
9 2 2.7 mm 55◦ 35 × 35
10 2 2.7 mm 0◦ 35 × 35
11 4 2.7 mm 55◦ 35 × 35
12 4 2.7 mm 0◦ 35 × 35

π
 , 0 ≤ ∆ΦC % 2π < π State 1


2
∆ΦQ = (6.1)
2 states 3π
π ≤ ∆ΦC % 2π < 2π

 ,
 State 2
2

5π π
0,
 ≤ ∆ΦC % 2π < State 1
3 3




2π π

∆ΦQ = , ≤ ∆ΦC % 2π < π State 2 (6.2)
3 states  3
 3
4π 5π



 ,
 π ≤ ∆ΦC % 2π < State 3
3 3

 7π π

0, ≤ ∆ΦC % 2π < State 1
4 4





 π π 3π
 , ≤ ∆ΦC % 2π < State 2


2 4 4
∆ΦQ = (6.3)
4 states 3π 5π
≤ ∆ΦC % 2π <

π,
 State 3



 4 4

 3π 5π 7π
 ,

 ≤ ∆ΦC % 2π < State 4
2 4 4

9π 9π
0, − ≤ ∆ΦC % 2π < State 1


32 32




π 9π 3π


, ≤ ∆ΦC % 2π < State 2






 4 32 8

 π 3π 5π

 , ≤ ∆ΦC % 2π < State 3
2 8 8




 3π
 5π 7π
 , ≤ ∆ΦC % 2π < State 4


4 8 8
∆ΦQ = (6.4)
8 states 7π 9π
≤ ∆ΦC % 2π <

π,
 State 5



 8 8

 5π 9π 11π

 , ≤ ∆ΦC % 2π < State 6
 4 8 8



3π 11π 13π


, ≤ ∆ΦC % 2π < State 7





 2 8 8
7π 13π 15π



 ,
 ≤ ∆ΦC % 2π < State 8
4 8 8
Here ∆ΦQ is the discrete quantized phase shift introduced by a unit cell, ∆ΦC is the
desired continuous phase from that particular unit cell as calculated using (5.25), and
% represents the modulo (remainder) operator. These states are realized through the
selection of unit cell dimensions based on measurements in Chapter 3.
(a) Negative mirrored mosaic, 0◦ . (b) Negative mirrored mosaic, 55◦ .
0 0
5 5 200
200

10 10
150 150
15 15
20 20 100
100
25 25
50 50
30 30
35 35 0
0 5 10 15 20 25 30 35 0 0 5 10 15 20 25 30 35
Unit cells in x-direction Unit cells in x-direction
(c) Phase distribution, 0◦ . (d) Phase distribution, 55◦ .
Figure 6.1: A 3 states phase quantized reflectarray at 0◦ and 55◦ beam pointing angles.
Fig. 6.1 displays some of the mosaics used for fabrication and the corresponding
aperture phase distributions of phase quantized reflectarrays. Mosaics shown are
negative and mirrored images of the reflectarrays as required by the photolithographic
fabrication process. One can clearly observe the difference in phase distribution over
the reflectarray aperture at various beam pointing angles in these figures. According
to the number of states in each case listed in Table 6.1, the corresponding number of
reflecting patches of different sizes in unit cells were used to realize phase states. For
instance a 3 states phase quantization was implemented using three different sizes of
patches. Similarly, 8 different sizes were used to realize an 8 states implementation.
Fabricated reflectarrays are shown in Fig. 6.2.
6.2. Millimeter Waves Antenna Measurement Setup 199
Figure 6.2: Phase quantized fabricated reflectarrays. Each reflectarray was mounted
on a plastic support to provide physical strength to the 10 mil thick RO5880 grounded
substrate.
6.2 Millimeter Waves Antenna Measurement Setup
The millimeter wave antenna measurement system was based on an HP/Agilent 85309A
frequency converter. It consists of two sections which are; a transmitter and a receiver.
A system block diagram is shown in Fig. 6.3 [200]. Its transmit section is made of
a synthesized RF source, an amplifier (HP 8349B), a mm-wave source module (HP
8355xA), and the transmit antenna. A standard gain horn is used as the transmit
antenna. A mm-wave stimulus signal for the receive antenna which is an antenna
under test (AUT) is provided by the mm-wave source module and the transmit antenna.
Through a general purpose interface bus (GPIB interface), the RF source module is
controlled by the receiver.
Part of Tx Rx AUT HP11970 Mixer

HP8355x A
HP85325A HP365A
mm-wave 5m
Coupler Isolator RF LO
source module
IF
LO out
HP365A
IF in
Part of
HP8349B
Isolator HP85326A Test module

Amplifier HP11970 60001
RF LO/IF
Ref. Mixer
IF out Ref. IF in
SMI interface
LO
IF
Frequency
LO in Ref. LO out converter LO in
IF in Reference
module Ref. Det. in
RF LO out Detector out HP85309A
out
Part of Ref. IF Test IF
RF HP85326A LO
source 60002 Receiver source
Figure 6.3: Antenna measurement setup.
HP 8355xA mm-wave source module accepts an RF power input in the range of +17
to +27 dBm. RF source and HP 8349B amplifier together provide an output of +17
dBm for HP 8355xA mm-wave source module. HP 8355xA mm-wave source module
up-converts the RF frequency by a multiplying factor of 2 to 6 based on the module
used. A source module interface (SMI) allows the mm-wave source module to become
an integral part of the RF source. Due to this interface (with its cables connected),
the RF source acknowledges the fact that multiplied frequencies are in use. Therefore,
when the receiver asks for frequencies above the RF source’s normal range, it responds
correctly. Using the STIMULUS controls on the receiver one can change the output
power of the mm-wave source module.
The receiver section of the measurement system consists of an AUT, a reference

antenna or a directional coupler (shown as a part of HP 85325A: mm-wave interface
kit), frequency converter unit, mixer modules (part of HP 85326A: Test-60001 and
Ref-60002), mm-wave mixers (HP 11970, two units; one as a reference mixer and the
second with AUT), isolators (HP 365A with both mixers), and a receiver (HP/Agilent
8510 or 8530). At the input of each mm-wave mixer there is an isolator to improve the
impedance matching between the antenna and its mixer. An isolator also blocks mixer
harmonic signals from radiating through the antenna.
Mixers down-convert the mm-wave signal to a 20 MHz IF frequency for the receiver
(HP/Agilent 8510 or 8530). Both mixers (test and ref.) are identical. However, they are
driven by different mixer modules. These mixers operate normally with an LO power in
the range of 16 ± 2 dBm. The reference mixer module (60002) measures the LO power
it receives, through an internal detector. The output voltage of the LO power detector
are fed to HP/Agilent 85309A LO/IF unit which controls the LO power. LO power to
the test module (60001) is not measured. The test module receives the same amount
of LO power as received by the reference module due to assumed similar electrical
distances. Therefore, test and reference LO powers are controlled simultaneously. Test
mixer module has a diplexer to allow LO and IF signals to travel through the same
cable therefore, permitting a single ring rotary joint for an AUT.
The LO signal is provided by an LO source unit. HP/Agilent 85309A LO/IF unit,

amplifies and distributes LO signals to mixer modules. HP/Agilent 85309A also
controls the LO power levels through an automatic level control circuitry. It receives IF
signals from both mixer modules, amplify them and send them to the receiver. At the
receiver 20 MHz IF signals are converted to the digital data, processed and displayed.
A receiver also controls the frequencies of LO and RF sources. For more details on the
system operation one can refer to [200].
In Fig. 6.4 the transmit and receive ends of the chamber are shown with antennas. The
transmit side consist of the transmit horn antenna and its related assembly is placed
behind it as shown in Fig. 6.4 (b). mm-wave module, its coupler and reference mixer
along with the transmit antenna polarization rotation motor can be seen in Fig. 6.4
(b). Whole electrical assembly shown in Fig. 6.4 (b) on transmit side rotates when a
polarization rotation is required. Fig. 6.4 (c and d) display the receive side antenna
positioner with AUT. For reflectarray measurements the mm-wave 1.8 mm cable from
WR-15 of AUT runs to the mixer input WR-15 adapter (through a waveguide isolator).
The antenna positioner is a roll over azimuth positioner. AUT assembly is mounted on
the positioner through a plastic interface plate with M8 bolts at 179.4 mm PCD.
(a) Transmit side (b) Tx assembly (c) AUT under test (d) AUT mounting
Figure 6.4: Antenna measurement chamber setup

.
(a) Plastic support (b) Plastic support (c) Rohacell support (d) Rohacell support
Figure 6.5: AUT support structures.

.
Fig. 6.5 display support structures used in reflectarray tests. Initially, a plastic holding
frame with unknown electrical properties was designed for holding the feed horn in
place as shown in Fig. 6.5 (a and b). During tests it was found that the sidelobe
levles were getting severely effected due to the plastic support structure. Therefore, a
Rohacell support was made during tests. In Fig. 6.5 (c and d) the Rohacell support
structure is shown to hold the feed assembly. In this structure the interface plate with
positioner is still plastic while front and lower portions are replaced by Rohacell.
During tests ‘3 antennas test method’ was used to find the gain. A reference antenna
with known gain was connected at the receive end and the received signal was measured
for comparison purpose to determine the gain of fabricated reflectarrays.
6.3. Measurement of Passive Demosntrators 203
The antenna test facility is a near field facility with a range (Tx-Rx distance) of 5
m. The receiver AUT positioner rotates in roll 360◦ and in azimuth by ±90◦ . The
positioner has two modes of operation which are: move with a given incremental step
and measure the sample or take measurements on the fly. During incremental step
based measurements, the positioners moves to the desired position, stops there, take
the measurement and then move to the next location. While in measurements on
the fly, measurements are taken when positioner is rotating. Due to sensing system
of the positioner, this method do not provide an accurate angular position. With
measurements on the fly, the radiation pattern can be found however, the exact angles
at which the maximum of radiation pattern occurs are not very accurate and calibrated.
Measurements on the fly is a fast way of measuring the radiation pattern in comparison
to incremental step approach which is more positionally accurate however very slow.
The measurement facility is mainly designed for 3D radiation pattern measurements
which are very time consuming. It was not possible to measure the 3D radiation
patterns of all of the fabricated reflectarrays in the allocated time which was only one
week.
Therefore, the fabricated reflectarrays benefited from 2D radiation pattern

measurements in the far field, where azimuth cuts (−90◦ ≤ θ ≤ 90◦ ) were recorded
at a fixed elevation. Only a few 3D patterns were recorded.
6.3 Measurement of Passive Demosntrators
The measurement results of the passive demonstrator reflectarrays fabricated for

their performance verification for various number of phase quantization states and
at various pointing angles are described in this section. Tables 6.2 and 6.3 list the
major performance parameters of these passive demonstrators. In these tables ‘# of
states’ = number of phase quantization states of the reflectarray, ‘Angle design’ = the
theoretical pointing angle used in design calculations, ‘Gain CST’ = gain achieved in
CST Microwave Studio 3D EM simulations using time domain (T) solver, ‘Angle CST’
= pointing angle achieved in CST simulations, ‘Bandwidth CST’ = achieved 3 dB
gain bandwidth in CST simulations, ‘Gain test’ = measured gain using 3 antenna test
method, ‘Angle test’ = measured pointing angle of the reflectarray, ‘Bandwidth test’ = 3
dB gain bandwidth from measurements, ‘States:angle’ = number of phase quantization
states and design pointing angle, ‘Beamwidth CST’ = 3 dB beamwidth achieved in
CST simulations, ‘SLL CST’ = sidelobe level in CST simulations, ‘XPD CST’ = cross
polarization discrimination achieved in CST simulations, ‘Beamwidth test’ = is the
measured 3 dB beamwidth, ‘SLL test’ = measured sidelobe levels, and ‘XPD test’ =
measured value of XPD in the main beam. Mainly, the CST results are based on 60
GHz center frequency while measured results are based on 61.5 GHz center frequency.
At feed horn port of reflectarrays the impedance matching was excellent (S11 ≤ −20
dB) and is not discussed here. One of the fabricated reflectarray ‘Array no. = 6’ in
Table 6.1 could not be fabricated well due to limitations of fabrication facility therefore,
was discarded and is not discussed further here.
All measurements were performed using the plastic support structure for reflectarrays
and the feed horn. The only exception was the case where the sidelobe level was
measured for a 3 states reflectarray pointed at 0◦ . For this particular case the sidelobe
level was measured using Rohacell support. Plastic support deteriorates the sidelobe
performance in addition to a gain reduction in the near boresight region. This loss
in gain reduces as the main beam is pointed away from the boresight. Similarly, the
sidelobe performance starts getting better while moving away from the blockage due to
plastic support. Near the boresight plastic support also effect the shape of the radiated
beam. Rohcell is a good solution to avoid blockage. However, as the Rohacell is a soft
material, it is hard to maintain the dimensional accuracy of distances and geometrical
pointing angles when support structure is build on the run. It is worth noting that these
distances and pointing angles are very important in reflectarray operation particularly
at mm-waves were a slight change can lead to a significant phase error. Therefore most
of the results were generated using the plastic support structure which offered a rigid
support. Another source of blockage in measurements was the feed horn and waveguide
(WR-15) adapter flange. It is significantly bigger (19.1 mm) as compared to the feed
horn’s aperture (7.2 mm) therefore, a gain reduction of 0.4 dB was observed in CST
simulations with and without the flange at 0◦ beam pointing. All the results from CST
simulations stated in these tables are without the effect of waveguide flange considered.
Table 6.2: A comparison of CST simulated and measured results for passive
demonstrators of reflectarrays at fixed beam pointing angles. (35 × 35 elements,
inter-element spacing = 2.7 mm, F = 70 mm, L = W = 94.5 mm, feed horn model
LB-15-10-A from A-Info, frequency for CST results = 60 GHz, frequency for test results
= 61.5 GHz)
# of Angle Gain Angle Bandwidth Gain Angle Bandwidth

states design CST CST CST test test test
(dBi) (GHz) (dBi) (GHz)
3 0◦ 33.17 0◦ 9.69 31.48 0◦ 10.21
3 15◦ 33 14.8◦ 7.04 31.21 15◦ 9.41
3 30◦ 32.51 29.8◦ 5.12 31.8 29.6◦ 6.22
3 45◦ 31.41 44.8◦ 4.07 31.2 44.8◦ 4.17
3 55◦ 29.54 54.5◦ 3.5 30.22 54.2◦ 3.41
2 0◦ 31.65 0◦ 9.78 29.54 0◦ 8.96
2 55◦ 27.38 54.5◦ 3.23 28.19 54.6◦ 3.73
4 0◦ 34.13 0◦ 7.83 31.28 0◦ 8.93
4 55◦ 30.2 54.3◦ 3.33 30.75 54.2◦ 3.8
8 0◦ 34.55 0◦ 8.92 33.26 0◦ 9.37
8 55◦ 30.7 54.4◦ 3.26 31.39 54.2◦ 3.51
These measurements are made at 5 m distance between transmit and receive antennas.
The measurements were made using the roll over azimuth antenna positioner at the
anechoice chamber. For a fixed value of roll (elevation in these cases), the azimuth was
varied in a limited range to capture the main beam and few sidelobes around it.
The CST simulated gain and measured values of gain are in close agreement considering
the loss due to support structure and the feed flange. At boresight in all cases the
difference is about a dB in addition to the losses stated above. This loss can be easily
related to the loss in unit cells due to substrate and copper losses. In CST simulations
with and without substrate losses, there was a negligible difference. However, copper
loss was not considered in CST simulations while simulating reflectarrays. In fact the
plastic support is too lossy at and near the boresight as is clear from the results at
55◦ pointing cases. To fully simulate the support structure in CST was not possible
due to its larger electrical size at 60 GHz. Therefore, only the feed horn support was
modeled in CST to observe the gain reduction effect at the reflectarray boresight. It
was observed that the sidelobes were significantly higher when support structure was
introduces in simulations. A gain reduction of about 2 dB was observed in simulations
at the boresight due to the feed support structure. In any way to have only around
a dB loss in gain near the boresight due to these losses, in an actual reflectarray
measurements, is a very promising result.
Table 6.3: A comparison of CST simulated and measured results for passive
demonstrators of reflectarrays at fixed beam pointing angles. (35 × 35 elements,
inter-element spacing = 2.7 mm, F = 70 mm, L = W = 94.5 mm, feed horn model
LB-15-10-A from A-Info, frequency for CST results = 60 GHz, frequency for test results
= 61.5 GHz, sidelobe level = SLL, cross polarization discremination = XPD.)
States: Beamwidth SLL XPD Beamwidth SLL XPD

angle CST CST CST test test test
(dB) (dB) (dB) (dB)
3: 0◦ 3◦ -19.8 > 40 2.87◦ -16 37
3: 15◦ 3.2◦ -19.1 > 40 4.11◦ - -
3: 30◦ 3.6◦ -18.9 > 40 3.65◦ - 38
3: 45◦ 4.4◦ -17.6 > 40 5.57◦ - -
3: 55◦ 5.5◦ -18.4 > 40 5.05◦ -17 38
2: 0◦ 3◦ -21.1 > 40 2.87◦ - -
2: 55◦ 5.2◦ -14.6 > 40 4.82◦ -12 35
4: 0◦ 3◦ -20.7 > 40 2.43◦ - 30
4: 55◦ 5.5◦ -17.1 > 40 5.2◦ -14.3 39
8: 0◦ 2.9◦ -23.7 > 40 2.75◦ - -
8: 55◦ 5.5◦ -17.2 > 40 5.06◦ -14.2 -
0
8 states
4 states
−1 3 states
2 states
Loss in gain (dB)
−2
−3
−4
−5
−6
0 10 20 30 40 50 60
Measured beam pointing angle (deg)
Figure 6.6: Measured loss in gain with achieved beam pointing angles for reflectarrays
having various number of phase qunatization states. Values are normalized w.r.t. the
gain of 8 states case at 0◦ . Measurement points are shown by geometric symbols at the
frequency of 61.5 GHz.
The loss is gain at various pointing angles for measured reflectarrays w.r.t. 8 states
case at 0◦ is shown in Fig. 6.6. The gain reduces as the beam is pointed away from the
boresight. Due to multiple 3 states reflectarrays for comparison, the gain reduction at
boresight is very obvious from this figure. In measurements only the combined effect
of gain improvement (less effect of structure and blockage etc.) and gain reduction
is observed when the beam is pointed away from the boresight. When the beam is
pointed significantly away from the boresight e.g. at 30◦ and 45◦ the reduction in
gain is dominated by the improvement in gain by avoiding the blockage. It is very
interesting to observe the performance of a 4 states phase quantized reflectarray to
be less than that of a 3 states phase quantized reflectarray at the boresight. The
degradation in 4 states case is not huge however, one should theoretically expect it to
perform a little better than the 3 states case. Two obvious reasons are the selection of
its phase states and the fabrication process. Although, a significant care was practiced
during measurements, it is quite possible to experience the physical mounting of the

4 states boresight reflectarray to go wrong during measurements which lead to a gain
reduction. The loss for a 2 states phase quantization case is significant in comparison
to the case of a 8 states phase quantization as was the case in theory. Table 6.4 lists the
scan loss in gain due to beam pointed away from the boresight for each type of phase
quantization in these reflectarrays. For a 3 states phase quantized reflectarray, when
the beam is pointed from 0◦ to 55◦ a scan loss of 1.26 dB is observed. This table also
lists the loss in comparison to two cases (0◦ and 55◦ ) of a 8 states phase qiunatized
reflectarray. A very graceful reduction in gain is observed when the beam is scanned
from 0◦ to 55◦ .
Table 6.4: Gain reduction versus scan angle
States: Scan loss (dB) Loss w.r.t. 8 states Loss w.r.t. 8 states
angle case at 0◦ (dB) case at 55◦ (dB)
8: 0◦ 0 0 -
8: 55◦ 1.87 1.87 0
4: 0◦ 0 1.98 -
4: 55◦ 0.53 2.51 0.64
3: 0◦ 0 1.78 -
3: 15◦ 0.27 2.05 -
3: 30◦ -0.32 1.46 -
3: 45◦ 0.283 2.06 -
3: 55◦ 1.26 3.033 1.17
2: 0◦ 0 3.72 -
2: 55◦ 1.35 5.07 3.2
From CST simulations and measurement results it is concluded that all the phase
quantized reflectarrays can achieve the required beam pointing angle with a minimal
mis-pointing as compared to the theoretical case of continuous phases as shown in Fig.
6.7. Numerical values of achieved pointing angles are listed in Table 6.2. An excellent
agreement can be observed among continuous phase (theoretical) and phase quantized
60
Achieved beam pointing angle (deg)

50
40
30
Continuous
20 8 states
4 states
3 states
10 2 states
0
0 10 20 30 40 50 60
Desired beam pointing angle (deg)
Figure 6.7: Measured beam pointing angles for reflectarrays having various number of
phase qunatization states. The continuous phase case is plotted for reference as was
calculated using Matlab
R
. Measurement points are shown by geometric symbols at
the frequency of 61.5 GHz.
reflectarrays in this figure.
The bandwidth response of phase quantized reflectarrays is listed in Table 6.2. A

general trend is: when the beam is scanned away from the boresight the bandwidth
at the desired particular pointing angle reduces. The bandwidth response shall be
discussed further in next section. From Table 6.3 it can be observed that the beamwidth
increases as the main beam is scanned away from the boresight which results in a
gain reduction. At 55◦ from the boresight, a 3 dB beamwidth of 5.5◦ was observed.
Sidelobe levels are good in CST simulations which are without the feed horn flange and
support structure. In measurements the sidelobe levels are within 4 dB (in the worst
case) of the CST simulated values. The main reason for this difference is the blockage
due to support and feed horn’s flange which was not considered in CST simulations.
For a 3 states reflectarray pointed at 55◦ this difference is only 1.4 dB. The sidelobe
values are not provided in cases where it is considered that the blockage is significantly
effecting the radiation pattern. To measure the sidelobe levels of a 3 states phase
quantized reflectarray pointed at 0◦ the feed horn support was replaced by the Rohacell
and a significant improvement of sidelobe levels was observed. It is observed that
2 states phase quantization leads to highest sidelobes among all the cases where as
3 states cases can achieve a significantly low sidelobe levels. The cross polarization
discrimination (XPD) was greater than 40 dB in CST simulations. XPD values close
40 dB were measured during the measurements. XPD values were measured for only a
few reflectarrays as listed in the Table 6.3. Based on the measured XPD results it can
be concluded that all the fabricated reflectarrays can achieve an excellent XPD value.
In the next section radiation pattern, pointing angle and bandwidth characteristics are
discussed in more details.
6.3.1 2D Radiation Pattern and Bandwidth Response
In this section the radiation pattern, bandwidth and frequency scanning performance
is discussed for fabricated reflectarrays for 2, 3, 4, and 8 states phase quantized
reflectarrays for various beam pointing angles. Extracted results from measurements
are shown in Figures 6.8 to 6.18. In each of these figures, the sub-figure (a) represents
the radiation pattern at center frequency with XPD results where possible. In sub-figure
(b) the bandwidth response for the desired pointing angle is shown. In this figure the
bandwidth responses from the CST simulation and measurements are shown. The
measured pointing angle is also shown at which the bandwidth response is found. The
gain values are normalized w.r.t. the peak gain values from Table 6.2.
In Fig. 6.8, results of the 3 states phase quantized reflectarray designed to point its
main beam at 0◦ are shown. The measured beam pointing can be observed at 0◦ . The
cross olarization response is very good. An increased sideleobes are due to the support
structure. Similarly, the bandwidth response from CST simulations match very well
with the measured bandwidth response.
In Fig. 6.9 the characteristics of the 3 states phase quantized reflectarray pointed at
15◦ are shown. Due to asymmetric blockage the radiation pattern is quite distorted.
0 0 5
Normalized gain (dB)

-2
Pointing angle (deg)

-10
-4
-20 Co-pol H plane
0
Co-pol E plane
-6
-30 Cross pol
Gain measured
-8 Gain CST simulated
Pointing angle
-40
-10 -5
-10 -5 0 5 10 55 60 65 70
Angle θ (deg) Frequency (GHz)
(a) Radiation pattern at 61.5 GHz (b) Bandwidth response at 0◦
Figure 6.8: Response of a V band 3 states phase quantized reflectarray designed to

point its beam at 0◦ , a comparison of measured results with CST simulated results.
0 0 20
18

-10
-5
16
-20
Co-pol
14
-30 -10
Gain measured
Gain CST simulated 12
Pointing angle
-40
-15 10
55 60 65 70
5 10 15 20 25
Frequency (GHz)
Angle θ (deg)
(a) Radiation pattern at 61.5 GHz (b) Bandwidth response at 14.8◦

When beam is scanned at 30◦ as shown in Fig. 6.10 the sidelobe performance gets
better.The bandwidth reduces as the beam is scanned away from the boresight which
is observable here for the beam pointed at 30◦ .

0 35

-10
-5
-20
30
Co-pol
Cross pol -10
-30 Gain measured
Gain CST simulated
-40 Pointing angle
-15 25
55 60 65 70
10 20 30 40 50
Frequency (GHz)
Angle θ (deg)
(b) Bandwidth response at 29.8◦
(a) Radiation pattern at 61.5 GHz

0 0 50
Co-pol
-5
48

-10 -5
46
-15
44
-20 -10
Gain measured
-25 Gain CST simulated 42
Pointing angle
-30 -15 40
30 40 50 60 55 60 65 70
(a) Radiation pattern at 58 & 61.5 GHz (b) Bandwidth response at 44.8◦

In Fig. 6.11 the response of the 45◦ , 3 states reflectarray is shown. The sidelobe
performance is significantly improved. Radiation pattern curves against frequency are
pointed at varying pointing angles. There is a significant reduction in bandwidth at the
desired pointing angle when compared to the 0◦ pointed reflectarray. It was observed
that the gain starts decreasing sharply at higher frequencies. This is mainly due to the
inter-element spacing design at 60 GHz for all the arrays.
Fig. 6.12 provides the details of 3 states phase quantized reflectarray pointed at 55◦ .
Due to inter-element spacing of 2.7 mm, the design can point the beam up to 55◦ at
60 GHz without generating grating lobes in the radiation pattern. The beam can be
further scanned up to 78◦ , however gain reduces severely. Here, the sidelobe and XPD
performance are good. Bandwidth at 55◦ is reduced as is clear from sub-figure (b) and
observed from CST simulations & measurement results.
0 0 60

-10 58
-5
56
-20
Co-pol
Cross pol 54
-30 -10
Gain measured
-40 Pointing angle
-15 50
40 50 60 70 55 60 65 70

The performance of a boresight pointed 2 states phase quantized reflectarray is shown

in Fig. 6.13. Here the unequal sidelobes on either side of main beam indicate a
physical mis-pointed array during measurements. The beam is pointed at 0◦ and the
bandwidth response is shown in sub-figure (b), where a good agreement of CST and
measured results can be observed.
When a 2 states phase quantized reflectarray is pointed at 55◦ the resulting performance
is shown in Fig. 6.14. Here relatively higher sidelobes at 55◦ are observed in comparison
to other reflectarrays discussed in this section. The simulated and measured bandwidth
response at 55◦ matches very well.
0 0 5

-5

-10 -5
-15 0
-20 -10
Gain measured
Gain CST simulated
-25 Co-pol
Pointing angle
-30 -15 -5
-20 -10 0 10 20 55 60 65 70

0 0 60
58

-10
-5
56
-20
Co-pol
54
Cross pol -10
-30 Gain measured
Pointing angle
-40
-15 50
55 60 65 70
40 50 60 70
Frequency (GHz)
Angle θ (deg)

A boresight pointed 4 states phase quantized reflectarray’s measured performencae is

shown in Fig. 6.15 where a very good agreement of CST simulation and test results
can be observed.
Similarly, the perfromance of the 4 states phase quantized reflectarray pointed at 55◦
is shown in Fig. 6.16, where CST and measured results can be observed in close
agreement.
0 0 5
-5 Co-pol

-10 -5
-15 0
-20 -10 Gain measured

Gain CST simulated
-25 Pointing angle
-30 -15 -5
-20 -10 0 10 20 55 60 65 70

0 0 60
58
-10
-5
56
-20
Co-pol
Cross pol 54
-30 -10
Gain measured
-40 Pointing angle
-15 50
40 50 60 70 55 60 65 70

The performance of two 8 states phase qunatized reflectarrays pointed at 0◦ and 55◦
is shown in Fig. 6.17 and 6.18. In both cases the simulated and measured results are
in a very good agreement.
0 0 5
Co-pol
-5

-10 -5
-15 0
-20 -10
Gain measured
Gain CST simulated
-25
Pointing angle
-30 -15 -5
-20 -10 0 10 20 55 60 65 70

0
0 60
-5
58
-10 -5 Pointing angle (deg)

56
-15
54
-20 -10
Gain measured
-25 Co-pol Gain CST simulated 52
Pointing angle
-30 -15 50
40 50 60 70 55 60 65 70


From Figs. 6.8 to 6.18 it can be observed that phase quantized reflectarrays can
significantly achieve the desired performance in terms of pointing angles and radiation
pattern characteristics. The simulated and measured bandwidth response at any
particular pointing angle are in an excellent agreement which validates the analysis
in CST. However, to perform one such a simulation in CST took 30 hours and required
an advanced computing cluster. Through these simulations and measurements it is
also found that array based approach used to analyze the reflectarrays is a significant
simplification for fast prediction of radiation pattern and pointing angle for continuous
as well as phase quantized reflectarrays. For gain prediction one can rely on either 3D
EM simulation in CST or can use simply the overall efficiency in the range of 0.4 to
0.6. Although the product of illumination and spillover efficiencies was significantly
high in designs. However, the phase quantization, unit cell losses, and blockage losses
etc. reduced the achievable overall efficiency in reference to the physical aperture
dimensions to a value of 0.45 for 3 states phase quantized reflectarray and 0.6 for a 8
states phase quantized reflectarray. Even by counting the unit cell and all other losses in
a reflectarray it can still achieve the efficiency performance of a reflector antenna which
is usually close to 0.6 which is an excellent result established through this research.
These efficiency values do not consider the gain reduction due to blockage effects in
measurements. The overall efficiency factor can be further improved by avoiding the
blockage due to support structures. This is achievable even in center fed reflectarrays
where the main beam can be pointed off boresight and feeding can be still in the
aperture center. Such a versatile combination of central feeding and off boresight main
beam pointing in reflectarrays do not have a counterpart in normal reflector antennas.
In Fig. 6.19 the bandwidth response for various pointing angles cases of the 3 states
phase quantized reflectarrays is shown. It is clear that the bandwidth reduces when
the main beam is pointed at the desired pointing angles w.r.t. the bandwidth at the
boresight.
Normalized gain (dB) -2
-4
-6
-8
Gain measured 0◦
-10
Gain measured 15◦
Gain measured 30◦
-12 Gain measured 45◦
Gain measured 55◦
-14
55 60 65 70
Frequency + 1.5 (GHz)
Figure 6.19: Measured normalized gain versus frequency for a 3 states phase quantized
reflectarray at various pointing angles to show bandwidth response at pointing angles.
As the fabricated reflectarrays were measured at a distance between transmit and

receive antennas which was just equal to the start of far field distance. Here
a comparison of results is presented with 3D measurements. On one hand, at
the stated distance one can regard measurements as far field measurements. On
the other hand the distance can be still considered as not too good for far field
measurements. Therefore, full 3D measurements were performed for the 55◦ pointed
3 states phase quantized reflectarray with Rohacell support considering them as near
field measurements. Similarly, the measurements were performed considering them as
the far field measurements at a fixed roll (elevation here) while varying the azimuth
in a limited range for the same reflectarray however with the plastic support. In both
cases the distance between Tx/Rx antennas was 5 m. The 3D near field data was
transformed to far field using SNIFTD software package from TICRA. A comparison
of both cases is presented in Fig. 6.20. An excellent agreement can be observed in both
cases. This also identifies that the support has negligible effects when the main beam
is pointed well away from the boresight.
-5
-10
-15
-20
-25
-30
-35
-40 Co-pol transformed far field

Co-pol near field at just far field distance
-45
0 10 20 30 40 50 60 70 80 90
Angle θ (deg)
Figure 6.20: A comparison of the transformed far field, and the near field for a 3 states
phase quantized reflectarray designed for 55◦ beam pointing. As the distance between
antennas is just equal to the start of far field distance, this measured near field is termed
as “near field at just far field distance”.
6.3.2 3D Radiation Pattern
The 3D radiation pattern of the 3 states phase quantized reflectarray pointed at 55◦
was measured and transformed to far field. A frequency step of 0.5 GHz was used in the
band of 55 GHz to 70 GHz for the measurement purpose. Only the results at 60 GHz
are plotted in Fig. 6.21 where various views can be seen for the same 3D radiation
pattern. In Fig. 6.21 (a), only one main lobe can be observed at θ = 54.2◦ which
clearly indicates the beam pointing capabilities of the phase quantized reflectarrays. It
can also be observed in Fig. 6.21 (b) that the sidelobe levels are significantly low. The
beamwidth in both planes is little different as the beam scanning takes place only in
one plane at a fixed angle in the second (orthogonal) plane. At 60 GHz the details of
2D radiation patterns in both planes is discussed in the next paragraph.
Fig. 6.22 to 6.24 display the 2D radiation pattern characteristics of the above
reflectarray. The far field was generated from Roll/Azimuth 3D near field measurements
and processed in Matlab
R
to generate these curves. Fig. 6.22 plots the 2D radiation
pattern in θ plane which here is called Plane 1. The main beam can be seen pointed at
56◦ (which in fact is the 54.2◦ due to 1.8◦ error of the antenna positioner) at 60 GHz.
In Fig. 6.23 the 2D radiation pattern in φ plane is plotted at a fixed angle θ = 55◦ .
Here the beam can be seen pointed at 180◦ which in fact is the φ = 0◦ when compared
with the general antenna co-ordinates. Because beam pointing takes plane in θ plane
only, therefore, the radiation pattern in Fig. 6.23 (Plane 2 = φ plane) has a 0.63 dB
higher gain in comparison to the gain in Fig. 6.22 (Plane 1 = θ plane).
Fig. 6.24 displays both plane results together with an inset focusing the difference in
peaks of both planes. In this figure, the θ axis is displayed as abscissa while φ axis is
shifted to bring both peaks together.
(a) 3D view of the radiation pattern.
(b) Theta-Phi view of the 3D radiation pattern.
Figure 6.21: Measured 3D radiation pattern at 60 GHz of a 3 states phase quantized

reflectarray designed for (θ, φ) = (55◦ , 0◦ ). The color bar represent the measured gain
(dBi).
30
Plane 1
20
Radiation pattern (dB)
10
-10
-20
-30
-40
-50
0 20 40 60 80 100 120 140 160 180
Measurement angle ( θ) (deg)
Figure 6.22: 2D radiation pattern in θ plane (called Plane 1).
30
Plane 2
20
10
-10
-20
0 50 100 150 200 250 300 350
Measurement angle ( φ) (deg)
Figure 6.23: 2D radiation pattern in φ plane (called Plane 2) at θ = 55◦ .

6.4. Active Reflectarrays 223
35
31
Plane 1
30
30 Plane 2
25
29
20
15 28
54 56 58
10
-5
-10
-15
-20
0 10 20 30 40 50 60 70 80 90
Measurement angles ( θ and φ ) (deg)
Figure 6.24: 2D radiation pattern in θ and φ planes (called Plane 1 and 2 respectively).
6.4 Active Reflectarrays
To implement active reflectarrays the passive unit cells are replaced by the active
3 states unit cells. The active unit cell structure remains same throughout the
reflectarray. The only difference among these unit cells would be the appropriate
provision of DC biasing for each unit cell. Appendix B provides the calculated
continuous reflection phases, discrete phases and required DC bias voltages for a 35 ×
35 unit cells reflectarray. In an active large reflectarray, there are mainly three problems
faced. These are losses due to active unit cells, provision of DC bias, and dissipation
of heat generated. A low loss active unit cell design was realized in Chapter 4 (section
4.3) to avoid excessive losses in unit cells. It was established in Chapter 5 (section 5.5)
through measurements in Chapter 4 (section 4.2) that a low DC drive is preferable in
active large reflectarrays which leads to a low power consumption therefore, less heat
6.4. Active Reflectarrays 224
to dissipate and control. From Chapter 5 a trade-off of power consumption with gain
and functionality (section 5.5) leads to a low power design. In case there are still issues
due to heat one can use heat sinks behind the reflectarray substrate when in space.
The heat generated may be useful to maintain the antenna physical temperature when
mounted on the satellite. By selecting a low power design presented in Chapter 5, the
heat dissipation issue gets resolved significantly.
The provision of DC biasing to individual unit cells in a large reflectarray is a challenging

task as there are as many DC bias lines as the number of unit cells. This issue can be
resolved by fabricating sub-arrays of a limited size where low voltage latch and hold D
type flip flops can be used to bias the diodes. Unit cells can be arranged in rows and
columns to make a matrix like structure. The control lines can be equal to the number
of columns while each row is enabled by a shift register in a serial fashion. From each
low voltage D type flip flop the output would pass through a switch which selects the
control voltage to be +DC or -DC for the unit cell. The function of these switches as
well as the column control lines and sequential row selection (using clock driven serial
shift registers which enable the D type flip flops of a row at a time) are controlled at the
host computer. Controlling the unit cells in a sequential fashion would lead to delay in
beamforming, however it greatly simplifies the DC bias design. As the digital circuits
can work to several MHz clock, therefore expected delay in beamforming would be not
a huge factor.
6.5 Chapter Summary
This chapter presented the design, fabrication and characterization of passive

demonstrators of V band phase quantized reflectarrays implemented through various
number of phase quantization states and pointed at various angles. Twelve reflectarrays
were fabricated for measurements. Out of twelve it was found that eleven reflectarrays
(inter-element spacing of 2.7 mm) were fabricated successfully using a low frequency,
low cost fabrication facility. One reflectarray which was not fabricated successfully
was implementing for inter-element spacing of 2.0 mm (to point its beam at the
maximum allowed angle of 78◦ ) which was too tight for the fabrication facility. Based
on the experience gained through the fabrication exercise, it was found that mm-wave
reflectarrays can be reliably and easily fabricated using low cost fabrication facilities
which are not necessarily high frequency facilities. For this to happen, a designer needs
to choose simple shapes of reflecting unit cells and a good inter-element spacing such
that the etching is possible.
Initial performance assessment of these reflectarrays was achieved in Matlab

R
using
array theory. EM simulation models were prepared and 3D EM simulations were
performed in CST which were very time consuming however, built an excellent level
of confidence in using Matlab
R
based analysis for fast prediction of results. These
reflectarrays were measured in the anechoic chamber. A very good correlation of
predicted and measured performance parameters was observed for all the eleven
reflectarrays. A great level of confidence was achieved through these measurement
results’ correlation with predicted results. All the phase quantized reflectarrays
achieved their performance goals and can point their beams to the desired angles while
maintaining their radiation pattern characteristics. A very close agreement of measured
and simulated gains verified the designs procedures and confidence in the unit cells’s
performance prediction and measurements. This chapter provided guidelines on the
implementation of active reflectarrays based on unit cells design presented in Chapter
4 (section 4.3).
Chapter 7
Conclusion and Future Research
This chapter presents conclusions drawn from the research on high gain wide angle
beamsteerable reflectarray smart antennas presented in this thesis. Additionally,
limitations of this research and possible potential research directions in the field of
high gain smart antennas are discussed in this chapter as well.
7.1 Thesis Summary
High gain smart antennas are considered as the latest frontier in the next generation
millimeter wave terrestrial / satellite communication systems as well as in radar tracking
applications. This research was focused at realizing an optimum solution for mm-waves
(V band = 60 GHz) high gain, wide angle electronic beamsteerable smart antennas. To
realize practically implementable designs of high gain smart antennas at mm-waves is
a potential challenge. At these frequencies tradiational beamforming practices do not
perform well due to a very large number of antenna elements in a mm-wave high gain
array. This thesis investigated a potential alternate to the mm-wave beamforming to
simultaneously achieve high gains and wide angle beam scanning capabilities. While
realizing these potentially desired objectives, the implementation perspective was given
a due consideration which resulted in a simplified yet high performance design.
This thesis examined high gain mm-waves printed smart reflectarrays at V band. High
gain and wide angle beamsteering capabilities were achieved by spatially illuminating
the thousands of comprising high performance unit cells in the reflectarray aperture
226
7.1. Thesis Summary 227
through a feed horn. Due to spatial illumination, the losses and complexities associated
with array beamformers were eliminated. Each unit cell was engineered to produce a
desired effect from its location in the reflectarray aperture to achieve a desired far field
objective. The presence of a large number of unit cells in the reflectarray enabled
diverse capabilities including the implementation of any phase only far field synthesis
technique.
As a part of this research mm-wave passive unit cells were characterized including their
selection, design, fabrication and measurements. A computationally fast approach for
parametric studies of unit cells for their performance characterization was realized.
To ensure a high accuracy at mm-waves, the analytical model included the effects of
fringing fields, metal conductivity, metal surface roughness, and surface waves. It was
established that parametric studies are highly important to achieve reliable performance
objectives from a unit cell design by exploiting the available degrees of freedom for
which the availability of such a fast technique proved very useful. A large number
of unit cells were designed, fabricated, and measured. A close agreement of measured
results with analytical and CST simulated results was observed. A procedure to extract
the RLC circuit based equivalent model of the mm-wave unit cells was formulated
in the form of closed form mathematical expressions. Through analytical analyses,
CST simulations and mm-wave measurements it was established that the extracted
lumped element circuit based model can represent a mm-wave unit cell with an excellent
agreement of results. Two new structures of unit cells were proposed to ease the
measurements at mm-waves. The measurement results of unit cell designs based on
these new structures were compared with the results of an ordinary unit cell design in
a waveguide measurement setup and an excellent agreement was observed. These unit
cell structures greatly simplify the measurements of active unit cells by facilitating the
provision of DC biasing for reconfiguration devices, and eliminating the alignment and
orientation issues faced in ordinary designs.
After formulating the design procedure and characterizing the passive unit cells,
an optimum technique to incorporate reconfiguration in unit cells was devised to
implement active/reconfigurable unit cells. These active unit cells with reconfiguration
mechanism ultimately realize an active smart reflectarray. The performance of these
active unit cells directly impacts the performance of an active smart reflectarray.
Therefore, the challenge to realize high performance active unit cells at mm-waves
was accomplished through this research. As a result, the design of a potentially
simplified high performance practically implementable mm-wave reconfigurable unit
cell was realized. To implement a continuous reflection phase in unit cells at mm-waves
is presently not possible due to the non-availability of electronic devices or these
devices being too lossy as well as exhibit a non-linear behavior which contradicts
the high performance objectives. Through a detailed investigation of reconfiguration
technologies at mm-waves the optimum technology was selected in the form of PIN
diodes. This research focused on the quantized reflection phase behavior of active
unit cells and the performance of resulting phase quantized smart reflectarrays. As
an outcome from these investigations a three states phase quantized active unit cell
with three reflection phases for dual polarization behavior controlled by a single DC
bias line was realized at V band. Due to these active unit cells the reflection phase
control mechanism implicitly gets implemented directly at the RF plane of the active
reflectarray at mm-waves which eliminates any explicit expensive, bulky and lossy phase
shifters while simultaneously achieving high performance.
This active unit cell design can reconfigure two similar or different frequencies having
orthogonal linear polarizations. Each active unit cell uses only one DC bias line to
simultaneously control its four PIN diodes connected in a fashion to produce any three
pre-selected (in design) phase states in the range of 0◦ to 360◦ . Having a least number
of DC bias lines greatly simplify the implementation as well as avoids its associated
performance degradations. The problem of polarization conversion and depolarization
of the incident EM field from the reflectarray aperture was resolved through this
research. A novel technique was realized to control and optimize the depolarization
of the incident EM field at active unit cells without any additional complexity. DC
bias line was exploited to optimize the polarization purity. Additionally, these unit
cells share their via holes with neighboring unit cells, therefore less number of via holes
are required in the active reflectarray based on these unit cells. The reflection loss
magnitude was minimized in the active unit cell design to achieve high performance
using PIN diodes. These active phase quantized unit cells were fabricated and measured
at mm-waves. A significant insight of their operation was achieved.
The optimally selected reconfiguration technology for mm-waves (PIN diodes) was
characterized at X and V bands for extraction of RLC values in ON/OFF states
when used in a series switch configuration. Insertion loss of PIN diode switches
and their DC power consumption were investigated in detail based on measurements
to establish a ground for reducing the DC power consumption in high gain smart
antennas. As a result of two mm-waves PIN diode’s measurements at various DC
drives it was found that DC power consumption in a PIN diode based switch can be
considerably reduced at the cost of a slight increase in insertion loss. On the basis
of these measurements, the power consumption of large active smart reflectarrays was
calculated and a functionality/performance versus DC power consumption trade-off
was made available.
Through this research a detailed design and analysis procedure for reflectarrays
was realized. Their design, efficiency and radiation pattern prediction, and
parametric studies to enhance the efficiency were investigated in detail. Performance
characterization of phase quantized reflectarrays, unit cells’ phase states selection
for optimum performance, effects of scattering of phase states around their nominal
pre-selected states were examined. A fast algorithm to build EM simulation model
of reflectarrays was realized. Through analyses and 3D EM simulations in CST,
various performance matrices were prepared for phase qunatized reflectarrays having
2, 3, 4, and 8 states of phase quantization. It was found that in all the cases of
phase quantization, the reflectarrays can point their beam at the desired pointing
angle with sufficient accuracy in comparison to the theoretical continuous phase case.
Similarly, the performance versus complexity trade-off resulted in the selection of 3
states phase quantization as the best case at mm-wave. Due to financial reasons, it was
decided to implement passive demonstrators of phase quantized reflectarrays. Finally,
to demonstrate the scan performance and achieveable gains various phase quantized
reflectarrays’ passive demonstrators were fabricated in a low cost low frequency PCB
fabrication facility. The fabricated set consisted of 12 passive reflectarrays consisting
of two 2 states (0◦ , 55◦ ), two 4 states (0◦ , 55◦ ), two 8 states (0◦ , 55◦ ), and six 3
states (0◦ , 15◦ , 30◦ , 45◦ , 55◦ , 90◦ ). All the reflectarrays except the one; 3 states at
90◦ , were fabricated acceptably. These 11 passive demonstrators were measured in an

anechoic chamber. Based on the measured results it was concluded that these passive
demonstrated behaved very well against the predicted results through EM simulations
and analyses. The performance matrices were validated through these measurements.
Through these exercises on reflectarrays a framework for reflectarray design, fabrication,
their performance estimation and measurements at mm-waves was accomplished to
build a strong ground for future research in this area.
Based on the findings of this research, phase quantized reflectarrays are the compelling
potential candidates for mm-waves high performance smart antennas to offer technically
simple and commercially less expensive solutions for the new era of extremely high
throughput communication systems. The designs conceived through this research
enable very high gain as well as wide angle beam scanning reflectarray antenna platform
at mm-waves. This design when implemented in an active smart reflectarray is capable
of continuous beam steering or switched beams by means of digital controls. It provides
a low DC power solution for high gain wide angle beamsteering smart antennas. The
design can be scaled to any frequency and gain requirements provided the required
switching devices are available. This reflectarray platform when implemented can steer
its orthogonally polarized radiation beams simultaneously at the two similar or different
frequencies. It provides a platform to implement any phase synthesis technique for
radiation pattern control including single pencil beam, multiple pencil beams, contoured
beam, and their scanning. This platform is suitable for multiple antenna configurations
including single center fed or offset fed case, dual Cassegrain or Gregorian, Ring focus
antenna, or folded antennas.
7.2. Future Research Directions and Limitations of this Research 231
7.2 Future Research Directions and Limitations of this

Research
LFR1: Further Measurements: It was experienced through this work that the support
structure plays a significant role in radiation pattern and gain measurements. Due
to time limitations full 3D measurements of all the passive demonstrators were not
possible. A limited set of measurements at fixed roll was performed, which may
not be the actual representative of the maximum gain of the arrays in cases where
main beam is mis-pointed. Therefore, either 3D measurements or first finding
the peak gain and then performing the measurements is recommended which
was not possible in the NPL’s roll/azimuth chamber which is mainly designed
for 3D measurements. The interface with the chamber positioner need to be
identified earlier in research. The required support structure should be made
using Rohacell and alignments should be verified before measurements. Based on
the 3D measurements with the Rohacell structure, higher gains (0.63 dB at least)
are expected than the stated values in this thesis which would result in a better
measured efficiency of the reflectarrays.
LFR2: Unit Cell Analytics: Current analytical approaches presented in this thesis are
limited to simple patch shapes of the unit cell. This is due to the fact that to find
analytical expressions for the radiated power from complex patch shapes is quite
challenging presently. Although, at mm-waves simple patch shapes on a single
layer structure are preferred due to fabrication issues. However, when a design
is scaled to low frequencies and /or multiple layers, other complex shapes can
be used which require new expressions for the radiated power. Once a suitable
expression for the radiated power is found in terms of unit cell parameters, the
analytical procedure presented here can be used to perform parametric analysis.
LFR3: Bandwidth Enhancement: In this research a 5 mil thick substrate was used
while fabricating active unit cells to achieve a larger phase range. Later during
the research it was found that a thinker substrate would have performed better as
was clear from the 10 mil passive unit cells. To demonstrate a better bandwidth
response a thicker substrate is required. The fabrication process plays a significant
role too. Therefore characterization of fabrication process is extremely important.

A one time fabrication may not achieve its goals. Therefore, one need to
do fabrication iterations while updating the design in correlation with what is
achieved in each fabrication step.
LFR4: Circuit Model Extension to Array Level: It is very time consuming to

perform 3D EM simulations of the full reflectarray. In this research a circuit
model of a unit cell was extracted and validated. Now it is required to extend
this circuit model to the reflectarray level. This is required when one needs to
perform optimization to achieve a certain objective from the reflectarray analysis
(e.g. XPD) which is currently not possible due to time considerations. It is to note
here that array based technique do not predict XPD values. As this circuit based
model when extended to array level would consider the real time optimization
w.r.t. unit cell parameters, therefore a faster method could be realized.
LFR5: Extension to Multi-Reflector Cases: The reflectarray analysis presented in

this research was only for a single main reflector case. In practice there are
multiple reflector configurations e.g. dual reflector antennas. Therefore, the
presented analytical procedures for reflectarrays need extension to incorporate
multi-reflector cases. A simple geomtery based ray tracing to find the distances
and then corresponding phases can be used to extend the analytical procedure.
This also includes the ring focus antenna case where blockage due to feed system
can be avoided.
LFR6: Fabrication of Active Reflectarrays: Due to budget limitations this research

ended with the passive demonstrators. Although it is expected based on
experiments of active unit cells and passive demonstrators that the active
reflectarrays shall perform with low loss. However, one needs to fabricate and
measure active reflectarrays to see how they really perform. Similarly, DC power
consumption calculations were based on PIN diodes’ measurements. One needs to
see in a real active reflectarray the DC power consumption. The cross polarization
was optimized in active unit cells. However, one need to see during measurements
of active reflectarrays if further improvement techniques presented in this research
are required to be implemented at reflectarray level. Presently mm-wave MEMS

are not available. However when they become available one could replace PIN
diodes with them to further reduce the DC power consumption.
LFR7: Space Qualification of Active Reflectarrays: The present work did not
explore the space qualification of the hardware. Although a reliable PIN diode
technology was selected however it is not space qualified. To use the design
in a satellite one need to space qualify it and test the reliability of the active
reflectarrays.
LFR8: Design of DC Biasing Control Circuits: To operate the active reflectarray

one needs to build DC bias control circuits based on the technique presented in
this research.
LFR9: Algorithm Development: To make active reflectarrays really smart,

algorithms to find direction of arrival and then to auto point the radiation pattern
towards the required direction are required. The reflection phase calculations and
the required DC bias control to produce the desired phase from a unit cell location
would act as the input of these algorithms. Based on the array approach, and
phase quantization method the generated DC bias matrix need implementation to
achieve a desired performance. When multiple pencil beams or contoured beams
are required, the algorithm need to operate the PIN diodes accordingly to produce
the required phase response from the reflectarray surface. When any other phase
only synthesis techniques are required to be implemented, the algorithm needs to
select the required DC voltages to realize the required phase response from the
reflectarray aperture.
LFR10: Overall Volume Reduction: Although, reflectarrays have a planar structure

of the reflecting surface however, the overall volume due to feed horn becomes
non compact. There is a possibility to eliminate the feed structure spatially
illuminating the surface. It can be made planar as in case of probe fed microstrip
patch antennas. The proposed active surface of reflectarray is capable for this
modification, one just need to choose the required phase distribution based on
calculations of planar surface excitation.
Appendix A
Passive and Active Unit Cells
A.1 Reflection Phase Control in Unit Cells
This section is about the features of various reflection phase control techniques
implemented in unit cells. Numerous techniques have been applied in literature and
can be categorized as passive or active phase control. Passive phase control techniques
benefit from changing the geometrical features of a unit cell while active phase control
techniques make use of control devices to reconfigure the surface currents in a unit cell
which then results in a desired reflection phase. Various types of passive and active
phase controls are discussed as follows:
1. Passive phase control by variable geometrical shapes

(a) Element loaded with variable stub lengths [24]
(b) Variable geometry elements [25]
(i) Single shape elements [107, 123, 124]
(ii) Compound shapes elements
• Double square ring, double cross and rectangle cross elements [125–
127]
• Double concentric ring [128]
• Concentric split ring embedded with a square [129]
• Double petal loops [130]
(c) Phoenix cell structures [21, 121]
(d) Variable length slot & multi-scale cell splitting [131]
(e) Patch physical rotation for circular polarization [109]
2. Phase control by electrical/electronic means
(a) Element electronic rotation for circular polarization [132]
(b) Varactor loaded cells, phase shifters, PIN diodes and MEMS switched stubs
[18]
(c) Liquid crystal cells [133] and elastomers [134] (change of εr )
(d) Reconfigurable Meta-materials
234
A.1. Reflection Phase Control in Unit Cells 235
Figure A.1: Various types and shapes of passive unit cells
A.1.1 Passive Reflection Phase Control
The original reflectarray concept based on short ended waveguides was revived by
developments in printed microstrip technology when the advantages of reflectarrays
were actually realized. To find alternative reflection phase control solutions gained a
significant interest. As a result various passive techniques were developed to produce
the required amount of reflection phase. This section explains the principle of operation
for the majority of passive phase control techniques used in reflectarrays.
A.1.1.1 Stub Loaded Patches It was realized that a microstrip patch can reflect
the incident field. These patches can be loaded with matched stub lengths to produce a
controlled amount of phase shift in the reflected field. A simple patch loaded with stubs
is shown in Fig. A.1 (a). It is one of the very early techniques developed to produce
controlled amount of reflection phase. The stub lengths can be made meandered to
get accommodated in short spaces in large arrays. To achieve a proportionate relation
(2 β l) of phase versus length (l), a stub is required to be well matched to the patch
impedance. Otherwise, the partial reflections from patch itself and the unmatched stub
will result in a disproportionate phase relation. Stubs introduce additional loss due to
impedance matching and losses in printed microstrip lines. The bandwidth of the
stub loaded phase control technique is limited by patch resonance and stub impedance
matching. Similarly, due to variable length stubs, the specular reflection is high for
oblique incidence which further increases the loss. One of the main disadvantages of
this technique is deteriorated cross polarization due to increased radiation leakage due
to stubs. A solution to improve cross polarization was proposed by [148] which orient
unit cells in mirroring fashion in four quadrants of a reflectarray.
A.1.1.2 Patches with Variable Geometric Features Metallic patches behave

like resonant circuits and their scattering impedance is a function of physical
dimensions. By changing physical dimensions, the reflected phase of an incident

field/wave can be controlled. A whole range of geometrical shapes is available as shown
in Fig. A.1 (b). The selection of a particular shape is mainly based on polarization
achievable bandwidth and ease of manufacture at operational frequency. Some shapes
like dipoles are very sensitive and very narrow band while others a more tolerant to
manufacturing tolerances. For wide band operation a smooth phase variations with
respect to geometric changes are required. Varying the patch geometry can achieve a
phase range of 0◦ − 180◦ without a ground plane. The reflected phase range, being
a function of separation between patch and its ground plane, can be extended up to
nearly 360◦ by backing the patch by a ground plane using very thin substrate (thickness
≤ λ/10). Due to high quality factor (Q) value of these patches, the phase variations
versus length are strongly nonlinear, exhibiting very sharp variations near resonance
and a slow change for off resonance [107]. Such rapid variations in reflected phase
make the phase distribution very sensitive to manufacturing tolerances. Moreover, due
to nonlinear behavior the reflected phase becomes very sensitive to frequency variations
which results in a narrow band performance of this technique. This non-linearity can
be reduced by using thick substrates. However, the reflected phase range reduces even
further. It is worth mentioning here that for a reflectarray design, all phase values
between 0◦ − 360◦ are normally required, which is not achievable by just using a thick
substrate. If no grating lobes or surface waves are generated, then the amplitude of
reflected field/wave is nearly equal to unity expect small dissipative losses in dielectric
separators and metal patches. However, the patch size may change significantly up
to 40% to achieve phase variations of nearly 360◦ . Such off-resonant operation of
patch increases specular reflections for oblique incident angles which reduces the overall
efficiency. Due to absence of extra stubs etc. the variable patch elements, can be easily
accommodated, and result in a superior cross polarization performance as compared
to the variable stub technique because there are no leakage radiations [25]. Using
trapezoidal, parallelogram transforms, and certain notches on the patch surface, the
cross polar performance can be further improved.
A.1.1.3 Compound Variable Geometry Patches The research for extended

phase range and smoother phase variations versus dimensional changes over a wide
frequency band using only single layer patches had led to the design of various
compound shapes e.g. double square or circular rings, double cross elements, and a
combination of cross & rectangle cells which benefit from multiple resonances of the
structure as shown in Fig. A.1 (c). Double ring structures exhibit two resonances as
compared to single resonance of a single ring, and dimensions of outer and/or inner
ring(s) can be changed as per requirements to obtain the required phase and frequency
response. Double square ring elements have shown a 1 dB bandwidth of 9% for the
reflectarray at 22 GHz using single layer cells. Similarly, an investigation of double cross
loops has shown a comparable bandwidth. An extended phase response in the range of
600◦ on different substrate thickness values of the substrate has been achieved [125,126].
In spite of extended phase range in both cases, the phase versus length curves tends to
be non-linear owing to resonances therefore, limiting the bandwidth.
Inspired by the fact that increasing the gap between loop elements can increase the
separation between their resonant frequencies, the linearity of phase response curves can
be improved to a certain level. After trying multiple elements, an optimum combination

consisting of a cross and rectangles was found as shown in top right corner in Fig. A.1 (c)
[127]. This compound element acts as a multi-resonant structure and phase variations
were achieved by varying the size (length) of loops. Moreover, the separation between
loops is adjusted to improve the linearity of phase response curves while keeping gaps
among loops and both arms of cross loop symmetric. For linearly polarized excitation,
an extended (500◦ ) and smoother phase response has been reported which led to a 1
dB percentage bandwidth of 24% in a Ku band reflectarray.
When sub-wavelength elements (size = λ/n where n > 2) are used as reflectarray
unit cells, wider bandwidths are achievable due to non-resonant behavior of these
elements. A study of sub-wavelength double concentric rings (Fig. A.1 (c)) using λ/4
inter-element spacing has demonstrated 2 dB bandwidth of about 33% at X band for
a single layer reflectarray design [128]. Although, the percentage bandwidth improves
using sub-wavelength unit cells, the reflectarray aperture efficiency severely degrades.
A value of 36.2% has been reported due to a limited achievable phase range versus
geometric changes (< 300◦ ). It has been observed that with an increase in n, the
phase range reduces therefore, reducing the overall reflectarray efficiency. Among
others concentric split ring embedded with a square unit cells [129] have demonstrated
a considerable bandwidth increase (67%).
An issue related to variable shape patch unit cells is the amount of scaling of selected
shape to achieve a desired phase range. It has been observed that the patch undergo
significant geometry scaling while providing phase from 0◦ − 360◦ . The size of patch
to produce 0◦ phase is significantly different from the patch providing a 360◦ phase.
Therefore, when a transition in phase from 360◦ to 0◦ occurs over the reflectarray
aperture, the geometry of patches undergo a significant transition. Due to these
geometric transitions the illuminating field (from a feed (elementary pattern)) gets
modulated at the reflectarray surface in the form of circular lines corresponding to these
transitions where electric field vanishes [21] as shown in Fig. A.2. It has been reported
that in large contour beam reflectarrays (which are very sensitive to random phase
errors) this transition deteriorates the radiated beam. Therefore, a sudden transition
in geometry of reflecting patches is not desired and should be avoided particularly in
contoured beam reflectarrays. Otherwise, the aperture efficiency gets lowered and the
antenna coverage region gets significantly changed. To avoid this problem phoenix cell
and multi-scale cell splitting concepts were evolved.
A.1.1.4 Phoenix Reborn Cells A phoenix cell [21] has the advantage of returning
to its original shape after going through a 360◦ reflection phase cycle (so called rebirth
of a phoenix bird). Therefore, avoiding sudden transitions in geometry. It is based
on square ring slot approach. An inner square ring starts growing from inner patch
towards outer ring in k intermediate step as shown in Fig. A.1 (f). It divides the one
big slot into two, and as a result reflected phase variations take place. The number of
intermediate steps can be controlled to achieve the required phase resolution. These
two slots have different resonant frequencies and when the square ring generating these
slots grows, both resonant frequencies shift downwards. The higher resonance shifts
downward and replaces the lower resonance. Whereas the lower resonance has already
(a) (b) (c)
Figure A.2: (a) A reflectarray made up of double cross elements, (b) Simulated incident
field, (c) Near field measured over the reflectarray aperture: (color indicates the field
strength with red being the maximum), at places where phase transition from 360◦
to 0◦ happens field can be seen vanishing in (c) due to unit cell sudden geometric
transitions.
shifted to a new lower resonance. This replacement of resonances by changing the slot
dimensions in a systematic way is the key principle of phoenix cells.
While undergoing k growth steps, the widths of both slots keep on inverting i.e. one
grows and other shrinks. The phoenix cell principle permits to reverse the dispersion
characteristics of both resonances, and as a result estimated percentage bandwidth
of 18% was achieved with dispersion ≤ 30◦ /GHz in simulations using a single layer
structure. The 3 dB bandwidth at C band was reported to be 27.5% [201]. In a
modified design by shrinking the cell size from its original, a 3 dB increase in gain and
7.4% bandwidth at X band was achieved practically [202]. Therefore, phoenix cells
are a fairly good choice to increase bandwidth of reflectarray in single layer designs.
Similarly, it avoids sudden changes in geometry, therefore is a preferred choice for
contoured beam reflectarrays.
In another approach the phoenix cells operate with complementary shunt and series
LC resonances [21]. A 37 element library of patches was selected to produce required
reflection phases. Each element in this library provided a 10◦ successive phase shift
step. The library starts with a fully metallic element providing a 180◦ phase shift.
Then a square aperture of growing size is introduced. Further, this aperture is filled
by a growing central patch which end up in a fully metallic patch as shown in lower
portion of Fig. A.1 (f). This cell design is focused on low loss and wideband operation.
The 360◦ phase shift range is divided and is covered by two complementary resonances.
Elements 1 - 10 roughly behave inductively and cover a phase range of 180◦ −80◦ while
operating with a shunt resonance. While capacitive series resonance for elements 11
- 36 cover a phase range of 80◦ − 180◦ . Because both resonances are operating in a
limited phase range, therefore can be softened to reduce losses.
A.1.1.5 Multi-Scale Cell Splitting: Variable Length Slots Variable length

slot cell splitting is another concept to avoid abrupt variations in reflect array elements
due to required reflection phase transitions. An initial patch is loaded by a slot of
variable length. When slot length equals patch width, it divides the patch into two
smaller patches. This process can be further repeated on these smaller patches and so
on as shown in Fig. A.1 (e). Each splitting provides a phase range of around 360◦ .
Therefore, a very large range of phases can be accommodated provided the splitted
pieces of cell remain manufacturable. However, the local behavior of reflection phase
at each splitting is sharp which makes it narrow band. For a pencil beam reflectarray,
the 3 dB bandwidth of around 10% at Ku (12 GHz) band was achieved in a single layer
design. The reported cross polarization level is very good (< 30 dB), but the sidelobes
are relatively high (around 16 dB below the peak) [131].
A.1.1.6 Element Rotation for Circular Polarization Element rotation is one

of the early techniques by which the reflected phase of a wave can be controlled by
rotating the element itself as shown in Fig. A.1 (d). The amount of reflection phase
is twice the rotation angle. This technique can achieve better cross polarization and
sidelobe performance due to averaging effect of a large number of elements arranged
pseudo-randomly. Specular reflections are low for oblique incidence because all elements
resonate at the same frequency [25]. This technique is only applicable to circular
polarized waves as the polarization also gets rotated. The bandwidth is limited by
patch itself and no significant increase has been observed.
A.1.1.7 Comparison of Passive Phase Control Techniques Table A.1 provides

comparison for common passive phase control techniques. Although, this table is very
comprehensive however, certain parameters are dependent on the selected shape(s)
therefore, few entries may seem to be off range.
Table A.1: A comparison of passive phase control techniques

stubs shapes shapes
|M | 0.78 0.9 0.85 0.9 0.85 0.8
∆Φ ≥ 360◦ ≈ 340◦ ≥ 600 ◦ 360◦ ≈ 360◦ ≥ 600◦
Φ∼F medium high 6-13 medium high low high
rad/GHz
Φ∼ 0.08 0.38 0.05 high low high
∆L /µm ◦
Φ∼ high high high high medium high

incident
angle
◦ /◦
Geomtery medium high medium low low low

Transition
Bandwidth low low medium low medium low 10%
25% 27%
Continued on next page ...
Continuation of Table 2.5

stubs shapes shapes
Co-pol any any any only any mainly
circular linear
XPD poor excellent excellent good good good
mm-waves yes yes no yes yes/no no
fabrication
Comment stub thick & multiple only difficult difficult
matching, multi resonances, circular mm-waves mm-waves
lossy, layering smoother polarized fabrication fabrication
Φ ∼ ∆L smooths Φ ∼ ∆L,
non-linear resonances difficult
Φ ∼ ∆L mm-waves
non-linear fabrication
A.1.2 Phase Control by Electrical/Electronic Means
This section describes the reconfigurable properties of an antenna and techniques to

facilitate it. Potential advantages and drawbacks of each technique are also discussed
here. The electromagnetic field radiated by an antenna is determined by its surface
current distribution. By intentionally altering this current distribution, antenna
characteristics can be changed. Reconfiguration is the deliberate change in operating
frequency, polarization, and/or radiation characteristics of an antenna by redistribution
of its surface currents. These reconfigurable antennas are becoming important in
modern communication systems by providing additional degrees of freedom for system
optimization, not available in fixed antennas. Due to the fact that the antenna
characteristics are not independent of each other, as a consequence changing one
property provokes changes in others to some extent. Therefore, while designing a
reconfigurable antenna following aspects are given due consideration [120].
1. Which properties of antenna are required to be made reconfigurable (frequency,

polarization, radiation pattern, or a combination of these)?
2. How to reconfigure antenna elements to achieve a desired reconfiguration

property?
3. Which reconfiguration technique minimally impacts the other characteristics

which are not being reconfigured in a particular reconfiguration?
4. Availability of selected technique at the frequency of operation?
Table A.2 lists reconfigurable properties for an antenna.

1
|M | = reflection magnitude, ∆Φ = achievable phase range, Φ ∼ F = phase versus frequency
dispersion (rad/GHz), Φ ∼ ∆L = phase change per unit change in dimensions (length, radius etc.)
Table A.2: Reconfigurable properties of an antenna

Property Possibilities
Continuous scan in a band
Frequency Band switching
Multi-band operation
Linear: Horizontal, Vertical, Slant
Polarization
Circular: Left hand CP, Right hand CP
Continuous scan
Fixed bean switching
Radiation Pattern Beamwidth control
Contour beam(s)
Single or multiple pencil beams
Combination of above A suitable combination of above
While the reconfigurable antennas enable multitude of functionality and automation to

fairly adapt to the operating requirements and environment, these capabilities comes
at the expense of following design issues.
• Complexity associated with biasing network to control the switches. It is

particularly problematic in arrays, where a large number of wires are required
to be routed, and result in performance degradation.
• Isolation of biasing lines with radio frequencies signal lines.
• Requirement of rapid tuning to ensure the system functionality.
• Increased power consumption due to integration of a huge number of active

elements. This leads to heat dissipation problems.
• Operation at high RF power may be problematic for devices.
• Generation of harmonics and inter-modulation products.
• Capacitance modulation for Varactors.
• Increased system cost and complexity.
Antenna reconfiguration can be invoked by either of the following technologies as shown

in Fig. A.3.
1. Electrical: By switching and tuning techniques e.g. RF-MEMS, PIN diode

switches, and Varactor diodes
2. Optical: By photoconductive switches
(◦ /µm) , Φ ∼ incident angle ◦ /◦ = change in phase versus wave incident angle , Co-pol = achievable
co-polarization , XPD = achievable level of cross polarization discrimination.
Antenna Reconfiguration
Figure A.3: Antenna reconfiguration technologies and their ingredients
3. Smart materials: Ferrites, liquid crystals, and elastomers (by change of εr ),

reconfigurable Meta-materials
4. Physical structural changes: By means of electric motors and re-orienting the

constituting elements of an antenna
A.1.2.1 Electronic Reconfiguration Electronic reconfiguration is the most

commonly found technique in antennas. Electronically reconfigurable antennas utilize
electronic switches e.g. RF-MEMS or PIN diodes to redistribute the antenna surface
currents in discrete steps. For a continuous change in surface currents the varactor
diodes are used. Ease of integration in an antenna structure is their main attractive
feature despite certain drawbacks associated with this technique. MEMS technology
has the advantage of good DC-RF isolation and low loss. However, MEMS switching
response is slow as compared to PIN and varactor diodes which have the response
time in nanoseconds. MEMS can be fabricated using a monolithic technique with the
antenna instead of soldering discrete components in antenna post manufacturing. In
addition to just adding these discrete components, a phase shifter can be implemented
with reflectarray elements as well. Application of varactors, PIN and MEMS to
reflectarray elements is shown in the Fig. A.4.
Although, these electronic components are very attractive, they suffer certain downsides
too. The resistive and capacitive non-linearities are exhibited in addition to high
operating voltage (MEMS + Varactor). DC bias lines required by these switches
deteriorate radiation pattern and may add additional losses. The requirement for
bypass capacitors and biasing inductors further complicate the antenna structure. A
major challenge in achieving continuous phase variations is the linearity of tuning
technology. In addition to losses, particularly at high operating RF powers these
switches may produce harmonics and intermodulation distortions. In a in satellite
operation, there exist strict limits for allowable adjacent channel interference and
harmonic levels. Varactor diodes based reconfiguration is susceptible to nonlinear
effects. Varactors diodes integrated in an aperture illuminated by moderate power
levels have shown to create a significant amount of odd-order harmonics [18, 203].
When a varactor diode is subjected to a large incident field, its capacitance modulation
occurs at the carrier frequency, leading to significant amount of intermodulation
distortion (IMD). It causes phase modulation which manifests itself as a distortion
(a) (b) (c)
Figure A.4: Electronic reconfiguration technologies: (a) Varactor diode loaded cell, (b)
Phase shifter loaded unit cell, (c) PIN/RF-MEMS switch loaded cell.
in the scattered signal [18]. In varactor diode based unit cell measurements [203] using
two tone test for the third order IMD as a function of incident power and tuning
voltage have shown the generation of IM3. IM3 was found to be a strong function
of bias voltage. The worst case was recorded when cell is biased at midpoint of the
phase transition. The reflected signal remains spur-free up to an input of -5 dBm,
below which the reflectarray is essentially linear exhibiting a spur-free dynamic range
of 40 dB. Therefore, varactor based reflectarrays are usually limited to receive only
apertures and are not very good choice for transmit application. In a reflectarray the
power is spatially distributed over a greater number of elements. Therefore, in a large
reflectarray operating with a limited transmit power, this may not be a problem in cases
where the individual diodes receive lower incident power. Using MEMS capacitors can
be an alternative however, the capacitance range is limited [203,204]. Currently, MEMS
technology is not considered mature enough to be used widely in space applications due
to their reliability constraints.
Reflectarrays based on semiconductor diodes suffer a significant reduction in reflection
efficiency due to diode losses, nonlinear effects caused by RF modulation of the varactor
capacitance, and significant package parasitics at higher frequencies. The PIN and
Varactor diodes are significantly lossy. PIN diodes suffer significant losses (5.6dB)
when used in reconfigurable unit cells [142, 143]. Similarly, varactor diodes based unit
cells suffer from considerable amplitude fluctuations about the center of tuning range
mainly due to power loss in diode which is more pronounced at unit cell resonance
frequency. Reflection losses of 3.5 dB at 5.8 GHz [203] and 4 dB at 8 GHz [138, 205],
have been reported. This unit cell loss is projected at the reflectarray level almost
by the same amount. Such a huge loss drastically lowers the reflection efficiency of
a reflectarray and is manifested as lower ultimate gain. Such increased losses present
a challenge to reconfigurable reflectarray to compete the non-reconfigurable counter
parts.
Silicon switches Optical fiber fixture

(a) (b) (c)
Figure A.5: Optical reconfiguration technologies: (a) Optical switching through

non-integral fiber optic, (b) Optical switching through integral fiber optic, (c) Directly
integrated LASER without a fiber optic.
A.1.2.2 Optical Reconfiguration Optically reconfigurable antennas rely on

optical switches. These optical switches work on the principle of exciting valance
electrons to conduction band in a semiconductor (Si, GaAs) by an incident laser light
to create a conductive path. Although, these optical switches are lossy in nature and
need LASER light to function, they have potential advantages of being linear and free
from harmonics and intermodulation distortions. In addition, no bias lines are required
and therefore, they eliminate any degradation of pattern and unwanted interferences.
The main issue with these switches is the activation of semiconductor i.e. how to
illuminate these by the LASER beam. For activation the LASER diode is coupled
to a fiber optic cable which may or may not be an integral part of the antenna. In
the non-integral case (Fig. A.5 (a)), these cables are held in place (on semiconductor)
using plastic clamps [206]. For the integral case (Fig. A.5 (b)), these cables are held
underneath the substrate by plastic fixtures, and holes are drilled in substrate to couple
the LASER light to the semiconductor [207]. Finally, the laser diodes can be directly
integrated into the antenna structure (Fig. A.5 (c)) thus eliminating the need of fiber
optic cables. In this case holes are drilled in the substrate to let the illuminating light to
reach the semiconductor switches. The LASER diode is activated by a current driver to
generate the required optical power [208]. These optical switching topologies are shown
in Figure 4 7. Despite of being linear and avoiding bias lines, these optical switches are
lossy and need complex activation mechanism. Additionally, there is significant power
consumption per switch activation mechanism. Although their applications are found
in low frequency single antennas however, no example has been found in literature
where such optical switches have been used in reflectarrays. For their integration in
reflectarrays they have to be low loss, extremely compact, and power efficient; neither
of these is true presently.
A.1.2.3 Reconfiguration Using Smart Materials Smart material e.g. ferrites

and liquid crystal (LC) are able to change their electrical characteristics. Ferrites can
change their permeability (µr ) in response to applied static magnetic field, and their
permittivity (εr ) in response to applied static electric field. Similarly, liquid crystal can
change their permittivity (εr ) in response to applied voltage by reorienting its molecules
as shown in Fig. A.6 [133]. Such change of electrical parameters enables a designer to
exploit these characteristics for reconfigurability.
Vbias < Vth
Vbias > Vth Vth Vmax

(a) (b)
Figure A.6: Operation of liquid crystals: (a) Liquid crystal dipole/molecules allignment
w.r.t. bias voltage , (b) resulting change in (εr ) and loss tangent (tan δ).
An LC cross section is shown in Fig. A.6 (a). There are two substrates acting as
microstrip line (top) and ground (bottom). The gap between these two substrates is
filled with LC which acts as a dielectric. A molecule of LC is ellipsoidal in shape. The
orientation of these molecules defines the electrical properties of an LC. Substartes
can be coated (e.g. with polyimide) to facilitate the initial orientation of these
molecules, parallel to electrode here. When electric field direction is perpendicular
to the semi-major axis of ellipse, the dielectric properties (εr,⊥ and tanδ⊥ ) are effective.
Molecules can keep their orientation for applied voltage less than threshold voltage
(Bbias < Vth ). When applied voltage exceed certain voltage (Bbias > Vmax ) then the
molecules get aligned perpendicular to the electrode in the direction of applied voltage
as shown in lower part of Fig. A.6 (a). As the electric field is still in the previously fixed
direction, therefore, the dielectric properties parallel to major axis of these molecules
(εr,k and tanδk ) become effective. When applied voltages are removed, the molecules
return to their original orientation. By varying the applied voltage from Vth to Vmax
a continuous tuning range is achieved. Normally the operating voltages are between
0 - 20 V. Although, LC have been reported at lower frequencies mm-wave bands in
antennas, they are very lossy [133, 209, 210]. They are more useful at E band (77
GHz) and above. The main disadvantage of these materials is their low efficiency and
common availability. Additionally, the reconfiguration time is large as compared to
PIN and MEMS based switches.
Similarly, ferrite materials are not a preferred choice for reflectarrays due to their
bulkiness, lossy behavior and magnetic control. Currently ferrite materials are use
in individual antennas however not common in large arrays. Graphene and carbon
nano-tubes may be a promising candidates for future, however not suitable presently.
Advantages of Graphenes have been realized at THz spectrum however, they are very
lossy at lower mm-wave bands. Additionally, the technology is still in developing phase.
Elastomer materials need very high actuation volatge in the order of KVs, theirfore not
a preferred choice for space applications.
A.1.2.4 Physical Structural Reconfiguration The structure of antenna is

changed by physical means e.g. by actuating motors as proposed for reflectarrays
A.2. Reconfigurable Unit Cells 246
(a) (b) (c)
Figure A.7: (a) A split ring unit cell, (b) Rotation of split ring by an angle ψ, (c)
MEMS based dual band implementation of rotation based technique.
by [24]. In large reflector antennas e.g. in FAST observatory (Fig. 1.1 (b)) it is
achieved by physically tilting the antenna panels at certain angles by motors. One
structural reconfiguration solution for reflectarrays is presented in [109]. An element
can be physically rotated by electric micro-motors as suggested in the first illustration
of this principle [25, 109]. However, element rotation is only applicable to circular
polarization due to the fact, the element rotation is coupled with electrical polarization
which gets rotated too. It is a very cumbersome solution in large arrays where a large
number of motors need to be integrated and powered. In advanced forms, the electronic
switches can be integrated in the unit cell which when activated result in rotation and
introduce the phase shift in the reflected field as shown in Fig. A.7. A dual band
implementation is demonstrated by [132] on a 4 inch quartz wafer consisting of 109 K
band and 124 Ka band elements. In this implementation each unit cell has integrates
six equally spaced series RF MEMS to provide sequential rotation. A progressive phase
shift of ±120◦ in both bands was achieved. Beam steering of ±35◦ in K band ±24◦ in
Ka band was accomplished.
A.1.2.5 Comparison of Reconfiguration Technologies When it comes to

choose a reconfiguration technology, there is no single answer to which one is the
best. The best is the one which is most suitable to constraints of the application
in consideration. One needs to consider the technology maturity, performance (loss),
availability, power handling, switching time, device linearity, and application to certain
frequency bands and the operating environment. Other design complexities e.g. ease of
integration and biasing control are given due consideration. Table A.3 compares major
reconfiguration technologies.
A.2 Reconfigurable Unit Cells
In recent years, there have been significant developments in the reconfiguration

technologies and their application. Although, there exist very exotic solutions (e.g.
graphene and photo conductive), however a majority of reconfiguration is based on
semiconductor devices (PIN and Varactor diodes) and MEMS. Due to reasons that
RF-MEMS are not commonly available as commercial off the shelf (COTS) and their
Table A.3: A comparison of reconfiguration technologies 2
Property MEMS PIN Optic Varactor Ferrite LC Graphene

Freq. GHz <40 <110 <20 >70 >1000
Voltage V 20-100 3-5 1.8-1.9 0-30 >100 - <35
kV
Current 0 3-20 0-87 100
mA
PDC mW 0.05-0.1 5-100 0-50 0-300 Low Low
Switching 1-200 1-100 3-9 uS nS mS S
speed uS nS
Isolation Very High High Medium
High
Loss dB 0.05-0.2 0.3-1.2 0.5-1.5 High 0 GHz - GHz, - GHz,
- THz + THz + THz
Linearity + 0 - 0 -
Analog / D D D A A A A
Digital
Reliability 0 + + + 0 0 -
Integration + - - + 0 +
Biasing
Cost Medium Low High High Low 0
Available Foundry COTS COTS Special Special
Design facility facility
mm-waves - + - - - - -
application
space heritage is not well established yet, therefore semiconductor devices are presently
the most suitable choice for reconfiguration designs. In this section an overview of unit
cell reconfiguration based on PIN, Varactor diodes and MEMS is presented.
A.2.1 Single Versus Dual Polarized Unit Cells
Polarization of incident field(s) plays a significant role in deciding the reconfiguration

design of a unit cell. To have multiple polarizations in an antenna is a potential
advantage. In satellite communications it is preferred to share the same antenna
aperture for orthogonal receive and transmit polarizations. It efficiently utilizes the
available real estate, saves cost and mass.
2
+ = good, 0 = neutral, - = poor, A= analog, D = digital, COTS = commercial off the shelf
2
d d
(a) (b)
Figure A.8: Inter-cell spacing for (a) dual polarized unit cells and (b) single polarized
unit cells. Dual polarized unit cells offer a compact solution
√ as the inter-cell spacing is
d as compared to single polarized unit cells where it is 2 d.
Additionally, the incident field of each polarization can be reconfigured independently.

Therefore, using the same aperture of an antenna a satellite can have same or different
receive and transmit coverage regions supported by two orthogonal polarizations.
To enable independent scanning/steering of dual polarized beams, a unit cell need
to integrate independent control for each polarization. A unit cell can have dual
polarization or two interlaced single polarized unit cells can be used to achieve dual
polarization operation. There are certain advantages of using a dual polarized unit cell
instead of two interlaced single polarized unit cells. It reduces the spacing between
elements which is more stringent in reconfigurable reflectarrays as compared to fixed
broadside arrays to avoid grating lobes. A single polarized unit cell lattice is inefficient
because of increased specular reflections since only a portion of its area is used to control
the reflection phase of each linear polarization. The size (‘d’) of a dual polarized unit
cell is mainly comparable with that of the single polarized counterpart as shown in
Fig. A.8 [138]. Both solutions are √ square lattice. Inter-element spacing in a single
polarized unit cell then becomes 2 d whereas in dual polarized unit cell it stays as
‘d’. Therefore, when dual polarized unit cells are used, the reflectarray lattice becomes
more compact and the array aperture is used more efficiently.
A.2.2 A Dual Polarized Tuneable Unit Cell
A dual polarized reconfigurable unit cell is shown in Fig. A.9 (a) [138, 205] that can
dynamically and independently control the reflection phase of two orthogonal linear
polarizations. This kind of unit cells enable dual polarized reconfigurable reflect arrays,
linear/circular polarization flexibility, and polarization twisting. It uses a pair of
varactor diodes controlled simultaneously for each polarization component. Consider a
case where the incident TM and TE polarized waves are assumed along x (horizontal)
and y (vertical) axis respectively. The reflection phase of TM wave is controlled by
diodes marked V2 only because rest of the diode pair marked is located in zeros of the
current. In case of TE waves the diodes marked V1 control the reflection phase while
diode pair marked has a minimal effect. The biasing of diodes has a special arrangement
as shown.
For a TM wave, the reflection phase is controlled only by V2 varactors. In the absence
of PIN diodes when the value of V2 is varied from 0 to 20 V a phase range of 300◦
was achieved using Arlon25 substrate with a Rohacell layer at X band. The reflection
phase sensitivity to variations in voltage V1 (which is for the other pair of varactor
diodes for TE wave) is only 15◦ when V1 varies from 0 to 20 V. It indicates that
both polarizations can be controlled independently without having much effects on
each other. Variations due to wave incident angle were recorded as 20◦ for an incident
angle of 40◦ as compared to the case of normal incidence. Maximum reflection loss
was observed to be 3.8 dB whereas the cross polarization discrimination is very high.
To extend the reflection phase range PIN diodes can be integrated in the unit cell
with Varactor diodes [138]. The PIN and Varactor diodes are DC decoupled by very
thin etched line to bias them for independent control. In this design switching the
PIN diodes ON/OFF, extends the phase range to 360◦ , while slightly increasing the
reflection loss by 0.2 dB in comparison to the design without PIN diodes.
It is a versatile design for low frequencies where unit cell size permits splitting and then
integrating the electronic components. Due to large dimensions of a low frequency unit
cell, the DC bias control lines for multiple components are relatively much easier to get
accommodated in comparison to mm-wave designs. One would be tempted to have such
versatile designs at mm-waves. However, presently it is almost impossible to integrate
multiple electronic components of the comparable size to a mm-wave unit cell.
Varactor R L
V switch 2 PIN
PIN L
V1+V2
V2
V Varactor
V1 V1
varactor
V2
PIN
V switch 1
V2
(a) (b)
1
Sw
h
itc
itc
h
Sw
2
YX
MEMS MEMS
(c) (d) (e)
Figure A.9: Various reconfigurable unit cells (a) Dual polarized unit cell reconfigured by
a combination of varactor and PIN diodes, (b) Single polarized frequency reconfigurable
unit cell using PIN switches and varactor tuning, (c) Reconfiguration using RF-MEMS,
(d) Polarization reconfiguration using PIN diode switches, (e) MEMS based circular
polarized unit cell based on rotation principle.
A.2.3 A Single Polarized Frequency Selectable Tuneable Unit Cell
Generally, antennas are desired to be broadband. However, instantaneous operational

bandwidth required to make the communication link possible is usually significantly
narrower than the full bandwidth covered by an antenna. Therefore, there is not much
harm, except the reconfiguration and complexity, in making the antenna bandwidth
instantaneously selectable/tuneable. Doing so enables instantaneous narrow band
communication in a certain wider bandwidth range using the same single antenna. In
addition to cover a desired frequency range when reconfiguration of radiation pattern is
desired; the antenna solution becomes further complicated. Such solution may involve
switching and tuning together. A frequency selectable and reflection phase tuneable
unit cell design is shown in Fig. A.9 (b) [117, 211]. This solution embodies two PIN
diodes and one varactor diode to configure and tune the unit cell. This unit cell
structure can switch the state of PIN diodes and tune varactor diode independently.
The operation is divided into two parts, namely the course frequency switching by
means of PIN diodes and fine frequency tuning by a varactor diode. Lengths of various
sections are designed such that a continuous frequency range is covered. To improve
the slope of phase frequency curves, rectangular structures can be changed to bowtie
structures. A phase range of 270◦ was achieved. A reflection loss of 4 dB for normal
incidence and 5 dB for oblique incidence was reported. Biasing arrangement consists
of three independently controlled serially connected voltage supplies. PIN diodes were
operated at 20 mA current each by adjusting voltages (V switch 1, 2) and resistors
(R). Whereas the voltage for varactor diode (V varactor) was varied from 2 to 21 volts.
RF chokes are realized by inductors. The biasing tracks are made perpendicular to
the operating polarization (y in this case). The overall structure is capable to cover a
percentage frequency bandwidth of 50% for a single linear polarization operation at S
band.
It is a highly versatile design to cover a significant amount of percentage bandwidth
by selectable sub-bands and then tuning the reflection phase at the selected band to
make reconfiguration possible. At low frequencies like S band, the idea works very well
where components are much smaller than the unit cell itself and a number of those can
be integrated by cell splitting which seems to be not the case for mm-waves.
A.2.4 Monolithic MEMS Based Single Polarized Unit Cell
Instead of mounting discrete control devices/switches in post fabrication, a monolithic

process provides the flexibility to manufacture MEMS with the unit cells. Therefore,
parasitic effects are reduced and reliability gets improved. A unit cell based on
monolithic MEMS technology for single linear polarization using five pair of MEMS
symmetrically distributed around E-plane (vertical) is shown in Fig. A.9 (c) [212, 213].
The monolithic process deposited a 500 nm thick gold layer and a 1500 nm thick
aluminum layer on quartz substrates. Aditionally, high resistivity poly-silicon bias lines
were patterned by the process as well, which do not impact the microwave performance.
To control the capacitance, MEMS membrane is actuated by connecting it to a DC
source. When DC voltage is applied to the membrane, all the metal layers except
membrane are connected to a common ground thus exerting an attraction force on the
membrane. By this way these MEMS act as a two state digital capacitor providing ON
and OFF state capacitances. There are 32 switching states due to 5 pairs of switches
(each pair activated simultaneously). A 360◦ phase range with low reflection loss has
been achieved at 12 GHz. An average phase error of 15◦ for a 10% bandwidth has
been claimed with reflection loss better than 1 dB. This low value of phase error is
achieved by careful cell design to avoid operating MEMS close to sharp resonances
of the element. Reflection phase deviation for oblique incidence is not significantly
deteriorated and full 360◦ phase range is maintained under even very large angle of
incidence (60◦ ). Owing to symmetry of the cell design, good cross polarization level is
maintained.
The issue with the achieved phase range is its non-linearity. Additionally, a significant
number of DC lines are required as there are 5 pairs of MEMS per each unit cell.
Although, this technique offers a low current solution however, the size of each MEMS
is very comparable to the reflecting patch in a mm-wave unit cell which is a nightmare
when such a technique is implemented at mm-waves.
A.2.5 A Polarization Flexible Unit Cell
The response of a unit cell in a periodic environment can be represented by a 2 port

network, where ports are associated with TE and TM polarized waves. For a normal
incidence assuming fields aligned with coordinate axis, the S parameters S11 (reflection
in x polarization) and S21 (reflection in y polarization) are defined as:
Exr Eyr

S11 = i and S21 = i (A.1)
E i
x Ey =0 E i
x Ey =0
where i = incident, r= reflected, E = electric field, x = horizontal polarized field, and y

= vertical polarized field. When S11 = 1 and S21 = 0, the polarization of the reflected
field is preserved i.e. reflected and incident fields have the same polarization. In the
case when S11 = 0 and S21 = 1 the polarization of the reflected field gets rotated by
90◦ . Therefore, if the incident field was horizontally polarized then the reflected field
would have vertical polarization. This is very useful function where polarization control
is required. In addition to polarization control, a two state phase control can also be
implemented in a unit cell. Such a unit cell capable of polarization control and two
state phase control is shown in Fig. A.9 (d) at 11 GHz [214]. It consists of two switches
(Switch 1 and 2). By operating these switches in opposite states (01 or 10) performs
the action of polarization twist by 90◦ while also providing a phase shift of either 0◦ or
180◦ . When both switches are operated in the same state (00, 11) the unit cell reflects
the incoming field with the same polarization while capable of providing either 0◦ or
180◦ phase control also. For an oblique incidence in a limited range of ±30◦ the unit
cell performance is acceptable with a phase deviation of around 20◦ .
The operation of unit cell is Fig. A.9 (e) is similar to the unit cells used for circular
polarization by means of electronic rotation as described for Fig. A.7 where three
pairs of RF MEMS were used to provide a phase resolution of 120◦ at each band (27
and 37 GHz). A beam scan range of ±35◦ and ±24◦ was demonstrated for operating
frequency bands respectively while achieving grating lobes of 10 dB down the peak. The
RF MEMS used in this research were purpose built and are not available commercially.
Appendix B
Unit Cell Reflection Phase and

DC Biasing
Here the reflection phase required in a reflectarray of 35 × 35 unit cells is shown. The
focal length F = 70 mm and side length of the square reflectarray L = W = D =
94.5 mm are used for these calculations. The continuous reflection phase required from
each unit cell’s location is shown in Fig. B.1. Then the reflection phase of each unit
cell is quantized to discrete values using a 3 states phase quantization which resulted
in three phases. Here 0◦ , 120◦ , 240◦ , phases are chosen as the required phase states.
The 3 states quantized phases in the reflectarray are shown in Fig. B.2. Finally, the
DC voltages required to produce these discrete reflection phases over the reflectarray
aperture are shown in Fig. B.3.
252
253
UC # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
UC # mm -45.9 -43.2 -40.5 -37.8 -35.1 -32.4 -29.7 -27 -24.3 -21.6 -18.9 -16.2 -13.5 -10.8 -8.1 -5.4 -2.7 0 2.7 5.4 8.1 10.8 13.5 16.2 18.9 21.6 24.3 27 29.7 32.4 35.1 37.8 40.5 43.2 45.9
1 -45.9 34 302 215 133 55 343 275 213 156 104 59 19 345 317 295 279 270 267 270 279 295 317 345 19 59 104 156 213 275 343 55 133 215 302 34
2 -43.2 302 210 121 38 319 245 177 113 56 3 317 276 242 213 191 175 166 163 166 175 191 213 242 276 317 3 56 113 177 245 319 38 121 210 302
3 -40.5 215 121 32 307 228 153 83 19 320 267 220 178 143 114 92 76 66 63 66 76 92 114 143 178 220 267 320 19 83 153 228 307 32 121 215
4 -37.8 133 38 307 222 141 65 354 289 229 175 127 85 50 20 358 341 331 328 331 341 358 20 50 85 127 175 229 289 354 65 141 222 307 38 133
5 -35.1 55 319 228 141 59 342 270 204 143 89 40 358 321 291 268 252 241 238 241 252 268 291 321 358 40 89 143 204 270 342 59 141 228 319 55
6 -32.4 343 245 153 65 342 264 191 124 63 7 318 275 238 208 184 167 157 154 157 167 184 208 238 275 318 7 63 124 191 264 342 65 153 245 343
7 -29.7 275 177 83 354 270 191 118 50 348 291 241 198 160 130 106 89 78 75 78 89 106 130 160 198 241 291 348 50 118 191 270 354 83 177 275
8 -27 213 113 19 289 204 124 50 341 278 221 171 126 89 58 33 16 5 2 5 16 33 58 89 126 171 221 278 341 50 124 204 289 19 113 213
9 -24.3 156 56 320 229 143 63 348 278 215 157 106 61 23 351 327 309 299 295 299 309 327 351 23 61 106 157 215 278 348 63 143 229 320 56 156
10 -21.6 104 3 267 175 89 7 291 221 157 99 47 2 323 292 267 249 238 234 238 249 267 292 323 2 47 99 157 221 291 7 89 175 267 3 104
11 -18.9 59 317 220 127 40 318 241 171 106 47 355 309 270 238 213 195 184 180 184 195 213 238 270 309 355 47 106 171 241 318 40 127 220 317 59
12 -16.2 19 276 178 85 358 275 198 126 61 2 309 263 224 191 166 148 137 133 137 148 166 191 224 263 309 2 61 126 198 275 358 85 178 276 19
13 -13.5 345 242 143 50 321 238 160 89 23 323 270 224 184 151 126 108 97 93 97 108 126 151 184 224 270 323 23 89 160 238 321 50 143 242 345
14 -10.8 317 213 114 20 291 208 130 58 351 292 238 191 151 119 93 74 63 60 63 74 93 119 151 191 238 292 351 58 130 208 291 20 114 213 317
15 -8.1 295 191 92 358 268 184 106 33 327 267 213 166 126 93 67 49 37 34 37 49 67 93 126 166 213 267 327 33 106 184 268 358 92 191 295
16 -5.4 279 175 76 341 252 167 89 16 309 249 195 148 108 74 49 30 19 15 19 30 49 74 108 148 195 249 309 16 89 167 252 341 76 175 279
17 -2.7 270 166 66 331 241 157 78 5 299 238 184 137 97 63 37 19 7 4 7 19 37 63 97 137 184 238 299 5 78 157 241 331 66 166 270
18 0 267 163 63 328 238 154 75 2 295 234 180 133 93 60 34 15 4 0 4 15 34 60 93 133 180 234 295 2 75 154 238 328 63 163 267
19 2.7 270 166 66 331 241 157 78 5 299 238 184 137 97 63 37 19 7 4 7 19 37 63 97 137 184 238 299 5 78 157 241 331 66 166 270
20 5.4 279 175 76 341 252 167 89 16 309 249 195 148 108 74 49 30 19 15 19 30 49 74 108 148 195 249 309 16 89 167 252 341 76 175 279
21 8.1 295 191 92 358 268 184 106 33 327 267 213 166 126 93 67 49 37 34 37 49 67 93 126 166 213 267 327 33 106 184 268 358 92 191 295
22 10.8 317 213 114 20 291 208 130 58 351 292 238 191 151 119 93 74 63 60 63 74 93 119 151 191 238 292 351 58 130 208 291 20 114 213 317
23 13.5 345 242 143 50 321 238 160 89 23 323 270 224 184 151 126 108 97 93 97 108 126 151 184 224 270 323 23 89 160 238 321 50 143 242 345
24 16.2 19 276 178 85 358 275 198 126 61 2 309 263 224 191 166 148 137 133 137 148 166 191 224 263 309 2 61 126 198 275 358 85 178 276 19
25 18.9 59 317 220 127 40 318 241 171 106 47 355 309 270 238 213 195 184 180 184 195 213 238 270 309 355 47 106 171 241 318 40 127 220 317 59
26 21.6 104 3 267 175 89 7 291 221 157 99 47 2 323 292 267 249 238 234 238 249 267 292 323 2 47 99 157 221 291 7 89 175 267 3 104
27 24.3 156 56 320 229 143 63 348 278 215 157 106 61 23 351 327 309 299 295 299 309 327 351 23 61 106 157 215 278 348 63 143 229 320 56 156
28 27 213 113 19 289 204 124 50 341 278 221 171 126 89 58 33 16 5 2 5 16 33 58 89 126 171 221 278 341 50 124 204 289 19 113 213
29 29.7 275 177 83 354 270 191 118 50 348 291 241 198 160 130 106 89 78 75 78 89 106 130 160 198 241 291 348 50 118 191 270 354 83 177 275
30 32.4 343 245 153 65 342 264 191 124 63 7 318 275 238 208 184 167 157 154 157 167 184 208 238 275 318 7 63 124 191 264 342 65 153 245 343
31 35.1 55 319 228 141 59 342 270 204 143 89 40 358 321 291 268 252 241 238 241 252 268 291 321 358 40 89 143 204 270 342 59 141 228 319 55
32 37.8 133 38 307 222 141 65 354 289 229 175 127 85 50 20 358 341 331 328 331 341 358 20 50 85 127 175 229 289 354 65 141 222 307 38 133
33 40.5 215 121 32 307 228 153 83 19 320 267 220 178 143 114 92 76 66 63 66 76 92 114 143 178 220 267 320 19 83 153 228 307 32 121 215
34 43.2 302 210 121 38 319 245 177 113 56 3 317 276 242 213 191 175 166 163 166 175 191 213 242 276 317 3 56 113 177 245 319 38 121 210 302
35 45.9 34 302 215 133 55 343 275 213 156 104 59 19 345 317 295 279 270 267 270 279 295 317 345 19 59 104 156 213 275 343 55 133 215 302 34
Figure B.1: Continuous reflection phase required from unit cells in a reflectarray of 35
× 35 unit cells. Top row and left most column indicate the number o unit cell in x and
y directions. Similarly unit cells’ central locations are listed in mm for both axis.
254
UC # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
UC # mm -45.9 -43.2 -40.5 -37.8 -35.1 -32.4 -29.7 -27 -24.3 -21.6 -18.9 -16.2 -13.5 -10.8 -8.1 -5.4 -2.7 0 2.7 5.4 8.1 10.8 13.5 16.2 18.9 21.6 24.3 27 29.7 32.4 35.1 37.8 40.5 43.2 45.9
1 -45.9 0 0 240 120 0 0 240 240 120 120 0 0 0 0 240 240 240 240 240 240 240 0 0 0 0 120 120 240 240 0 0 120 240 0 0
2 -43.2 0 240 120 0 0 240 120 120 0 0 0 240 240 240 240 120 120 120 120 120 240 240 240 240 0 0 0 120 120 240 0 0 120 240 0
3 -40.5 240 120 0 0 240 120 120 0 0 240 240 120 120 120 120 120 120 120 120 120 120 120 120 120 240 240 0 0 120 120 240 0 0 120 240
4 -37.8 120 0 0 240 120 120 0 240 240 120 120 120 0 0 0 0 0 0 0 0 0 0 0 120 120 120 240 240 0 120 120 240 0 0 120
5 -35.1 0 0 240 120 0 0 240 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 240 0 0 120 240 0 0
6 -32.4 0 240 120 120 0 240 240 120 120 0 0 240 240 240 240 120 120 120 120 120 240 240 240 240 0 0 120 120 240 240 0 120 120 240 0
7 -29.7 240 120 120 0 240 240 120 0 0 240 240 240 120 120 120 120 120 120 120 120 120 120 120 240 240 240 0 0 120 240 240 0 120 120 240
8 -27 240 120 0 240 240 120 0 0 240 240 120 120 120 0 0 0 0 0 0 0 0 0 120 120 120 240 240 0 0 120 240 240 0 120 240
9 -24.3 120 0 0 240 120 120 0 240 240 120 120 120 0 0 0 0 240 240 240 0 0 0 0 120 120 120 240 240 0 120 120 240 0 0 120
10 -21.6 120 0 240 120 120 0 240 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 240 0 120 120 240 0 120
11 -18.9 0 0 240 120 0 0 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 0 0 120 240 0 0
12 -16.2 0 240 120 120 0 240 240 120 120 0 0 240 240 240 120 120 120 120 120 120 120 240 240 240 0 0 120 120 240 240 0 120 120 240 0
13 -13.5 0 240 120 0 0 240 120 120 0 0 240 240 240 120 120 120 120 120 120 120 120 120 240 240 240 0 0 120 120 240 0 0 120 240 0
14 -10.8 0 240 120 0 240 240 120 0 0 240 240 240 120 120 120 120 120 0 120 120 120 120 120 240 240 240 0 0 120 240 240 0 120 240 0
15 -8.1 240 240 120 0 240 240 120 0 0 240 240 120 120 120 120 0 0 0 0 0 120 120 120 120 240 240 0 0 120 240 240 0 120 240 240
16 -5.4 240 120 120 0 240 120 120 0 0 240 240 120 120 120 0 0 0 0 0 0 0 120 120 120 240 240 0 0 120 120 240 0 120 120 240
17 -2.7 240 120 120 0 240 120 120 0 240 240 240 120 120 120 0 0 0 0 0 0 0 120 120 120 240 240 240 0 120 120 240 0 120 120 240
18 0 240 120 120 0 240 120 120 0 240 240 240 120 120 0 0 0 0 0 0 0 0 0 120 120 240 240 240 0 120 120 240 0 120 120 240
19 2.7 240 120 120 0 240 120 120 0 240 240 240 120 120 120 0 0 0 0 0 0 0 120 120 120 240 240 240 0 120 120 240 0 120 120 240
20 5.4 240 120 120 0 240 120 120 0 0 240 240 120 120 120 0 0 0 0 0 0 0 120 120 120 240 240 0 0 120 120 240 0 120 120 240
21 8.1 240 240 120 0 240 240 120 0 0 240 240 120 120 120 120 0 0 0 0 0 120 120 120 120 240 240 0 0 120 240 240 0 120 240 240
22 10.8 0 240 120 0 240 240 120 0 0 240 240 240 120 120 120 120 120 0 120 120 120 120 120 240 240 240 0 0 120 240 240 0 120 240 0
23 13.5 0 240 120 0 0 240 120 120 0 0 240 240 240 120 120 120 120 120 120 120 120 120 240 240 240 0 0 120 120 240 0 0 120 240 0
24 16.2 0 240 120 120 0 240 240 120 120 0 0 240 240 240 120 120 120 120 120 120 120 240 240 240 0 0 120 120 240 240 0 120 120 240 0
25 18.9 0 0 240 120 0 0 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 0 0 120 240 0 0
26 21.6 120 0 240 120 120 0 240 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 240 0 120 120 240 0 120
27 24.3 120 0 0 240 120 120 0 240 240 120 120 120 0 0 0 0 240 240 240 0 0 0 0 120 120 120 240 240 0 120 120 240 0 0 120
28 27 240 120 0 240 240 120 0 0 240 240 120 120 120 0 0 0 0 0 0 0 0 0 120 120 120 240 240 0 0 120 240 240 0 120 240
29 29.7 240 120 120 0 240 240 120 0 0 240 240 240 120 120 120 120 120 120 120 120 120 120 120 240 240 240 0 0 120 240 240 0 120 120 240
30 32.4 0 240 120 120 0 240 240 120 120 0 0 240 240 240 240 120 120 120 120 120 240 240 240 240 0 0 120 120 240 240 0 120 120 240 0
31 35.1 0 0 240 120 0 0 240 240 120 120 0 0 0 240 240 240 240 240 240 240 240 240 0 0 0 120 120 240 240 0 0 120 240 0 0
32 37.8 120 0 0 240 120 120 0 240 240 120 120 120 0 0 0 0 0 0 0 0 0 0 0 120 120 120 240 240 0 120 120 240 0 0 120
33 40.5 240 120 0 0 240 120 120 0 0 240 240 120 120 120 120 120 120 120 120 120 120 120 120 120 240 240 0 0 120 120 240 0 0 120 240
34 43.2 0 240 120 0 0 240 120 120 0 0 0 240 240 240 240 120 120 120 120 120 240 240 240 240 0 0 0 120 120 240 0 0 120 240 0
35 45.9 0 0 240 120 0 0 240 240 120 120 0 0 0 0 240 240 240 240 240 240 240 0 0 0 0 120 120 240 240 0 0 120 240 0 0
Figure B.2: Discrete 3 states quantized reflection phase required from unit cells in a
reflectarray of 35 × 35 unit cells. Top row and left most column indicate the number
o unit cell in x and y directions. Similarly unit cells’ central locations are listed in mm
for both axis.
255
UC # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
UC # mm -45.9 -43.2 -40.5 -37.8 -35.1 -32.4 -29.7 -27 -24.3 -21.6 -18.9 -16.2 -13.5 -10.8 -8.1 -5.4 -2.7 0 2.7 5.4 8.1 10.8 13.5 16.2 18.9 21.6 24.3 27 29.7 32.4 35.1 37.8 40.5 43.2 45.9
1 -45.9 0 0 -5 5 0 0 -5 -5 5 5 0 0 0 0 -5 -5 -5 -5 -5 -5 -5 0 0 0 0 5 5 -5 -5 0 0 5 -5 0 0
2 -43.2 0 -5 5 0 0 -5 5 5 0 0 0 -5 -5 -5 -5 5 5 5 5 5 -5 -5 -5 -5 0 0 0 5 5 -5 0 0 5 -5 0
3 -40.5 -5 5 0 0 -5 5 5 0 0 -5 -5 5 5 5 5 5 5 5 5 5 5 5 5 5 -5 -5 0 0 5 5 -5 0 0 5 -5
4 -37.8 5 0 0 -5 5 5 0 -5 -5 5 5 5 0 0 0 0 0 0 0 0 0 0 0 5 5 5 -5 -5 0 5 5 -5 0 0 5
5 -35.1 0 0 -5 5 0 0 -5 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 -5 0 0 5 -5 0 0
6 -32.4 0 -5 5 5 0 -5 -5 5 5 0 0 -5 -5 -5 -5 5 5 5 5 5 -5 -5 -5 -5 0 0 5 5 -5 -5 0 5 5 -5 0
7 -29.7 -5 5 5 0 -5 -5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 -5 -5 0 5 5 -5
8 -27 -5 5 0 -5 -5 5 0 0 -5 -5 5 5 5 0 0 0 0 0 0 0 0 0 5 5 5 -5 -5 0 0 5 -5 -5 0 5 -5
9 -24.3 5 0 0 -5 5 5 0 -5 -5 5 5 5 0 0 0 0 -5 -5 -5 0 0 0 0 5 5 5 -5 -5 0 5 5 -5 0 0 5
10 -21.6 5 0 -5 5 5 0 -5 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 -5 0 5 5 -5 0 5
11 -18.9 0 0 -5 5 0 0 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 0 0 5 -5 0 0
12 -16.2 0 -5 5 5 0 -5 -5 5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 5 -5 -5 0 5 5 -5 0
13 -13.5 0 -5 5 0 0 -5 5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 5 -5 0 0 5 -5 0
14 -10.8 0 -5 5 0 -5 -5 5 0 0 -5 -5 -5 5 5 5 5 5 0 5 5 5 5 5 -5 -5 -5 0 0 5 -5 -5 0 5 -5 0
15 -8.1 -5 -5 5 0 -5 -5 5 0 0 -5 -5 5 5 5 5 0 0 0 0 0 5 5 5 5 -5 -5 0 0 5 -5 -5 0 5 -5 -5
16 -5.4 -5 5 5 0 -5 5 5 0 0 -5 -5 5 5 5 0 0 0 0 0 0 0 5 5 5 -5 -5 0 0 5 5 -5 0 5 5 -5
17 -2.7 -5 5 5 0 -5 5 5 0 -5 -5 -5 5 5 5 0 0 0 0 0 0 0 5 5 5 -5 -5 -5 0 5 5 -5 0 5 5 -5
18 0 -5 5 5 0 -5 5 5 0 -5 -5 -5 5 5 0 0 0 0 0 0 0 0 0 5 5 -5 -5 -5 0 5 5 -5 0 5 5 -5
19 2.7 -5 5 5 0 -5 5 5 0 -5 -5 -5 5 5 5 0 0 0 0 0 0 0 5 5 5 -5 -5 -5 0 5 5 -5 0 5 5 -5
20 5.4 -5 5 5 0 -5 5 5 0 0 -5 -5 5 5 5 0 0 0 0 0 0 0 5 5 5 -5 -5 0 0 5 5 -5 0 5 5 -5
21 8.1 -5 -5 5 0 -5 -5 5 0 0 -5 -5 5 5 5 5 0 0 0 0 0 5 5 5 5 -5 -5 0 0 5 -5 -5 0 5 -5 -5
22 10.8 0 -5 5 0 -5 -5 5 0 0 -5 -5 -5 5 5 5 5 5 0 5 5 5 5 5 -5 -5 -5 0 0 5 -5 -5 0 5 -5 0
23 13.5 0 -5 5 0 0 -5 5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 5 -5 0 0 5 -5 0
24 16.2 0 -5 5 5 0 -5 -5 5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 5 -5 -5 0 5 5 -5 0
25 18.9 0 0 -5 5 0 0 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 0 0 5 -5 0 0
26 21.6 5 0 -5 5 5 0 -5 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 -5 0 5 5 -5 0 5
27 24.3 5 0 0 -5 5 5 0 -5 -5 5 5 5 0 0 0 0 -5 -5 -5 0 0 0 0 5 5 5 -5 -5 0 5 5 -5 0 0 5
28 27 -5 5 0 -5 -5 5 0 0 -5 -5 5 5 5 0 0 0 0 0 0 0 0 0 5 5 5 -5 -5 0 0 5 -5 -5 0 5 -5
29 29.7 -5 5 5 0 -5 -5 5 0 0 -5 -5 -5 5 5 5 5 5 5 5 5 5 5 5 -5 -5 -5 0 0 5 -5 -5 0 5 5 -5
30 32.4 0 -5 5 5 0 -5 -5 5 5 0 0 -5 -5 -5 -5 5 5 5 5 5 -5 -5 -5 -5 0 0 5 5 -5 -5 0 5 5 -5 0
31 35.1 0 0 -5 5 0 0 -5 -5 5 5 0 0 0 -5 -5 -5 -5 -5 -5 -5 -5 -5 0 0 0 5 5 -5 -5 0 0 5 -5 0 0
32 37.8 5 0 0 -5 5 5 0 -5 -5 5 5 5 0 0 0 0 0 0 0 0 0 0 0 5 5 5 -5 -5 0 5 5 -5 0 0 5
33 40.5 -5 5 0 0 -5 5 5 0 0 -5 -5 5 5 5 5 5 5 5 5 5 5 5 5 5 -5 -5 0 0 5 5 -5 0 0 5 -5
34 43.2 0 -5 5 0 0 -5 5 5 0 0 0 -5 -5 -5 -5 5 5 5 5 5 -5 -5 -5 -5 0 0 0 5 5 -5 0 0 5 -5 0
35 45.9 0 0 -5 5 0 0 -5 -5 5 5 0 0 0 0 -5 -5 -5 -5 -5 -5 -5 0 0 0 0 5 5 -5 -5 0 0 5 -5 0 0
Figure B.3: DC bias voltage for discrete 3 states quantized reflection phase in a
reflectarray of 35 × 35 unit cells. Top row and left most column indicate the number
o unit cell in x and y directions. Similarly unit cells’ central locations are listed in
mm for both axis. Here, 0 V, 5 V and -5 V are shown. However, these diodes can be
operated by a ±1.5 V supply instead of ±5 V.
Bibliography
[1] Harvey Lehpamer. Millimeter-wave radios in backhaul networks. Communication

Inrastructure Corporation, 2008.
[2] J. Zhang, X. Ge, Q. Li, M. Guizani, and Y. Zhang. 5G millimeter-wave antenna
array: Design and challenges. IEEE Wireless Communications, 24(2):106–112,
April 2017.
[3] J. G. Andrews, T. Bai, M. N. Kulkarni, A. Alkhateeb, A. K. Gupta, and
R. W. Heath. Modeling and analyzing millimeter wave cellular systems. IEEE
Transactions on Communications, 65(1):403–430, Jan 2017.
[4] Dhruva Teja Jangam, V Chaitanya Reddy, TSR Prasad, and T Ravi. Mm waves
role in satellite communication. IOSR Journal of Electronics and Communication
Engineering (IOSR-JECE), 4:01–04, 2013.
[5] Potturi Raghavendra Rao. Role of mm waves in terrestrial and satellite
communications. International Journal of Systems and Technologies, 3(1):25–34,
2010.
[6] Next generation wireless: Riding on the back of millimeter waves.
[Online]. Available: [http://www.rec-usa.com/press/Next%20Generation%
20Wireless.pdf], (accessed Feb. 2015).
[7] Tasuku Teshirogi. Modern millimeter-wave technologies. IOS Press, 2001.
[8] Lockheed martin pushes forward on additive manufacturing for satellite
construction. [Online]. Available: [http://www.rapidreadytech.com/2014/
07/lockheed-martin-pushes-forward-on-additive-manufacturing-for-:
satellite-construction/], (accessed 27 Mar. 2018).
[9] Barrie D Dunn. Requirements for spacecraft materials. In Materials and
Processes, pages 7–53. Springer, 2016.
[10] Hervé Legay, Yann Cailloce, Olivier Vendier, Gérard Caille, Julien
Perruisseau-Carrier, Markku Lathi, Jean Philippe Polizzi, Ulf Oestermann,
Patrick Pons, and Nathalie Raveu. Satellite antennas based on MEMS tunable
reflectarrays. IET, 2007.
[11] Rendong Nan, Di Li, Chengjin Jin, Qiming Wang, Lichun Zhu, Wenbai Zhu,
Haiyan Zhang, Youling Yue, and Lei Qian. The five-hundred-meter aperture
spherical radio telescope (fast) project. International Journal of Modern Physics
D, 20(06):989–1024, 2011.
[12] Atacama large millimeter array. [Online]. Available: [http://www.
almaobservatory.org/en/images-tags/antennas/], (accessed 27 Mar. 2018).
256
Bibliography 257
[13] The solid-state phased array radar system (SSPARS) in protective

dome at clear air force station. [Online]. Available: [http://www.
airforce-technology.com/projects/clear-air-force-station-alaska/
attachment/clear-air-force-station-alaska2/], (accessed 27 Mar. 2018).
[14] Wael M Abdel-Wahab, Safieddin Safavi-Naeini, and Dan Busuioc. Low cost 60
GHz millimeter-wave microstrip patch antenna array using low-loss planar feeding
scheme. In Antennas and Propagation (APSURSI), 2011 IEEE International
Symposium on, pages 508–511. IEEE, 2011.
[15] Samet Zihir, Ozan D Gurbuz, Arjun Karroy, Sanjay Raman, and Gabriel M
Rebeiz. A 60 GHz single-chip 256-element wafer-scale phased array with EIRP
of 45 dBm using sub-reticle stitching. In Radio Frequency Integrated Circuits
Symposium (RFIC), 2015 IEEE, pages 23–26. IEEE, 2015.
[16] S Zihir, OD Gurbuz, A Karroy, S Raman, and GM Rebeiz. A 60 GHz 64-element
wafer-scale phased-array with full-reticle design. In Microwave Symposium (IMS),
2015 IEEE MTT-S International, pages 1–3. IEEE, 2015.
[17] Phased array antenna. [Online]. Available: [http://www.radartutorial.eu/
06.antennas/Phased%20Array%20Antenna.en.html], (accessed 27 Mar. 2018).
[18] Sean Victor Hum and Julien Perruisseau-Carrier. Reconfigurable reflectarrays
and array lenses for dynamic antenna beam control: A review. IEEE Transactions
on Antennas and Propagation, 62(1):183–198, 2014.
[19] Sean Victor Hum. Reflectarrays. In Aperture Antennas for Millimeter and
Sub-Millimeter Wave Applications, pages 143–189. Springer, 2018.
[20] Payam Nayeri, Fan Yang, and Atef Z Elsherbeni. Reflectarray Antennas: Theory,
Designs, and Applications. Wiley Online Library, 2018.
[21] Hervé Legay, Daniele Bresciani, Eric Labiole, Renaud Chiniard, and Raphaël
Gillard. A multi facets composite panel reflectarray antenna for a space contoured
beam antenna in Ku band. Progress In Electromagnetics Research, 54:1–26, 2013.
[22] Richard E Hodges, Daniel J Hoppe, Matthew J Radway, and Nacer E Chahat.
Novel deployable reflectarray antennas for cubesat communications. In Microwave
Symposium (IMS), 2015 IEEE MTT-S International, pages 1–4. IEEE, 2015.
[23] Richard E Hodges, Matthew J Radway, Armen Toorian, Daniel J Hoppe,
Biren Shah, and Andrew E Kalman. ISARA-integrated solar array and
reflectarray cubesat deployable Ka-band antenna. In Antennas and Propagation
& USNC/URSI National Radio Science Meeting, 2015 IEEE International
[24] John Huang. Bandwidth study of microstrip reflectarray and a novel phased
reflectarray concept. In Antennas and Propagation Society International
Symposium, 1995. AP-S. Digest, volume 1, pages 582–585. IEEE, 1995.
[25] John Huang. Reflectarray antenna. Wiley Online Library, 2008.
[26] DM Pozar. Bandwidth of reflectarrays. Electronics Letters, 39(21):1490–1491,
2003.
[27] David M Pozar, Stephen D Targonski, and HD Syrigos. Design of millimeter
wave microstrip reflectarrays. IEEE transactions on antennas and propagation,
45(2):287–296, 1997.
Bibliography 258
[28] Hervé Legay, Daniele Bresciani, Eric Labiole, Renaud Chiniard, Etienne Girard,
Gérard Caille, Daniel Calas, Raphaël Gillard, and Giovanni Toso. A 1.3
m facetted reflectarray in Ku band. In Antenna Technology and Applied
Electromagnetics (ANTEM), 2012 15th International Symposium on, pages 1–4.
IEEE, 2012.
[29] Min Zhou, Stig B Sørensen, Oleksiy S Kim, Erik Jørgensen, Peter Meincke, and
Olav Breinbjerg. Direct optimization of printed reflectarrays for contoured beam
satellite antenna applications. IEEE Transactions on Antennas and Propagation,
61(4):1995–2004, 2013.
[30] Min Zhou, Stig B Sorensen, Oleksiy S Kim, Erik Jorgensen, Peter Meincke, Olav
Breinbjerg, and Giovanni Toso. The generalized direct optimization technique
for printed reflectarrays. IEEE Transactions on Antennas and Propagation,
62(4):1690–1700, 2014.
[31] Oscar Borries, Stig Busk, Hans-Henrik Viskum, et al. Efficient surface
optimization of large dual reflector systems. In Antennas and Propagation
[32] ISARA (integrated solar array and reflectarray antenna). [Online]. Available:
[https://directory.eoportal.org/web/eoportal/satellite-missions/i/
isara], (accessed 27 Mar. 2018).
[33] Integrated solar array and reflectarray antenna (ISARA). [Online]. Available:
[https://www.nasa.gov/directorates/spacetech/small_spacecraft/
isara_project.html], (accessed 27 Mar. 2018).
[34] William Hart Hayt and John A Buck. Engineering electromagnetics, volume 6.
McGraw-Hill New York, 1981.
[35] Balanis Constantine A. Antenna theory: analysis and design. John wiley & sons,
third edition, 2005.
[36] Harald T Friis. A note on a simple transmission formula. Proceedings of the IRE,
34(5):254–256, 1946.
[37] Natalia K. Nikolova. Friis transmission equation and radar range equation, lecture
6. Department of Electrical and Computer Engineering, McMaster University,
Canada, 2018.
[38] F. Khan, Z. Pi, and S. Rajagopal. Millimeter-wave mobile broadband with large
scale spatial processing for 5G mobile communication. In 2012 50th Annual
Allerton Conference on Communication, Control, and Computing (Allerton),
pages 1517–1523, Oct 2012.
[39] Sophocles J Orfanidis. Electromagnetic waves and antennas. Rutgers University
New Brunswick, NJ, 2002.
[40] Final acts WRC-15 world radiocommunicati on conference, Geneva, International
Telecommunication Union, 2015.
[41] M. Marcus and B. Pattan. Millimeter wave propagation; spectrum management
implications. IEEE Microwave Magazine, 6(2):54–62, June 2005.
[42] Millimeter wave propagation: Spectrum management implications, FCC office
of engg. and tech. [Online]. Available: [http://transition.fcc.gov/Bureaus/
Bibliography 259
Engineering_Technology/Documents/bulletins/oet70/oet70a.pdf],
(accessed February 20, 2015).
[43] 4G Americas’ MIMO and smart antennas for mobile broadband systems. 4G
Americas, Bellevue, Washington, 2012.
[44] Constantine A Balanis and Panayiotis I Ioannides. Introduction to smart
antennas. Synthesis Lectures on Antennas, 2(1):1–175, 2007.
[45] Ivica Stevanovic, Anja Skrivervik, and Juan R Mosig. Smart antenna systems for
mobile communications. Technical report, 2003.
[46] Tinith Pitakandage, Milos Milosavljevic, Pandelis Kourtessis, and John M Senior.
Cooperative 5G switched and adaptive receiver beamforming for fibre wireless
networks. In Transparent Optical Networks (ICTON), 2014 16th International
Conference on, pages 1–4. IEEE, 2014.
[47] CG Christodoulou, M Ciaurriz, Y Tawk, J Costantine, and SE Barbin. Recent
advances in randomly spaced antenna arrays. In Antennas and Propagation
(EuCAP), 2014 8th European Conference on, pages 732–736. IEEE, 2014.
[48] Tanuja Satish Dhope and Dina Simunic. On the performance of DoA estimation
algorithms in cognitive radio networks:-a new approach in spectrum sensing.
In Information & Communication Technology Electronics & Microelectronics
(MIPRO), 2013 36th International Convention on, pages 507–512. IEEE, 2013.
[49] Murat Torlak. Spatial array processing. The University of Texas at Austin, 2018.
[50] Nihad AA Elhag, Isam M Osman, Alaa A Yassin, and Tsneem B Ahmed.
Angle of arrival estimation in smart antenna using music method for wideband
wireless communication. In Computing, Electrical and Electronics Engineering
(ICCEEE), 2013 International Conference on, pages 69–73. IEEE, 2013.
[51] Moustafa M Abdalla, Mostafa B Abuitbel, and Mohamed A Hassan. Performance
evaluation of direction of arrival estimation using music and esprit algorithms for
mobile communication systems. In Wireless and Mobile Networking Conference
(WMNC), 2013 6th Joint IFIP, pages 1–7. IEEE, 2013.
[52] Yuanming Guo, Wei Li, Yanyan Zuo, and Junyuan Shen. A nyström-based esprit
algorithm for DOA estimation of coherent signals. In Antennas & Propagation
(ISAP), 2013 Proceedings of the International Symposium on, volume 1, pages
426–429. IEEE, 2013.
[53] S Snigdha, SV Geethanjali, MSG Prasad, and TV Rama Krishna. Analysis of
two different DOA algorithms for the estimation of desired signal using smart
antenna technology. 2013.
[54] Nizar Tayem, Syed Ahmed Raza, Muhammad Omer, Mohamed El-Lakki, and
Jamal F Nayfeh. Hardware implementation of a proposed QR-TLS DOA
estimation method and Music, ESPRIT algorithms on Ni-Pxi platform. Progress
In Electromagnetics Research, 45:203–221, 2013.
[55] Xu-Hu Wang, Jian-Feng Chen, Jing Han, and Qin-Li An. Performance analysis
of DOA estimation of vector hydrophone array based on ESPRIT algorithm.
Systems Engineering and electronics, 35(3):481–486, 2013.
[56] Kunlai Xiong, Zheng Liu, and Wenli Jiang. SAGE-based algorithm for
Bibliography 260
direction-of-arrival estimation and array calibration. International Journal of

Antennas and Propagation, 2014, 2014.
[57] Mallikarjun C Sarsamba and Mallikarjun Talwar. Evaluation of smart antenna
for 3G network: A survey. History, 10(21):24–32, 2014.
[58] Thomas Jost, Wei Wang, Fernando Pérez-Fontán, and Uwe-Carsten Fiebig. A
satellite-to-indoor channel model for pedestrian applications. In Antennas and
Propagation (EuCAP), 2014 8th European Conference on, pages 1259–1263.
IEEE, 2014.
[59] Pratibha B Anantharamu, O Borio, and Gérard Lachapelle. Space-time
equalization techniques for new GNSS signals. GPS World, 22(10):36–41, 2011.
[60] Christopher Taylor. 802.11ac wave 2 with MU-MIMO:
The next mainstream Wi-Fi standard. [Online].
Available: [https://www.qualcomm.com/media/documents/files/
strategy-analytics-802-11ac-wave-2-with-mu-mimo-the-next-:
mainstream-wi-fi-standard.pdf], 2015.
[61] Wireless communication: 60 GHz wireless communication, IMEC.
[Online]. Available: [http://www2.imec.be/be_en/research/
wireless-communication/60ghz-wireless-communication.html], (accesssed
Oct. 2014).
[62] Taeyoung Kim, Jeongho Park, Ji-Yun Seol, Suryong Jeong, Jaeweon Cho,
and Wonil Roh. Tens of Gbps support with mmWave beamforming systems
for next generation communications. In Global Communications Conference
(GLOBECOM), 2013 IEEE, pages 3685–3690. IEEE, 2013.
[63] Wonil Roh, Ji-Yun Seol, Jeongho Park, Byunghwan Lee, Jaekon Lee, Yungsoo
Kim, Jaeweon Cho, Kyungwhoon Cheun, and Farshid Aryanfar. Millimeter-wave
beamforming as an enabling technology for 5G cellular communications:
Theoretical feasibility and prototype results. IEEE communications magazine,
52(2):106–113, 2014.
[64] C Hillman, PA Stupar, JB Hacker, Z Griffith, M Field, and M Rodwell.
An ultra-low loss millimeter-wave solid state switch technology based on the
metal-insulator-transition of vanadium dioxide. In Microwave Symposium (IMS),
2014 IEEE MTT-S International, pages 1–4. IEEE, 2014.
[65] Sukomal Dey and Shiban K Koul. Design and development of a surface
micro-machined push–pull-type true-time-delay phase shifter on an alumina
substrate for Ka-band T/R module application. Journal of Micromechanics and
Microengineering, 22(12):125006, 2012.
[66] Tanku. Y. Sule. MEMS and GaN based S-band active electronically steerable
phased array antenna for microsatellites. PhD thesis, Faculty of Engg. and
Physical Sciences, Univ. of Surrey, 2014.
[67] Pai-Yen Chen, Christos Argyropoulos, and Andrea Alu. Terahertz antenna phase
shifters using integrally-gated graphene transmission-lines. IEEE Transactions on
Antennas and Propagation, 61(4):1528–1537, 2013.
[68] H Aliakbari, A Abdipour, A Mohammadi, and R Mirzavand. A continues 360
CMOS phase shifter for 60 GHz phased array applications. PIERS Proceedings,
Moscow, Russia, 2012.
Bibliography 261
[69] Behzad Biglarbegian, Mohammad Reza Nezhad-Ahmadi, Mohammad

Fakharzadeh, and Safieddin Safavi-Naeini. Millimeter-wave reflective-type
phase shifter in CMOS technology. IEEE Microwave and Wireless components
letters, 19(9):560–562, 2009.
[70] Chunyuan Zhou, Lei Zhang, Yan Wang, Zhiping Yu, and He Qian. A 4-bit CMOS
phase shifter for millimeter-wave phased arrays. Analog Integrated Circuits and
Signal Processing, 79(3):461–468, 2014.
[71] Dmitry Chicherin, Mikael Sterner, Dmitri Lioubtchenko, Joachim Oberhammer,
and Antti V Räisänen. Analog-type millimeter-wave phase shifters based
on MEMS tunable high-impedance surface and dielectric rod waveguide.
International Journal of Microwave and Wireless Technologies, 3(5):533–538,
2011.
[72] Val Dyadyuk, Xiaojing Huang, Leigh Stokes, Joseph Pathikulangara, Andrew
Weily, Nasiha Nikolic, John Bunton, and Y Guo. Adaptive antenna arrays for
ad-hoc millimetre-wave wireless communications. In-Tech, 2011.
[73] Pietro Romano, O Araromi, Samuel Rosset, Herbert Shea, and Julien
Perruisseau-Carrier. Tunable millimeter-wave phase shifter based on dielectric
elastomer actuation. Applied Physics Letters, 104(2):024104, 2014.
[74] Waveguide manual phase shifter, rf-lambda. [Online]. Available: [http:
//www.rflambda.com/pdf/wgphaseshifter/WGPhaseShifter.pdf], (accessed
Oct, 2014).
[75] R. Singel. Ruckus smart antennas may be key to nationwide Wi-Fi. [Online].
Available: [http://www.wired.com/2010/12/ruckus-wifi-3g/], (accessed Jan
10, 2015).
[76] Simply the smartest wireless for a mobile world, ruckus wireless. [Online].
Available: [http://www.ruckuswireless.com/], (accessed Jan 10, 2015).
[77] Ruckus zoneflex 7762-s, dual-band 802.11n smart Wi-Fi outdoor sector ap,
ruckus wireless. [Online]. Available: [http://www.ruckussecurity.com/
ZoneFlex-7762-S.asp], (accessed Jan 10, 2015).
[78] Adant WiFi smart antenna system, Adant Technologies Inc. [Online]. Available:
[http://www.adant.com/solutions], (accessed Oct. 2014).
[79] A revolutionary antenna system, Adant Technologies Inc. [Online]. Available:
[http://www.adant.com/technology], (Accessed Oct. 2014).
[80] 2.4 GHz reconfigurable meta-material antenna, adant Technologies Inc. [Online].
Available: [http://www.adant.com/files/adant-datasheet_2.4GHz.pdf],
(accessed Oct. 2014).
[81] 5 GHz reconfigurable meta-material antenna, Adant Technologies Inc.
[Online]. Available: [http://www.adant.com/files/adant-datasheet_5GHz.
pdf], (accessed Oct. 2014).
[82] Aruba 130 series access point, Aruba networks. [Online]. Available: [http://
www.arubanetworks.com/pdf/products/DS_AP130Series.pdf], (accessed Oct.
2014).
[83] Cisco aironet 1140 series access point data sheet, Cisco. [Online]. Available:
[http://www.cisco.com/c/en/us/products/collateral/wireless/
Bibliography 262
aironet-1130-ag-series/datasheet_c78-502793.html], (accessed Oct.

2014).
[84] Masataka Ohira, Amane Miura, and Masazumi Ueba. 60-GHz wideband
substrate-integrated-waveguide slot array using closely spaced elements for
planar multisector antenna. IEEE Transactions on Antennas and Propagation,
58(3):993–998, 2010.
[85] Masataka ohira: 60 GHz-band sector-switched antenna for wireless LAN.
[Online]. Available: [http://sirius.reso.ees.saitama-u.ac.jp/ohiram-e.
html], (accessed Oct. 2014).
[86] Nuri Celik, Magdy F Iskander, Rudy Emrick, Steven J Franson, and John
Holmes. Implementation and experimental verification of a smart antenna system
operating at 60 GHz band. IEEE Transactions on Antennas and Propagation,
56(9):2790–2800, 2008.
[87] Haitao Liu, Steven Gao, and Tian Hong Loh. Frequency agile small smart
antenna. In Antennas and Propagation (EuCAP), 2010 Proceedings of the Fourth
European Conference on, pages 1–4. IEEE, 2010.
[88] Haitao Liu, Steven Gao, and Tian Hong Loh. Compact-size electronically
steerable parasitic array radiator antenna. In Antennas & Propagation
Conference, 2009. LAPC 2009. Loughborough, pages 265–268. IEEE, 2009.
[89] Haitao Liu, Steven Gao, and Tian-Hong Loh. Small director array for low-profile
smart antennas achieving higher gain. IEEE Transactions on Antennas and
Propagation, 61(1):162–168, 2013.
[90] Hai-Tao Liu, Steven Gao, and Tian-Hong Loh. Electrically small and low cost
smart antenna for wireless communication. IEEE Transactions on Antennas and
Propagation, 60(3):1540–1549, 2012.
[91] Sungjoon Lim, Christophe Caloz, and Tatsuo Itoh. Metamaterial-based
electronically controlled transmission-line structure as a novel leaky-wave antenna
with tunable radiation angle and beamwidth. IEEE Transactions on Microwave
Theory and Techniques, 52(12):2678–2690, 2004.
[92] Christophe Caloz and Tatsuo Itoh. Novel microwave devices and structures based
on the transmission line approach of meta-materials. In Microwave Symposium
Digest, 2003 IEEE MTT-S International, volume 1, pages 195–198. IEEE, 2003.
[93] Sungjoon Lim, Christophe Caloz, and Tatsuo Itoh. Electronically-controlled
metamaterial-based transmission line as a continuous-scanning leaky-wave
antenna. In Microwave Symposium Digest, 2004 IEEE MTT-S International,
volume 1, pages 313–316. IEEE, 2004.
[94] Christophe Caloz and Tatsuo Itoh. Electromagnetic metamaterials: transmission
line theory and microwave applications. John Wiley & Sons, 2005.
[95] MF Jamlos, O Abdul Aziz, TA Rahman, MR Kamarudin, P Saad, MT Ali,
and MN Md Tan. A beam steering radial line slot array (rlsa) antenna with
reconfigurable operating frequency. Journal of Electromagnetic Waves and
Applications, 24(8-9):1079–1088, 2010.
[96] Massimo Donelli and Pascal Febvre. An inexpensive reconfigurable planar array
for Wi-Fi applications. Progress In Electromagnetics Research, 28:71–81, 2012.
Bibliography 263
[97] Daniel Rodrigo, Edgar Diaz, et al. Reconfigurable pixel-layer isolator for
frequency-tunable on-frequency repeaters. IEEE Antennas and Wireless
Propagation Letters, 13:475–478, 2014.
[98] Daniel Rodrigo, Bedri A Cetiner, et al. Frequency, radiation pattern and
polarization reconfigurable antenna using a parasitic pixel layer. IEEE
transactions on antennas and propagation, 62(6):3422–3427, 2014.
[99] Muzammil Jusoh. Reconfigurable Beam Steering Parasitic Patch Antenna with
Embedded PIN Diode. PhD thesis, School of Computer and Communication
Engineering, 2013.
[100] Zhouyuan Li, Daniel Rodrigo, Luis Jofre, and Bedri A Cetiner. A new class
of antenna array with a reconfigurable element factor. IEEE transactions on
antennas and propagation, 61(4):1947–1955, 2013.
[101] Xiaoyan Yuan, Zhouyuan Li, Daniel Rodrigo, Hema Swaroop Mopidevi, Oguz
Kaynar, Lluı́s Jofre, and Bedri A Cetiner. A parasitic layer-based reconfigurable
antenna design by multi-objective optimization. IEEE Transactions on Antennas
and Propagation, 60(6):2690–2701, 2012.
[102] D Berry, R Malech, and W Kennedy. The reflectarray antenna. IEEE
Transactions on Antennas and Propagation, 11(6):645–651, 1963.
[103] CS Malagisi. Microstrip disc element reflect array. In EASCON’78; Electronics
and Aerospace Systems Convention, pages 186–192, 1978.
[104] Carmen S Malagisi. Electronically scanned microstrip antenna array, October 11
1977. US Patent 4,053,895.
[105] J Montgomery. Scattering by an infinite periodic array of microstrip elements.
IEEE Transactions on Antennas and Propagation, 26(6):850–854, 1978.
[106] JP Montgomery. A microstrip reflectarray antenna element. In Annales
d’Astrophysique, 1978.
[107] DM Pozar and TA Metzler. Analysis of a reflectarray antenna using microstrip
patches of variable size. Electronics Letters, 29(8):657–658, 1993.
[108] Thomas Allan Metzler. Design and analysis of a microstrip reflectarray. PhD
thesis, University of Massachusetts Amherst, 1993.
[109] John Huang and Ronald J Pogorzelski. A Ka-band microstrip reflectarray with
elements having variable rotation angles. IEEE transactions on antennas and
propagation, 46(5):650–656, 1998.
[110] J-M Colin. Phased array radars in france: Present and future. In Phased
Array Systems and Technology, 1996., IEEE International Symposium on, pages
458–462. IEEE, 1996.
[111] AA Tolkachev, VV Denisenko, AV Shishlov, and AG Shubov. High gain
antenna systems for millimeter wave radars with combined electronical and
mechanical beam steering. In Phased Array Systems and Technology, 1996., IEEE
International Symposium on, pages 266–271. IEEE, 1996.
[112] Chulmin Han, John Huang, and Kai Chang. A high efficiency offset-fed
X/Ka-dual-band reflectarray using thin membranes. IEEE Transactions on
Antennas and Propagation, 53(9):2792–2798, 2005.
[113] Chulmin Han, Christopher Rodenbeck, John Huang, and Kai Chang. A
Bibliography 264
C/Ka dual frequency dual layer circularly polarized reflectarray antenna with
microstrip ring elements. IEEE Transactions on Antennas and Propagation,
52(11):2871–2876, 2004.
[114] Wenfei Hu, Robert Cahill, Jose A Encinar, Raymond Dickie, Harold Gamble,
Vincent Fusco, and Norman Grant. Design and measurement of reconfigurable
millimeter wave reflectarray cells with nematic liquid crystal. IEEE Transactions
on Antennas and Propagation, 56(10):3112–3117, 2008.
[115] Atef Z Elsherbeni, Payam Nayeri, and Fan Yang. Reflectarray antennas for space
applications. In Ultra-Wideband (ICUWB), 2012 IEEE International Conference
on, pages 362–365. IEEE, 2012.
[116] Fan Yang, Payam Nayeri, and Atef Z Elsherbeni. Recent advances in
beam-scanning reflectarray antennas. In General Assembly and Scientific
Symposium (URSI GASS), 2014 XXXIth URSI, pages 1–4. IEEE, 2014.
[117] Daniel Rodrigo, Lluı́s Jofre, and Julien Perruisseau-Carrier. Unit cell for
frequency-tunable beamscanning reflectarrays. IEEE transactions on antennas
and propagation, 61(12):5992–5999, 2013.
[118] MT Zhang, Steven Gao, JX Wan, BN Tian, and CB Wu. Dual-polarized unit-cell
of continuous reflective phase-shift for reconfigurable reflectarrays. In PIERS
Proceedings, 2014.
[119] C Tienda, JA Encinar, S Montori, R Vincenti Gatti, M Arrebola, and
R Sorrentino. Dual-reflectarray antenna for bidirectional satellite links in
Ku-band. In Antennas and Propagation (EUCAP), Proceedings of the 5th
European Conference on, pages 1404–1407. IEEE, 2011.
[120] Christos G Christodoulou, Youssef Tawk, Steven A Lane, and Scott R Erwin.
Reconfigurable antennas for wireless and space applications. Proceedings of the
IEEE, 100(7):2250–2261, 2012.
[121] Lina Moustafa, Raphael Gillard, Federico Peris, Renaud Loison, Herve Legay,
and Etienne Girard. The phoenix cell: A new reflectarray cell with large
bandwidth and rebirth capabilities. IEEE Antennas and wireless propagation
letters, 10:71–74, 2011.
[122] Julien Perruisseau-Carrier and Anja K Skrivervik. Requirements and challenges in
the design of high-performance MEMS reconfigurable reflectarray (RRA) cells. In
Antennas and Propagation, 2007. EuCAP 2007. The Second European Conference
on, pages 1–6. IET, 2007.
[123] Maurizio Bozzi, Simone Germani, and Luca Perregrini. Performance comparison
of different element shapes used in printed reflectarrays. IEEE Antennas and
Wireless Propagation Letters, 2(1):219–222, 2003.
[124] Maurizio Bozzi, Simone Germani, and Luca Perregrini. A figure of merit for
losses in printed reflectarray elements. IEEE Antennas and Wireless Propagation
Letters, 3(1):257–260, 2004.
[125] MR Chaharmir, J Shaker, M Cuhaci, and A Ittipiboon. A broadband reflectarray
antenna with double square rings. Microwave and Optical Technology Letters,
48(7):1317–1320, 2006.
[126] MR Chaharmir, J Shaker, M Cuhaci, and A Ittipiboon. A broadband reflectarray
antenna with double square rings as the cell elements. In Antennas and
Bibliography 265
Propagation, 2006. EuCAP 2006. First European Conference on, pages 1–4.
IEEE, 2006.
[127] MR Chaharmir and J Shaker. Broadband reflectarray with combination of cross
and rectangle loop elements. Electronics Letters, 44(11):658–659, 2008.
[128] Lu Guo, Peng Khiang Tan, and Tan-Huat Chio. Bandwidth improvement
of reflectarrays using single-layered double concentric circular ring elements.
Progress In Electromagnetics Research, 46:91–99, 2014.
[129] SH Yusop, N Misran, MT Islam, and MY Ismail. Analysis of concentric split
ring square element for broadband reflectarray antenna. International Journal of
Integrated Engineering, 1(3), 2009.
[130] Li-Shi Ren, Yong-Chang Jiao, Fan Li, Jin-Juan Zhao, and Gang Zhao.
A novel double-petal loop element for broadband reflectarray. Progress In
Electromagnetics Research, 20:157–163, 2011.
[131] Hassan Salti, Raphaël Gillard, Renaud Loison, and Laurent Le Coq. A
reflectarray antenna based on multiscale phase-shifting cell concept. IEEE
Antennas and Wireless Propagation Letters, 8:363–366, 2009.
[132] Caner Guclu, Julien Perruisseau-Carrier, and Ozlem Civi. Proof of concept of
a dual-band circularly-polarized RF MEMS beam-switching reflectarray. IEEE
[133] Onur Hamza Karabey, Saygin Bildik, Carsten Fritzsch, Sebastian Strunck,
Alexander Gaebler, Rolf Jakoby, and Atsutaka Manabe. Liquid crystal based
rekonfigurable antenna arrays. In Proceedings of the 32nd ESA Antenna
workshop, 2010.
[134] OA Araromi, Pietro Romano, Samuel Rosset, Julien Perruisseau-Carrier, and
HR Shea. A tunable millimeter-wave phase shifter driven by dielectric elastomer
actuators. In Electroactive Polymer Actuators and Devices (EAPAD) 2014,
volume 9056, page 90562M. International Society for Optics and Photonics, 2014.
[135] Julien Perruisseau-Carrier and Apostolos Georgiadis. Efficient optimization
of the phase diagram in digitally-controlled reflective cells. In Antennas and
Propagation, 2009. EuCAP 2009. 3rd European Conference on, pages 1230–1233.
IEEE, 2009.
[136] Mohsen Yousefbeiki and Julien Perruisseau-Carrier. A practical technique
for accurately modeling reconfigurable lumped components in commercial
full-wave solvers [EurAAP corner]. IEEE Antennas and Propagation Magazine,
54(5):298–303, 2012.
[137] P Hannan and M Balfour. Simulation of a phased-array antenna in waveguide.
IEEE transactions on Antennas and Propagation, 13(3):342–353, 1965.
[138] Julien Perruisseau-Carrier. Dual-polarized and polarization-flexible reflective cells
with dynamic phase control. IEEE Transactions on Antennas and Propagation,
58(5):1494–1502, 2010.
[139] Rudi H Phillion and Michal Okoniewski. Measuring the reflection from a
polarization selective reflectarray element. In Antennas and Propagation Society
International Symposium 2006, IEEE, pages 4343–4346. IEEE, 2006.
[140] A unit cell model of a single periodic waveguide phased-array antenna, cst
Bibliography 266
mirowave studio. [Online]. Available: [https://www.cst.com/Applications/

ArticlePDFid=132], (accessed Feb. 2015).
[141] Skyworks. Varactor diodes. Online, 2008. Skyworks Solutions, Inc.
[142] Hirokazu Kamoda, Toru Iwasaki, Jun Tsumochi, and Takao Kuki. 60-GHz
electrically reconfigurable reflectarray using PIN diode. In Microwave Symposium
Digest, 2009. MTT’09. IEEE MTT-S International, pages 1177–1180. IEEE,
2009.
[143] Hirokazu Kamoda, Toru Iwasaki, Jun Tsumochi, Takao Kuki, and Osamu
Hashimoto. 60-GHz electronically reconfigurable large reflectarray using
single-bit phase shifters. IEEE Transactions on Antennas and Propagation,
59(7):2524–2531, 2011.
[144] Siamak Ebadi, Roberto Vincenti Gatti, and Roberto Sorrentino. Linear
reflectarray antenna design using 1-bit digital phase shifters. In Antennas and
IEEE, 2009.
[145] Michel Clenet and Gilbert Morin. Graphical investigation of quantisation
effects of phase shifters on array patterns. Technical report, Defence research
establishment OTTAWA (ONTARIO), 2000.
[146] Billy Wu, Adrian Sutinjo, Mike E Potter, and Michal Okoniewski. On the
selection of the number of bits to control a dynamic digital MEMS reflectarray.
IEEE antennas and wireless propagation letters, 7:183–186, 2008.
[147] Xavier Artiga, Julien Perruisseau-Carrier, Daniele Bresciani, et al. Polarimetric
control of the scattering parameters of reflectarray cells. In 7th European
Conference on Antennas and Propagation, number EPFL-CONF-182856, 2013.
[148] Dau-Chyrh Chang and Ming-Chih Huang. Multiple-polarization microstrip
reflectarray antenna with high efficiency and low cross-polarization. IEEE
[149] H Hasani, M Kamyab, and A Mirkamali. Low cross-polarization reflectarray
antenna. IEEE Transactions on Antennas and Propagation, 59(5):1752–1756,
2011.
[150] Payam Nayeri, Fan Yang, and Atef Z Elsherbeni. Beam-scanning reflectarray
antennas: A technical overview and state of the art. IEEE Antennas and
Propagation Magazine, 57(4):32–47, 2015.
[151] Min Zhou, Erik Jorgensen, Oleksiy S Kim, Stig B Sorensen, Peter Meincke, and
Olav Breinbjerg. Accurate and efficient analysis of printed reflectarrays with
arbitrary elements using higher-order hierarchical legendre basis functions. IEEE
[152] Ming-Yang Zhao, Guang-Qiu Zhang, Xue Lei, Jun-Mo Wu, and Ji-Yong Shang.
Design of new single-layer multiple-resonance broadband circularly polarized
reflectarrays. IEEE Antennas and Wireless Propagation Letters, 12:356–359,
2013.
[153] Selahattin Nesil, Filiz Güneş, and Ufok Özkaya. Phase characterization of a
reflectarray unit cell with minkowski shape radiating element using multilayer
perceptron neural network. In Electrical and Electronics Engineering (ELECO),
2011 7th International Conference on, pages II–219. IEEE, 2011.
Bibliography 267
[154] G Perez-Palomino, JA Encinar, M Barba, and E Carrasco. Design and

evaluation of multi-resonant unit cells based on liquid crystals for reconfigurable
reflectarrays. IET microwaves, antennas & propagation, 6(3):348–354, 2012.
[155] Hamed Hasani, Custódio Peixeiro, Anja K Skrivervik, and Julien
Perruisseau-Carrier. Single-layer quad-band printed reflectarray antenna
with dual linear polarization. IEEE Transactions on Antennas and Propagation,
63(12):5522–5528, 2015.
[156] Xiaoyue Xia, Qi Wu, Haiming Wang, Chen Yu, and Wei Hong. Wideband
microstrip reflectarray using dual-resonance unit cell. In Microwave Conference
(APMC), 2015 Asia-Pacific, volume 2, pages 1–3. IEEE, 2015.
[157] Muhammad M Tahseen and Ahmed A Kishk. High efficiency Ka-band single
layer air vias reflectarray: Design and analysis. In Antennas and Propagation
[158] Hoda Nematollahi, Jean-Jacques Laurin, JE Page, and José A Encinar. Design of
broadband transmitarray unit cells with comparative study of different numbers
of layers. IEEE Transactions on Antennas and Propagation, 63(4):1473–1481,
2015.
[159] Zuoxing Dai, Yuanbo Shang, Fengwei Yao, and Xiaobo Xuan. Analysis and design
of beam-scanning reflectarray with circular polarization. In PIERS Proceedings,
2014.
[160] Muhammad Kamran Arshad and FA Tahir. Optimum microstrip reflectarray
unit cell design for wide-band operation. In Frontiers of Information Technology
(FIT), 2013 11th International Conference on, pages 150–153. IEEE, 2013.
[161] Harish Rajagopalan and Yahya Rahmat-Samii. On the reflection characteristics
of a reflectarray element with low-loss and high-loss substrates. IEEE Antennas
and Propagation Magazine, 52(4):73–89, 2010.
[162] Kalyan K Karnati, Yazid Yusuf, Siamak Ebadi, and Xun Gong. Theoretical
analysis on reflection properties of reflectarray unit cells using quality factors.
IEEE Transactions on antennas and propagation, 61(1):201–210, 2013.
[163] Farooq Khan and Zhouyue Pi. mmWave mobile broadband (MMB): Unleashing
the 3–300 GHz spectrum. In Sarnoff Symposium, 2011 34th IEEE, pages 1–6.
IEEE, 2011.
[164] Theodore S Rappaport, Shu Sun, Rimma Mayzus, Hang Zhao, Yaniv Azar, Kevin
Wang, George N Wong, Jocelyn K Schulz, Mathew Samimi, and Felix Gutierrez.
Millimeter wave mobile communications for 5G cellular: It will work! IEEE
access, 1:335–349, 2013.
[165] Hermann A Haus. Waves and fields in optoelectronics. Prentice-Hall,, 1984.
[166] David M Pozar. Microwave engineering. John Wiley & Sons, 2009.
[167] Mehmet Kara. Formulas for the computation of the physical properties of
rectangular microstrip antenna elements with various substrate thicknesses.
Microwave and Optical Technology Letters, 12(4):234–239, 1996.
[168] Randy Bancroft. Microstrip and printed antenna design. The Institution of
Engineering and Technology, 2009.
Bibliography 268
[169] Width and effective dielectric constant equations for design of

microstrip transmission lines, Rogers Corporation. [Online].
Available: [https://www.rogerscorp.com/documents/783/acs/
Design-Data-for-Microstrip-Transmission-Lines-of-TMM-Laminates.
pdf], (accesseed 23 Mar. 2016).
[170] Harold A Wheeler. Transmission-line properties of parallel strips separated by
a dielectric sheet. IEEE Transactions on Microwave Theory and Techniques,
13(2):172–185, 1965.
[171] Erik O Hammerstad. Equations for microstrip circuit design. In Microwave
Conference, 1975. 5th European, pages 268–272. IEEE, 1975.
[172] Keith Carver and James Mink. Microstrip antenna technology. IEEE transactions
on antennas and propagation, 29(1):2–24, 1981.
[173] Copper foils for high frequency circuit materials, Rogers Corporation.
[Online]. Available: [https://www.rogerscorp.com/documents/749/acs/
Copper-Foils-for-High-Frequency-Circuit-Materials.pdf], (accessed 28
Mar. 2018).
[174] Henning Braunisch, Xiaoxiong Gu, Alejandra Camacho-Bragado, and Leung
Tsang. Off-chip rough-metal-surface propagation loss modeling and correlation
with measurements. In Electronic Components and Technology Conference, 2007.
ECTC’07. Proceedings. 57th, pages 785–791. IEEE, 2007.
[175] Scott Hinaga, M Koledintseva, P Anmula, and J Drewniak. Effect of conductor
surface roughness upon measured loss and extracted values of PCB laminate
material dissipation factor. In Proc. Tech. Conf. IPC Expo/APEX 2009, pages
S20–2, 2009.
[176] Kai Fong Lee and Wei Chen. Advances in microstrip and printed antennas. Wiley,
1997.
[177] Ramesh Garg. Microstrip antenna design handbook. Artech house, 2001.
[178] Ghulam Ahmad, Tim WC Brown, Craig I Underwood, and Tian H Loh. An
investigation of millimeter wave reflectarrays for small satellite platforms. Acta
Astronautica, 2018.
[179] James R James and Peter S Hall. Handbook of microstrip antennas, volume 28.
IET, 1989.
[180] WF Richards, YT Lo, P Simon, and DD Harrison. Theory and applications for
microstrip antennas. In Proc. Workshop on Printed Circuit Antenna Technol.,
pages 8–1, 1979.
[181] RJ Collier and PD White. Surface waves in microstrip circuits. In Microwave
Conference, 1976. 6th European, pages 632–636. IEEE, 1976.
[182] David M Pozar. Rigorous closed-form expressions for the surface wave loss of
printed antennas. Electronics Letters, 26(13):954–956, 1990.
[183] Ghulam Ahmad, Craig Underwood, Tim Brown, and TH Loh. Role of surface
waves in improving the reflection properties of a millimetre wave reflectarray unit
cell. LAPC 2017 Proceedings, 2017.
[184] Ghulam Ahmad, Tim Brown, Craig Underwood, and TH Loh. Millimetre wave
reflectarray antenna unit cell measurements. LAPC 2017 Proceedings, 2017.
Bibliography 269
[185] WE Doherty and RD Joos. The PIN diode circuit designers handbook. Microsemi
Corporation, 1:1–137, 1998.
[186] Agilent 4395A, Network/Spectrum/Impedance Analyzer Operation Manual, 6th
ed. Agilent Technolgies Japan, Feb. 2008.
[187] T Boles, J Brogle, D Hoag, and D Carlson. AlGaAs anode heterojunction PIN
diodes. Physica status solidi (c), 10(5):786–789, 2013.
[188] T Boles, J Brogle, D Hoag, and D Curcio. AlGaAs PIN diode multi-octave, mmW
switches. In Microwaves, Communications, Antennas and Electronics Systems
(COMCAS), 2011 IEEE International Conference on, pages 1–5. IEEE, 2011.
[189] Jian Song, Jun Wang, Kewu Peng, Changyong Pan, and Zhixing Yang.
Quantization error reduction for the phased array with 2-bit phase shifter.
Wireless personal communications, 52(1):29, 2010.
[190] Siew Bee Yeap, Xianming Qing, and Zhi Ning Chen. 77-GHz dual-layer
transmit-array for automotive radar applications. IEEE transactions on antennas
and propagation, 63(6):2833–2837, 2015.
[191] RF E’qab and Derek A McNamara. Angle of incidence effects in reflectarray
antenna design: Making gain increases possible by including incidence angle
effects. IEEE Antennas and Propagation Magazine, 58(5):52–64, 2016.
[192] Ang Yu, Fan Yang, Atef Z Elsherbeni, John Huang, and Yahya Rahmat-Samii.
Aperture efficiency analysis of reflectarray antennas. Microwave and Optical
Technology Letters, 52(2):364–372, 2010.
[193] John Huang. Analysis of a microstrip reflectarray antenna for microspacecraft
applications. Technical report, 1995.
[194] Payam Nayeri, Atef Z Elsherbeni, and Fan Yang. Radiation analysis
approaches for reflectarray antennas. IEEE Antennas and Propagation Magazine,
55(1):127–134, 2013.
[195] Min Zhou. Accurate modeling of advanced reflectarrays. PhD thesis, Technical
University of Denmark (DTU), 2013.
[196] Ming-Tao Zhang, Steven Gao, Yong-Chang Jiao, Ji-Xiang Wan, Bu-Ning Tian,
Chun-Bang Wu, and Andrew-John Farrall. Design of novel reconfigurable
reflectarrays with single-bit phase resolution for Ku-band satellite antenna
applications. IEEE Transactions on Antennas and Propagation, 64(5):1634–1641,
2016.
[197] Ghulam Ahmad, Tim WC Brown, Craig I Underwood, and Tian Hong Loh.
How coarse is too coarse in electrically large reflectarray smart antennas? In
Electromagnetics: Applications and Student Innovation Competition (iWEM),
2017 International Workshop on, pages 135–137. IEEE, 2017.
[198] Ghulam Ahmad, Tim WC Brown, Craig I Underwood, and Tian Hong Loh. An
efficient algorithm for electrically large reflectarray antenna design automation.
In Electromagnetics: Applications and Student Innovation Competition (iWEM),
2017 International Workshop on, pages 133–134. IEEE, 2017.
[199] Ghulam Ahmad, Tian H Loh, Tim WC Brown, and Craig I Underwood. On
the phase selection of millimeter wave quantized reflectarrays. In Applied
Bibliography 270
Computational Electromagnetics Society Symposium (ACES), 2017 International,

pages 1–2. IEEE, 2017.
[200] Agilent Technologies 85325A R/Q/U/V/W millimeter subsystem: Operating
and service manual. [Online]. Available: [https://literature.cdn.keysight.
com/litweb/pdf/85325-90061.pdf?id=1000002822-1:epsg:man], (accessed 28
Mar. 2018).
[201] Kai Zhang, Yangyu Fan, Jiadong Xu, and Chen Qu. Design of broadband, low
cost single layer reflectarray using phoenix cell. In TENCON 2013-2013 IEEE
Region 10 Conference (31194), pages 1–4. IEEE, 2013.
[202] Sarah Adel and Hany Hammad. Modified phoenix cell for microstrip reflectarray
antennas. In Antennas and Propagation (MECAP), 2012 Middle East Conference
on, pages 1–3. IEEE, 2012.
[203] Sean V Hum, G McFeetors, and M Okoniewski. Integrated MEMS reflectarray
elements. In Antennas and Propagation, 2006. EuCAP 2006. First European
[204] R Sorrentino, R Vincenti Gatti, and L Marcaccioli. Recent advances on millimetre
wave reconfigurable reflectarrays. In Antennas and Propagation, 2009. EuCAP
2009. 3rd European Conference on, pages 2527–2531. IEEE, 2009.
[205] Julien Perruisseau-Carrier and Pablo Pardo. Unit cells for dual-polarized and
polarization-flexible reflectarrays with scanning capabilities. In Antennas and
IEEE, 2009.
[206] Chinthana J Panagamuwa, Alford Chauraya, and JC Vardaxoglou. Frequency
and beam reconfigurable antenna using photoconducting switches. IEEE
[207] Y Tawk, Alex R Albrecht, S Hemmady, Gunny Balakrishnan, and Christos G
Christodoulou. Optically pumped frequency reconfigurable antenna design. IEEE
antennas and wireless propagation letters, 9:280–283, 2010.
[208] Y Tawk, J Costantine, SE Barbin, and CG Christodoulou. Integrating laser
diodes in a reconfigurable antenna system. In Microwave & Optoelectronics
Conference (IMOC), 2011 SBMO/IEEE MTT-S International, pages 794–796.
IEEE, 2011.
[209] G Perez-Palomino, M Barba, JA Encinar, R Cahill, R Dickie, and P Baine.
Liquid crystal based beam scanning reflectarrays and their potential in SATCOM
antennas. In Antennas and Propagation (EUCAP), 2017 11th European
[210] MY Ismail and M Inam. Liquid crystal based reconfigurable reflectarray antenna
design. World Academy of Science, Engineering and Technology, International
Journal of Electrical and Computer Engineering, 3(1), 2016.
[211] Daniel Rodrigo, Luis Jofre, and Julien Perruisseau-Carrier. A frequency
reconfigurable cell for beam-scanning reflectarrays. In Antennas and Propagation
Society International Symposium (APSURSI), 2013 IEEE, pages 1666–1667.
IEEE, 2013.
[212] Tomislav Debogovic and Julien Perruisseau-Carrier. MEMS-reconfigurable
Bibliography 271
metamaterials and antenna applications. International Journal of Antennas and

Propagation, 2014, 2014.
[213] Julien Perruisseau-Carrier and Anja K Skrivervik. Monolithic MEMS-based
reflectarray cell digitally reconfigurable over a 360 degree phase range. IEEE
[214] Tomislav Debogović, Srećko Sušac, and Julien Perruisseau-Carrier. A simple
reflectarray cell with 1-bit phase control and polarization flexibility. In Antennas
and Propagation (EuCAP), 2013 7th European Conference on, pages 2717–2720.
IEEE, 2013.

Im 2

Uploaded by

Copyright:

Available Formats

Im 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Im 2

Uploaded by

Copyright:

Available Formats

Novel Design and Measurement Methodologies of

Millimeter Wave Smart Antennas

Submitted for the Degree of

Surrey Space Center

Email: g.ahmad@surrey.ac.uk; gasajid48@gmail.com

2 Background to Millimeter Wave Beamsteering and Reflectarrays 13

2.8.2 Phase Control by Electrical/Electronic Means . . . . . . . . . . . 39

3 Characterization of Passive Reflectarray Unit Cells 56

4 Characterization of Active Reflectarray Unit Cells 97

4.3 Active Unit Cell Characterization . . . . . . . . . . . . . . . . . . . . . . 117

5 Reflectarray Design and Analysis 144

6 Reflectarray Measurements 194

7 Conclusion and Future Research 226

A Passive and Active Unit Cells 234

B Unit Cell Reflection Phase and DC Biasing 252

1.1 High gain antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1 Millimeter wave spectrum. Currently under consideration mm-wave

3.1 A reflectarray unit cell (front and side views). . . . . . . . . . . . . . . . 60

3.4 Quality factors Qo and Qext as a function of the substrate thickness.

3.20 Reflection coefficient’s (Γ) phase versus frequency (f ) of a square patch

4.6 Characterization of MACOM PIN diodes MA4AGBLP912 and

5.1 Co-ordinate system for reflectarray analysis. . . . . . . . . . . . . . . . . 147

5.22 Normalized directivity loss in reflectarrays with scan angle simulated in

6.1 A 3 states phase quantized reflectarray at 0◦ and 55◦ beam pointing

6.8 Response of a V band 3 states phase quantized reflectarray designed

6.21 Measured 3D radiation pattern at 60 GHz of a 3 states phase quantized

A.1 Various types and shapes of passive unit cells . . . . . . . . . . . . . . . 235

B.1 Continuous reflection phase required from unit cells in a reflectarray of

2.1 Unlicensed mm-wave band and maximum transmit power in various

3.1 A comparison of unit cell measurement techniques/structures. . . . . . . 92

4.1 A comparison of pin diode switches . . . . . . . . . . . . . . . . . . . . . 108

5.1 Reflectarray formulation definitions . . . . . . . . . . . . . . . . . . . . . 148

6.1 Passive demonstrators of reflectarrays at fixed beam pointing angles. (F

6.2 A comparison of CST simulated and measured results for passive

A.1 A comparison of passive phase control techniques . . . . . . . . . . . . . 239

PCB Printed Circuit Board

A continuous quest to achieve unprecedented extremely high data transfer rates in

At mm-waves the communication path loss increases rapidly. Therefore, to maintain

A rapid growth of the wireless communication market, the advent of internet of

bandwidths. A continuously increasing high resolution demand in observation systems

To overcome these propagation losses, mm-waves systems demand high values of

(a) Satellite antennas (b) FAST observatory (c) ALMA observatory

(d) Phased array radar (e) Phased arrays control

Figure 1.1: High gain antennas

1000s of Unit Cells

Printed Reflectarray Surface

(a) Reflectarray geometry (b) Folded reflectarray on a cube sat

Figure 1.2: Reflectarray antennas

Another attractive advantage of reflectarrays is their ability to be made foldable [21] to

In spite of these disadvantages, reflectarrays are still considered as a potentially

reflectarrays and smart reflectarrays are used interchangeably.

1.2 Aims and Objectives of Research

O2: To comprehensively examine the available reconfiguration technologies and select

O3: Characterization of the selected reconfiguration technology at mm-waves.

O5: Design novel solutions suitable for mm-waves smart antennas.

O6: Implementation and measurements of novel designs to demonstrate the proposed

O9: To propose considerations to economize power consumption of the selected

1.3 Major Research Contributions

Research work presented in this thesis brings following contributions.