Radio Arch SP Short 2pp

Radio Architectures and Signal Processing - short pp. 1 / 84 Radio Architectures and Signal Processing - short pp.
2 / 84
M. Valkama & M. Renfors, TUT/DCE 4 October 2011 M. Valkama & M. Renfors, TUT/DCE 4 October 2011
Some Contents
1. Background and introduction
2. Basic receiver architectures and their properties
An Introduction to - Main components; frequency translations and filtering
- Receiver architectures: superhet, direct conversion, low IF
- Mirror-frequency considerations and effects of I/Q
Radio Architectures and imbalance
- Non-idealities and performance measures of the analog
Signal Processing front-end: sensitivity, noise figure, intermodulation, phase
noise
3. A few notes about transmitters and receiver-transmitter
Mikko Valkama & Markku Renfors co-operation
4. Sampling of communications waveforms
Department of Communications Engineering - Sampling theorem for complex signals
Tampere University of Technology, Finland - Sampling alternatives
mikko.e.valkama@tut.fi, markku.renfors@tut.fi - Non-idealities:quantization noise and aperture jitter
5. Increasing to role of DSP in receiver front-ends
- Receiver architectures for software radio
- RF challenges in cognitive radio
6. Dirty-RF signal processing (separate material)
- Behavioral modeling of RF impairments
- DSP methods to reduce the effects of RF impairments in
radios
- Case studies
Radio Architectures and Signal Processing - short pp. 3 / 84 Radio Architectures and Signal Processing - short pp. 4 / 84
Short Overview Principal Transmitter Functionalities
Radio transmitters and receivers are essentially I

interfaces between digital data (bits) and D/A LPF
electromagnetic waves. RF PA
BITS IN
DSP I/Q LO
This can be further divided to front-end processing
and baseband processing. Q
(a) D/A LPF
DIGITAL BB/IF ANALOG BB/IF ANALOG RF
- RF/IF PROCESSING
BASEBAND PROCESSING Digital BB/IF
- CHANNEL EQUALIZATION
- DOWN-CONVERSION
T/R - I/Q DEMODULATION
- DETECTION
- CHANNEL DECODING
- baseband waveform generation (coding,
- SAMPLING
- SOURCE DECODING constellation mapping, pulse-shaping, etc.)
SYNCHRONIZATION - sample rate conversions
- possibly frequency translation to an
So in addition to antennas, the front-end includes intermediate frequency (IF)
- Amplification stages
- Frequency translation stages Analog BB/IF/RF
- Filtering stages - further bandlimitation (if needed)
- Sampling and A/D/A interface - convert the signal to the final RF range,
possibly through an IF stage
In practice, the front-end usually includes both analog
and digital signal processing stages (not only analog). - power amplification and final band-limitation
Different radio architectures then mean how the

above functionalities are organized in the radio chain.
Principal Receiver Functionalities Focus in this presentation

AGC
Here we focus on the basics and fundamentals of
LPF A/D
I frequency translations, filtering and sampling in
RF LNA
radios.
BITS OUT
BPF I/Q LO DSP
AGC
Also different radio architectures, like
(b)
A/D
Q superheterodyne, direct-conversion and low-IF, are
LPF
reviewed, and how the above functionalities are
ANALOG RF ANALOG IF/BB DIGITAL IF/BB
deployed in them.
Analog RF/IF/BB Anothe central theme is related to different

- translates the RF signal to lower frequencies nonidealities of the most essential circuit modules
- also implements some band-limitation and (amplifiers, mixers, oscillators, samplers, ADC’s), and
amplification how they affect the radio.
- typically part of selectivity In general, we focus mostly on the receiver side,
since the challenges are somewhat bigger compared
Digital IF/BB to transmitters
- final selectivity and frequency translation to BB - detecting the weak desired signal(s) in the
(if not done already in analog) presence of much stronger (easily tens of dB’s)
- sample rate conversions neighboring signals
- channel equalization, detection and decoding
Notice: Interactive demonstrations and additional
(supporting) material available at
Also extracting synchronization information (carrier &
timing) is one central element on the receiver side. http://bruce.cs.tut.fi/invocom/index.htm
Essential System Characteristics Main Targets in Radio Parts in Wireless

Communication Systems
The essential system characteristics effecting on the
front-end design are:
1. Terminals
- Signal bandwidth
o 25 kHz in DAMPS  small-size
o 200 kHz in GSM  low-cost
o 1.25 MHz in IS-95
 low power consumption
o ~5 MHz in UMTS/WCDMA
o 8 MHz in DVB-T  multiband & multimode capabilities
o Up to 20 MHz in 3GPP-LTE and WiMAX
o 25 MHz in 802.11a/g 2. Base-stations
o in the order of 100 MHz in emerging 4G systems
 system performance may not be compromised.
- Signal-to- Interference-Plus-Noise -ratio needed to
 size and cost also important
detect the signal properly.
 several parallel RX/TX systems typically in
- Dynamic range, depending on minimum desired operation simultaneously (multiradio BS).
signal level (receiver sensitivity) and the maximum
level of other signals to be tolerated in nearby
frequencies in the radio. On both sides, multi-antenna techniques coming to or
Relatively small distortion effects in a strong non- already in practical use
desired signal component may cause huge relative  The need for multiple RF chains increases the
distortion for a weak desired signal!! importance of cost and size aspects.
And due to technological limitations, the weak  RF impairment effects in multi-antenna systems
desired signal can be separated form the adjacent need to be carefully examined.
spectral components only at a rather late stage in
the processing chain.
Key Areas of Development Possible Solutions

 Analog and digital VLSI technologies
Regarding Multimode Capabilities
 MEMS (micro electro-mechanical systems) Supporting multiple radio access technologies in a

technologies single device (like mobile phone or BS):
 Packaging  Separate highly integrated radios for different

systems
 Analog RF-ASIC design (innovative circuit
topologies, etc.) or
 High-speed & low-power DSP architectures  Configurable HW platform, possibly with separate
analog front-ends for different frequency bands
 DSP algorithms
 Energy efficiency at all levels It helps if most of the radio functionality is defined in
DSP software or easily reconfigurable DSP hardware
 Analog/DSP-tradeoffs: role of DSP increasing
o Relatively low progress of ADC technologies as => Software Radio: a single hardware solution
the main bottleneck for increased use of DSP adaptable to different system standards by
changing software and/or reconfiguring digital HW
o Place for innovative RX/TX architectures
o Discrete-time analog processing is also an also called
interesting possibility (e.g. DRP by TI)
Software configurable radio
 Dirty RF: There are possibilities to relax the analog
Software defined radio
component requirements by compensating analog
Flexible radio
distortion effects through advanced DSP techniques
Practical solutions are compromises of the above, at
least so far.
Cognitive Radio Concepts

Contents
With a spectrum analyzer, it is easy to observe that 1. Background and introduction
the licensed bands of the radio spectrum are not in a 2. Basic receiver architectures and their properties
very efficient use. For example, there are gaps in the - Main components; frequency translations and filtering
spectrum which are seldom used, some of the - Mirror-frequency considerations and effects of I/Q
services are not operating all the time, etc. imbalance
The terms front-end: sensitivity, noise figure, intermodulation, phase
noise
o cognitive radio
3. A few notes about transmitters and receiver-transmitter
o flexible spectrum use
co-operation
o spectrum pooling
refer to ideas of taking most parts of the radio - Sampling theorem for complex signals
spectrum into more efficient use by relaxing the strict - Sampling alternatives
allocation of different services to different frequency - Non-idealities:quantization noise and aperture jitter
bands and allowing opportunistic secondary use of 5. Increasing to role of DSP in receiver front-ends
the empty parts of the spectrum, using flexible - Receiver architectures for software radio
communications waveforms that would be most - RF challenges in cognitive radio
suitable for a particular environment. 6. Dirty-RF signal processing (separate material)
For implementing the cognitive radio ideas, effective - DSP methods to reduce the effects of RF impairments in
radios
software radio type of transceiver implementation is
- Case studies
vital.
What is needed in the receiver front-end? Main Components for Receiver Front-Ends
 Amplification to compensate for transmission losses  Amplifiers

- Low-noise amplifiers (LNAs) in the first stages.
 Selectivity to separate the desired signal from others
- Automatic gain control (AGC) needed to cope with
 Tunability to select the desired signal different signal levels.
AGC
 Conversion to digital domain I
LPF A/D
RF LNA
BITS OUT
In the following, we examine different receiver architectures BPF I/Q LO DSP
and non-idealities affecting in the different building blocks AGC

(b) Q
used in the receivers. LPF A/D
ANALOG RF ANALOG IF/BB DIGITAL IF/BB
 Filters
- Impossible to achieve sufficient selectivity by
tunable RF filters (operating in the RF frequency
band of the modulated signal) to separate the
desired signal from others.
- Sufficient selectivity can be achieved by fixed (RF or IF)
filters based on special technologies (SAW,
Surface Acoustic Wave, ceramic, crystal,
mechanical)
or analog filters operating on baseband or relatively
low bandpass center frequencies
Main Components for Receiver Front-Ends Main Components for Receiver Front-Ends
(...cont’d...)  Mixers
or multirate digital filters up to some hundreds of MHz - Complex (I/Q, quadrature) mixer: pure frequency
range. translation by the local oscillator frequency:
- Special complex filters, phase splitters (related also
to Hilbert transformers) can be used to suppress Special case with
certain frequency range from the negative part of real input signal:
the frequency axis. Such filters find application in e j

LO
t
I I
certain special receiver architectures. cos( LO t)
f
fc
sin( LOt)
f I Q
fc+fLO
- Real mixer produces the combination of frequency

translations in both directions:
f
fc 0 fc
cos(LO t)
f
fcfLO fcfLO 0 fcfLO fcfLO
More details in the separate material ! (earlier

complex signals and systems material)
Main Components for Receiver Front-Ends Classical Receiver Architecture:

The Superheterodyne
 Oscillators
- Voltage (or current or digitally) controlled oscillators AGC
(VCO, ICO, DCO) are used to generate the local RF-STAGES MIXER IF-STAGES
oscillator (LO) signals in a tunable manner.
- In a communications transceiver (receiver+transmitter,
RX+TX), the frequency synthesizer is one of the main
blocks. It is used for generating all the needed LO LO
signals in a controllable manner.
RF-FILTER
 Analog-to-digital interface
0 fRF fLO fRF +2fIF
- Various ADC technologies, like Flash, Successive
Approximation and Sigma-Delta
+fLO fLO
- Key performance metrics:
o Number of bits / Dynamic range / SNR
IF-FILTER
o Sampling jitter effects
0 fIF=fLOfRF
- Main bottleneck in advanced DSP-based architectures.
- Room for innovative solutions Example: One common choice in GSM900 receivers has
been 1st IF = 71 MHz, 2nd IF = 13 MHz
Majority of all the receivers have been based on the

superheterodyne principle in the past.
Filtering Requirements in Superheterodynes Some Drawbacks of the Superheterodyne

Architecture
Selectivity is achieved at the IF stage(s) working at fixed
center frequency using special filter technologies. Some parts are difficult to integrate
 The RF filter should provide sufficient attenuation for the  IF-filter

image band at the distance of 2xfIF in frequency.  RF-filter
 Oscillators
 The final IF stage should have sufficient selectivity to
suppress the neighbouring channels sufficiently. Power consumption high
 In case of double (or triple) super heterodyne, the first  External components => parasitics
(and second) IF stage should provide enough attenuation  Several submodules => low impedance (e.g., 50 )
at twice the next IF frequency. levels used for matching the modules
 Image reject (I/Q-) mixer is one possibility to reduce the
RF filter requirements (but not sufficient as the only Complicated structure
solution)
 There is great interest for simpler architectures
which could be integrated more easily.
Spurious responses
 LO and IF signals and harmonics and mixtures leaking to

different places may cause problems.
Direct-Conversion Receiver Architecture Direct-Conversion Receiver Architecture

TUNED Advantages
TO RF I
 No image bands, or the image is the signal itself at
negative frequencies => RF-filtering not so critical in this
LO LOWPASS sense
CHANNEL
FILTERS  Not so much spurious responses
90°
 Simple structure, no IF filters
 Selectivity can be easily partitioned between analog and
Q
digital filters
 If most selectivity in analog parts, low requirements for
ZERO IF sampling and ADC blocks
“Zero IF” –principle, simply I/Q down-convert the target Problem: Difficulties in implementation: dc offsets, second-
radio channel directly to baseband. order intermodulation, leakage between RX and TX in full
duplex operation
Direct-conversion principle has become quite popular

in recent years in mobile terminals, and is also one
f
widely-used solution in base-stations!
f
DC-Offsets in Direct-Conversion Receivers Low-IF Receiver Architecture

The idea is to use I/Q down-conversion and a low IF frequency
DC-offsets appear mainly due to finite isolation between which is just high enough to cope with the DC-offset problem
the mixer RF and LO ports: (e.g., 250 kHz in case of GSM).
SELF-MIXING
OF INTERFERER
f
0
fLO
LO SELF-MIXING LO
f
Constant DC-offset can be compensated by measuring it fIF 0 fIF A COMPLEX
BANDPASS
without signal and then subtracting it during reception. FILTER
The channel filtering and final demodulation can be done after

In TDMA systems, different channels/bursts may have A/D-conversion in the DSP domain. Channel selectivity can also
different signal levels and different AGC-values and hence partitioned between analog and digital front-ends.
different DC-offsets => compensation is difficult. As we shall see on the next pages, I/Q down-conversion in the
analog domain may not, in practice, provide sufficient attenuation
Also 1/f -type of noise appearing in active components may for the image band.
be a problem. This effect also appears at DC and low More attenuation can be obtained by using a phase splitter
frequencies. attenuating the image band on the negative part of the frequency
axis.
Other issues Even more attenuation could possibly be achieved through
baseband digital signal processing.
Signal leakage from antenna to the surroundings takes On the other hand, the low-IF concept is facilitated by the fact that
place more easily than in superhet. system specifications (like GSM) don't allow the maximum signal
level to appear in the nearest adjacent channels in case of a very
Sensitivity to 2nd-order intermodulation is another drawback low desired signal level.
that will be discussed later in more details.
Low-IF Receiver by Steyert et al. About Analytic Bandpass Signals

Real bandpass signal and the corresponding ideal analytic
I bandpass signal (obtained ideally through Hilbert
MIRROR I transform):
SIGNAL
PASSIVE SUPPRESSION
POLYPHASE +
FILTER FINAL Q f
Q DOWN-
CONVERSION
0
W
FILTERING 
I Q f
0
HF LO
Other side of the desired signal spectrum suppressed by a

 
I in I out practical phase splitter with finite attenuation (e.g., for
improving image attenuation in case of I/Q down-
 
Q in Q out conversion):
 
I in I out
f
0
 
Q in Q out

In this architecture, the image frequencies are suppressed

f
25 … 30 dB by the phase splitter (Hilbert) implemented as 0 fc
a polyphase RC network, and another 25 … 30 dB by
the I/Q downconversion approach. To realize such a system, it is sufficient to design an
allpass filter which approximates 90° phase shift in the
In this case, full complex mixer (4 real mixers) are needed passband and stopband with sufficient accuracy,
as illustrated in the block-diagram. depending on the stopband attenuation requirement.
About Image Rejection as a Function of Gain and

Gain and Phase Imbalance in I/Q Systems Phase Imbalance
Let’s take a single complex exponential
Quadrature I/Q downconversion is trying to produce a pure
frequency translation which would suppress the image cos ct  j sin ct   e jct
band completely.
as a starting point.
AGC
I
Assuming now that g is the gain imbalance ratio and  is
LPF A/D the phase difference due to imbalance, we can write:
RF LNA
BITS OUT
BPF I/Q LO DSP cos ct  jg sin ct   
AGC
(b) Q
 ...
LPF A/D
1  ge j 1  ge j
ANALOG RF ANALOG IF/BB DIGITAL IF/BB  e jct  e jct
2 2
In practice, there is some mismatch (imbalance) of gain From the latter form, we can identify that due to imbalance,
and/or phase in the components involved (oscillator, the ideal exponential is now split to two mirror exponentials
amplifiers, mixers).
=> this means cross-talk between the mirror or image
Consequently, the image suppression is, in practice, far frequencies in general
from complete.
The ratio of the image and desired signal powers is
obtained as (image rejection ratio, IRR)
2
1  ge j
2 1  g 2  2 g cos 
R2  
1  ge j
2
1  g 2  2 g cos 
2
Image Rejection as a Function of Gain and Examples of Mirror-Frequency Crosstalk with

Phase Imbalance Different Signal Types
10
IDEAL SIGNAL IDEAL SIGNAL IDEAL SIGNAL IDEAL SIGNAL
Amplitude imbalance in dB SPECTRUM SPECTRUM SPECTRUM SPECTRUM
0
-10
f f f f
Image attenuation [dB]
2
-20 (a) 0 (b) 0 (c) 0 (d) 0
1
I/Q IMBALANCED I/Q IMBALANCED I/Q IMBALANCED I/Q IMBALANCED
-30 0.5 SIGNAL SPECTRUM SIGNAL SPECTRUM SIGNAL SPECTRUM SIGNAL SPECTRUM
0.2
-40
0.1
0.05
-50
f f f f
0 0 0 0
0
-60
-70
-1 0 1 2
10 10 10 10
Phase imbalance in degrees
This can be applied to all cases of quadrature mixing where

gain and/or phase imbalance appears (dimensioning).
Frequency-Dependent Imbalance Model Effects of Gain and Phase Imbalance in

Different Architectures
Especially in wide-band systems, the imbalance
parameters may be frequency dependent and can be In case of direct conversion-receiver (or final demodulation of an
written as g(f) and (f) I/Q-signal), gain mismatch and phase errors cause “self images”:
Then also the image suppression ratio is frequency-

dependent:
f
1  g ( f )2  2 g ( f )cos  ( f ) 0
R( f )2 
1  g ( f )2  2 g ( f )cos  ( f ) +fLO fLO
Measured example:
f
40 0
39 This is usually not a problem with low-order modulations, for
which an image attenuation of, e.g., 20 dB doesn’t essentially
38
effect the system performance. For high-order modulations, this
37 issue is more critical.
In other cases of I/Q down-conversion, like the low-IF receiver, the
Attenuation [dB]
36
image signal may be at a considerably stronger level (up to 100
35 dB!!) than the desired signal, and I/Q imbalance is very critical:
34
33
32 f
0
31
fLO fLO
A COMPLEX
30 BANDPASS
−8 −7 −6 −5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8 FILTER
Frequency [MHz]
f
fIF 0 fIF
Selectivity Requirements in System Example: GSM Interference Mask

Specifications
GSM Interferer Mask
Radio system specifications (e.g., for cellular systems) −20
don't allow a strong adjacent channel signal to be present −30
when a weak desired signal is to be received. (Radio

−40
Resource Management functionalities of wireless systems
take care of this!) −50
Signal Power [dBm]

−60
For example, the GSM specifications give:

−70
 The maximum levels for the 3 adjacent channels on both −80
sides (at 200, 400 and 600 kHz from the carrier) in case
−90
of a GSM interferer and desired signal 20 dB above the
← Desired Signal Level for the Blocking Test
reference sensitivity level of -102 dBm. −100
← Reference Sensitivity Level
 Maximum levels for more distant signals (>600 kHz from

−110
carrier), blocking signals, in case of a sinusoidal interferer −120

−5000 −4000 −3000 −2000 −1000 0 1000 2000 3000 4000 5000
Frequency Offset from the Carrier Frequency [kHz]
and desired signal 3 dB above the reference sensitivity
level.
These are usually stemming from linearity constraints but
help also in the image problem, especially when the IF is
reasonably low.
Summary: Essential Non-Idealities and Performance

Techniques for Providing Image Rejection in Measures of the Analog Front-End
Different Architectures Distortions in a receiver front-end are mostly caused by
1. RF, first IF filters - Nonlinearities in the front-end that distort the desired
signal and create new spectral components as
- Challenges to get sufficient performance in integrated
solutions.
harmonic or intermodulation products of the various
spectral components of the received signal.
- Not applicable in low-IF or direct conversion cases.
- Insufficient attenuation of spectral components that are
2. Quadrature down-conversion translated in the spectrum due to imaging or aliasing
- Mostly for low-IF and direct conversion cases. caused by the (inevitable) sampling and A/D-
- Utilized sometimes also in superheterodynes ("image reject conversion stages.
mixer") to simplify the RF filter. - Noise produced by the analog stages, sampling
- Rejection limited by gain and phase imbalance. process, and A/D-conversion.
- Produces complex signal and the consecutive signal
processing blocks must be duplicated, i.e., implemented for Depending on the system specifications, the quality of
both for I and Q branches. the front-end should be good enough in two senses:
3. Phase splitter (passive polyphase RC network) 1. The possible distortion in the desired signal (without
other signals) itself should not cause any significant
- Mostly for low-IF and direct conversion cases, but can be
used also in superhets.
degradation of bit-error-rate (BER) in detection.
- Produces complex signal and the consecutive signal 2. The distortions in the front-end should not cause new
processing blocks must be duplicated. interfering signal components to appear in the signal
band that would significantly affect the bit-error-rate
4. Advanced baseband digital signal processing (BER) in detection.
- If the imbalance parameters can be estimated, the effects
can be compensated fairly well. In recent years, various
effective and practical solutions have been developed for this
purpose.
Essential Non-Idealities and Performance Essential Non-Idealities and Performance

Measures of the Analog Front-End Measures of the Analog Front-End
Earlier low-order modulations, like QPSK, GMSK, or /4  Sensitivity of the receiver is mainly determined by the
QPSK, were used in wireless communications, and the first noise produced by the receiver front-end components. It
criterion was not very critical. However, increasingly higher- determines the minimum detectable signal in noise-
order modulations have been taken into use limited situation. In general:
 8PSK in Edge Receiver sensitivity = thermal noise power (dBm)
 16QAM, e.g., in WCDMA/HSDPA + noise figure (dB)
+ implementation loss (dB)
 64QAM, e.g., in DVB-T, 802.11a/g, WiMAX, 3GPP-LTE
+ required SNR (dB)
In room temperature:
=> Bigger and bigger demands for the quality of receiver
Thermal noise power (dBm) = -174 dBm+10log10B
front-end processing.
where B is the equivalent noise bandwidth of the RX.
For example in GSM, minimum S/N is 9 dB, B=200 kHz,
and required sensitivity is -102 dBm => NF<10 dB.
 With low received signal levels (as above), the receiver
front-end components can be assumed to be linear. With
higher signal levels, the nonlinearity of the amplifiers and
other components produce harmful intermodulation
products. In this way the nonlinearity limits the dynamic
range of received signals from above.
Intermodulation is measured by 1 dB compression point
or the so-called IP3 figure.
In RF circuit design, there are always tradeoffs between
noise figure, linearity, and power consumption.
N
Essential Non-Idealities and Performance Noise Figure

Measures of the Analog Front-End The noise factor of an amplifier stage (or some other
 Frequency accuracy and stability are determined by component) is determined by the ratio of S/N
the local oscillators of the receiver. In systems like GSM, ratios at the input and output:
the receiver is locked to the network, and the frequency SNRin
stability is very good. However, to guarantee that the F
receiver is able to synchronize to the network, certain SNRout
frequency accuracy, stability, and settling time The noise figure is:
requirements are set for the components. NF  10 log10 F .
The short term instability of the oscillators appear as
phase noise, and it is very critical for the performance For a cascade of n stages, the overall noise factor is
of the system.
F 1 F3  1 Fn  1
 Leakage effect means that strong signals, especially FT  F1  2  
g1 g1g 2 g1g 2  g n 1
local oscillator signals are connected, e.g., through
spurious capacitances to places where they are not where Fi and gi are the noise factor and power gain of
supposed to be connected. This means that various stage i.
harmonics, subharmonics, and mixtures of the local
oscillator frequencies are usually added to the signal. Typically the noise figure of the first components (before
significant amplification) is most critical in receivers,
In receiver design, the frequencies of the strongest
because the target signal level is weakest in those stages.
spurious frequencies can be calculated. By selecting the
local oscillator frequencies properly, most of the spurious
frequencies can be placed outside the desired frequency
bands at RF and IF.
 In the case of analog I/Q signal processing, amplitude
and phase responses of the I and Q branches are never
exactly the same. The effects of the gain and phase
imbalance depend greatly on the receiver architecture.
Intermodulation Intercept points - IP2, IP3

Consider a test where there are two nearby frequencies f 1 The device-specific IP2, IP3-values can be used to
and f 2 in the system frequency band (like in the determine the strength of second- and third-order
neighbouring channels). intermodulation products, in reference to the fundamental
signal component.
In general, nonlinearities produce intermodulation
frequencies of the form IP3
k1 f1  k2 f 2 1dB
OUTPUT LEVEL, dBm

where k1 , k2 are integers.
L
TA
T
UC
EN
DYNAMIC RANGE
M
SPURIOUS-FREE
Third-order intermodulation produces (among some others)
OD
A
ND
PR
FU
frequencies 2 f1  f 2 , 2 f 2  f1 which may easily fall in
IM
ER
RD
the desired signal band.
D-O
3R
INTERCEPT POINT (IP3)

COMPRESSION
INPUT-REFERRED
INPUT LEVEL, dBm
POINT
f
2f1 f2 f1 f2
Second-order intermodulation produces frequencies The overall IP2 and IP3 values for a cascade of stages can
f1  f 2 , f 2  f1 . be calculated in a fairly straightforward way from IP values
of the components.
In general, with low-enough signal levels, the levels of
second-and third-order intermodulation products are These calculations are mostly relevant in the first receiver
proportional to the 2nd and 3rd power of the fundamental stages, before the channel selection filtering, where there
signal level, respectively. are strong adjacent channel/blocking components present.
Third-order IMD of RF modules is always a major concern. Specifications for the IP values can be derived from the
Depending on the radio architecture, also second- and knowledge of the maximum signal levels in adjacent
third-order products of analog IF modules may be frequency channels and the minimum signal level and
problematic. required SNR of the wanted signal.
IP2, IP3 (cont’d) Example of IP3 Calculations

In general, if the fundamental (target) signal level (at a Assume that the IIP3 value of a receiver is 10 dBm.
certain point in the receiver chain) is x [dBm] and we know Assume further that there are two strong blocking signals at
the IP3-value calculated from the preceding stages, then nearby frequencies, causing 3rd-order intermodulation to
third-order intermodulation products are 2(IP3-x) [dB] below the desired signal band.
the fundamental signal level (assuming, of course, that
x<<IP3). Suppose the radio system specifications allow the blocking
signals to be at 60 dB higher level than the desired signal.
This gives the basis for relatively easy dimensioning in
receiver system calculations. Finally, we assume that 10 dB signal to interference ratio
(SIR) is sufficient at detection interface.
These calculations are mostly relevant in the receiver
stages before the channel selection filtering, where there Then we can calculate that if the blocking signal level is 35
are strong adjacent channel/blocking components. In the dB below the IP3 point, the intermodulation product is 70
calculations, the fundamental signal level is determined by dB below the blocking signal level, and the 10 dB SIR
the worst-case blocker level. target is exactly satisfied.
This means that the maximum tolarated blocking signal
level at the input of the receiver is -25 dBm.
In practice, some IP3
implementation margin
1dB
against other interference
OUTPUT LEVEL, dBm

and noise is of course
needed. TA
L
T
EN
UC
DYNAMIC RANGE
M
SPURIOUS-FREE
OD
N DA
PR
FU
IM
ER
RD
D-O
3R
INTERCEPT POINT (IP3)

COMPRESSION
INPUT-REFERRED
INPUT LEVEL, dBm
POINT
Effects of 2nd and 3rd-Order Intermodulation Non-Idealities in Oscillators

In-band effects (i.e., may exist even if no adjacent channels
2nd-order intermodulation products are generally clearly and blockers are present):
stronger than 3rd-order products.  Constant phase error rotates the constellation; This can be
corrected by baseband processing afterwards
=> If 2nd-order intermodulation products of some  Phase noise: random fluctuations in the instantaneous
strong signals (blocking signals) appear on the phase/frequency of the oscillator cause random constellation
signal band, then better linearity is required. rotations, reducing the noise margin and increasing BER:
4
 In typical superheterodyne receivers, only 3rd-order 3
intermodulation is a problem, because signals causing 2
2nd-order products on the signal band are attenuated by Q 0
the RF filter. -1
-2
-3
 In wideband superhet with relatively low IF, also the 2nd- -4

-4 -3 -2 -1 0 1 2 3 4
order intermodulation may become a problem. I

The RF effects of phase noise tend to be far more critical
 2nd-order intermodulation is always a problem in direct- than in-band effects:
conversion and low-IF receivers.  Mixing products of the phase noise spectrum and strong
adjacent channel signals (reciprocal mixing) may produce
spurious signals which overlap the desired signal. The following
figure shows the noisy LO spectrum, RF-spectrum, and
spectrum after mixing with practical LO with phase noise.
f f
fLO fc
f
fIF
Example of Phase Noise Calculations Contents

1. Background and introduction
VCO phase noise in a GSM 900 terminal 2. Basic receiver architectures and their properties
- Main components; frequency translations and filtering
 wanted signal -102 dBm + 3 dB = -99 dBm - Receiver architectures: superhet, direct conversion, low IF
- Mirror-frequency considerations and effects of I/Q
 blocking signal -43 dBm @ 600 kHz
imbalance
 Minimum S/N=9 dB - Non-idealities and performance measures of the analog
 target signal bandwidth 200 kHz front-end: sensitivity, noise figure, intermodulation, phase
noise
Assume that the phase noise spectrum is flat within the 3. A few notes about transmitter and receiver-transmitter
noise bandwidth with x dBc/Hz (i.e., the noise power in 1 co-operation
Hz bandwidth is x dB in reference to the power of the VCO 4. Sampling of communications waveforms
at the LO frequency) - Sampling theorem for complex signals
- Sampling alternatives
 noise power at target signal bandwidth: - Non-idealities:quantization noise and aperture jitter
-43 dBm+x+10 log10 200000 < -99 - 9 dBm 5. Increasing to role of DSP in receiver front-ends
 x < -118 dBc/Hz (@600 kHz) - RF challenges in cognitive radio
- DSP methods to reduce the effects of RF impairments in
radios
- Case studies
f
fIF fIF +600 kHz
About the Co-Operation of Simultaneous Operation of

Transmitter and Receiver Transmitter and Receiver
There are two ways of sharing the antenna between the
Duplexing Techniques transmitter and receiver:
Uplink and downlink signals can be separated using  Switching: Connecting the antenna to RX or TX as
different carrier frequencies or different time slots: needed.
 In Time Division Duplex (TDD), TX and RX use the same  Duplexer: Using a band-splitting filter to connect the RX
frequency but different time slots band to the receiver and transmitter to the TX band.
 In Frequency Division Duplex (FDD), RX and TX are As a passive device, the duplexer attenuates the received
working at the same time but use different carrier signal and thus increases the noise figure of the receiver
frequencies. considerably. Better noise figure can be achieved with
switching, even though passive filtering is usually needed
The GSM system is based on FDD, but it has also a TDD
also in this case both in RX and TX.
element in a sense through nonoverlapping sequencing of
TDMA slots in uplink and downlink (in individual terminal). If the TX and RX are not working at the same time, then
 In the basic system, terminal TX and RX are not working  Either switching or duplexer can be used.
at the same time  The same frequency synthesizer can be used for RX and
 In multislot services (HSCSD, GPRS) with high data rate, TX (if the settling is fast enough).
the RX and TX may need to operate at the same time.  Less spurious frequencies, because TX is silent when RX
is working.
UMTS system, including the enhancements (GPRS, LTE),
is based on the FDD principle and the transmitter and If the TX and RX are working at the same time, then
receiver are usually operating at the same time. (There is  Duplexer has to be used.
also a TDD variant of UMTS, but it is not currently in use.)  The frequency synthesizer has to produce all the
WLAN and WiMAX systems are based on TDD. frequencies needed by TX and RX at the same time.
 Controlling the spurs gets more difficult; isolation between
TX and RX is a big concern.
Critical Issues on Transmitter Side Power Amplifier (PA)

Sufficiently bandlimited radio waveform at target center-  For constant envelope modulations (FSK, GFSK, MSK,
frequency (tunability) and at desired output power. GMSK) nonlinear power amplifiers can be used and good
power efficiency is achieved.
Main performance criteria - With exactly constant envelope signals, nonlinearities
produce only harmonics, but no other new spectral
- Spectral purity components.
- power efficiency  In other cases (linear modulations, and especially CDMA
- Signal distortion and OFDM), linear power amplifier is needed.
- Nonlinearity causes spectral regrowth and signal
Transmitter architecture distortion.
 Back-off measures the headroom between the average
- Two-stage conversion (with IF) is a safe solution, but transmitted signal power and the maximum (saturated)
difficulties in integration. output power of the PA.
- Increasing back-off improves linearity, but reduces
- I/Q imbalance is a problem in "low-IF" type of power-efficiency.
architectures. 20
Spectral Regrowth
windowed OFDM,
IBO = 0 dB
6 dB
0
- Direct conversion has problems, like leakage from power 12 dB
18 dB
100 dB
amplifier output to LO may have bad effects. -20
- Direct Digital Synthesis (DDS) is a promising approach, -40
PSD in dB
but the DA-converter performance is still a problem in -60
mobile systems.
-80
- DSP based enhancement of the signal quality can be -100
used also on the transmitter side; digital pre-distortion.

-120
-60 -40 -20 0 20 40 60
frequency in MHz
Sampling Theorem
Contents
1. Background and introduction The sampling theorem says that a (real or complex)
lowpass signal limited to the frequency band [-W, W] can
2. Basic receiver architectures and their properties represented completely by discrete-time samples if the
sampling rate (1/T) is at least 2W.
- Mirror-frequency considerations and effects of I/Q In case of a complex signal, each sample is, of course, a
imbalance complex number.
front-end: sensitivity, noise figure, intermodulation, phase In general, discrete-time signals have periodic spectra,
noise where the continuous-time spectrum is repeated around
3. A few slides about transmitter and receiver-transmitter frequencies 1 T , 2 T , 3 T , 
co-operation
- Sampling theorem for complex signals
- Sampling alternatives
- Non-idealities:quantization noise and aperture jitter f
2fs fs 0 fs 2fs
- Receiver architectures for software radio In case of complex signals, it is not required that the
- RF challenges in cognitive radio original signal is located symmetrically around 0.
Any part of the periodic signal can be considered as the
- DSP methods to reduce the effects of RF impairments in useful part. This allows many possibilities for multirate
radios processing of bandpass signals.
- Case studies In general, the key criterion is that no destructive aliasing
effect occur on top of the desired part of the spectrum.
Real vs. Complex Discrete-Time Signals Real Bandpass Sampling

Down-conversion can also be implemented by sampling a
Real signal: bandpass signal. Any part of the periodic spectrum can be
selected for further processing.
fs
f
fs 0 W fs
T/H
f
Here 2W real samples per second are sufficient to fcW/2 fc
represent the signal.
f
Complex signal: kfsfs/2 kfs kfs+fs/2
(in above assumed that fc = kfs + W/2)
Concerning the sampling frequency, it is sufficient that no

f aliasing appears on top of the desired band.
fs 0 fs=W
In general, the feasible sampling frequencies are
Here W complex samples per second are sufficient. determined from W, B (useful signal bandwidth), and fs.
Minimum sampling frequency is B+W, which is adequate in
 The resulting rates of real-valued samples are the the case where the center frequency of the desired signal
same. is kfs fs/4:
W+B=fs
 However, the quantization effects may be quite
different. (Recall from the standard treatment of
SSB that Hilbert-transformed signals may be
difficult.)
f
kfsfs/4 kfs kfs+fs/4
Quadrature Sampling Second-Order Sampling

In this case we are sampling the complex analytic signal
Quadrature sampling can be approximated by the following
obtained by a phase-splitter:
structure:
T/H I T/H I
fs fs
90 T/H Q =1/4fc
f ANALYTIC
fcW/2 fc BANDPASS
T/H Q
SIGNAL
At the carrier frequency, the sampling time offset

corresponds exactly to the 90o phase shift. Farther away
f from the center frequency this is only approximative, but for
kfs k+1fs relatively narrowband signals, it works.
(in above assumed again that fc = kfs + W/2)
The nonideality can be evaluated using the phase
imbalance analysis.
Of course, in practise the image bands in between the
shown periodic replicas are not completely attenuated but
only suppressed to a level that is determined by the
amplitude and phase imbalances in the phase splitter &
sampler & ADC blocks.
The gain and phase imbalance analysis of quadrature

down-conversion applies also to this case.
Analysis of Second-Order Sampling Problems with Wideband Sampling

The second-order sampling concept works perfectly at the Analog to Digital Converter (ADC)
carrier frequency (ignoring the other sources of I/Q Sampling a wideband signal, containing several channels is
imbalance) but only approximately at other frequencies. At a tempting approach for designing a flexible radio receiver.
frequency f c  f  , a time-shift of 1 (4 f c ) corresponds to a However, there are some great challenges to do this.
phase shift of The strongest signal in the ADC input signal band should
1 be in the linear range of the ADC. When the desired signal
4 fc  f 
 2  1    rads is weak, a large ADC dynamic range is needed, the
1  fc  2 resolution of the converter has to be many bits, e.g., 14 ...
( fc  f  )
17 bits.
We are actually dealing with phase imbalance and the MAGNITUDE
image rejection formula for quadrature mixing can be
utilized. The resulting image rejection is:
f 
1  cos   
REQUIRED
DYNAMIC
1  cos   fc 2  RANGE
R 
1  cos  f 
1  cos    f
 fc 2 
STRONG NEIGHBORING WEAK DESIRED
CHANNELS CHANNEL
Example: fc=1 GHz
Sampling
f f Phase Image rejection The sampling to get a discrete time signal is done usually
fc imbalance with a track-and-hold circuit (T/H).
0.1 MHz 0.0001 0.009o 82.1dB In practical sampling clocks and sampling circuits, there are
1 MHz 0.001 0.09o 62.1 dB unavoidable random variations in the sampling instants,
10 MHz 0.01 0.9o 42.1 dB sampling aperture jitter. In bandpass sampling, the
100 MHz 0.1 9o 22.1 dB requirements for aperture jitter become very hard.
Quantization Noise in ADCs Spurious-Free Dynamic Range

In general, the maximum S/N-ratio for an A/D-converter is
Practical ADC's have also discrete spectral frequency
SNR  6.02n  4.76  CFdB  10log10  f s / 2 B  (dB) components, spurious signals (or spurs), in addition to the
where n is the number of bits flat quantization noise.
CFdB is the Crest Factor in dB In many applications, the spurious-free dynamic range,
B is the useful signal bandwidth SFDR, is the primary measure of the dynamic range of the
f s is the sampling rate. converter.
Crest factor for a voltage signal is defined as the ratio of the peak
absolute value and the RMS-value. The maximum SNR is
achieved when the signal utilizes the A/D-converter’s full voltage
range, which is assumed to be symmetric around DC. For a
sinusoidal signal, the crest factor is 3 dB.
The last term takes into account the processing gain due to
oversampling in relation to the useful signal band. When the
quantization noise outside that useful signal band is filtered away,
the overall qantization noise power is reduced by the factor fs/2B.
The number of additional bits needed to quantize a wideband
signal can be estimated by:
 CF 2  P 
10log10  B B /6 bits
2
 CFd  Pd 
where PB is the worst case power in the full band
Pd is the minimum useful signal power
CFB is the crest factor in the worst case test situation
CFd is the crest factor of the desired signal.
Usually, in communications receivers, the worst case power is
determined from the adjacent channel or blocking specifications.
For many types of communications signals (e.g., CDMA, OFDM),
the crest factor is much higher than that of the sinusoid, which
might be a valid assumption in the blocking test situation.
Track&Hold Circuit Nonidealities Aperture Jitter

Advanced bandpass sampling approaches could Aperture jitter is the variation in time of the exact
mean that we are sampling a tens-of-MHz to GHz- sampling instant, that causes phase modulation and
range signal with a relatively low sampling rate. results in an additional noise component in the
Noise Aliasing
sampled signal.
Wideband noise at the sampling circuitry will be Aperture jitter is caused both by the sampling clock
aliased to the signal band. In case of bandpass and the sampling circuit.
sampling, aliasing increases with increasing
subsampling (fc/fs) factor. Basically, the noise figure
depends on the subsampling factor.
Therefore, it is important to have a good noise figure
for the track&hold circuit and/or to have sufficient
amplification in the analog front-end.
f
SNR Due to Sampling Jitter About A/D-Conversion for SW Radio

The noise produced by aperture jitter is usually modelled It is obvious that the requirements for the T/H-circuit and
as white noise, which results in a signal-to-noise ratio of A/D-converter are the main bottlenecks for implementing
receiver selectivity with DSP.
 1 
SNRaj  20 log10   One promising A/D-converter technology in this context is
 2 f max a 
T
the sigma-delta (principle.
where fmax is the maximum frequency in the sampler input
and Ta is the RMS value of the aperture jitter. - This principle involves low-resolution, high-speed
conversion in a noise-shaping configuration, together
This model is derived for a sinusoidal input signal, but with decimating noise filtering.
applied also more generally. In critical test cases of the
wideband sampling receiver application, the blocking signal - In case of lowpass and bandpass sampling with suitable
is often defined as a sinusoidal signal, and the model is fixed center frequency, this principle can be combined
expected to work reasonably well. nicely with the selectivity filtering part of the receiver.
The processing gain due to oversampling effects in the Noise filtering in basic ADC:
same way as in case of quantization noise.
Example of Sampling Jitter Effects
110
Aperture jitter effect f
- 14 bits 100 fc
Quantization effect
- 1 ps RMS jitter Joint effect
90 Noise filtering in sigma-delta converter:
SNR in dB
80
70
60 f
fc
50
40
0 1 2 3
10 10 10 10
Frequency in MHz
Summarizing Selectivity Tradeoffs

Contents
1. Selectivity at IF (superhet) 1. Background and introduction
- high-cost IF filters 2. Basic receiver architectures and their properties
- less demands for analog circuits after IF
- simple A/D converter - Mirror-frequency considerations and effects of I/Q
2. Selectivity by analog baseband processing imbalance
(direct conversion case)
front-end: sensitivity, noise figure, intermodulation, phase
- no costly IF filters noise
- more RF gain needed 3. A few notes about transmitter and receiver-transmitter
=> RF has to very linear to avoid intermodulation effects co-operation
- simple A/D converter
3. Selectivity by baseband digital filtering - Sampling theorem for complex signals
(direct conversion and low-IF cases) - Sampling alternatives
- Non-idealities:quantization noise and aperture jitter
- no costly IF filters
- more flexible than case 2: selectivity can be easily
adapted to different systems
- RF challenges in cognitive radio
- more RF gain with high linearity needed
- high dynamic range A/D-converters (14…17 bits) 6. Dirty-RF signal processing (separate material)
4. Selectivity by digital filtering after wideband IF - DSP methods to reduce the effects of RF impairments in
sampling radios
- simplified IF filter - Case studies
- high flexibility
- suitable for multi-channel receivers with common
analog parts
- high dynamic range A/D-converters (14…17 bits)
- very strict demands for low jitter sampling clock
Classification of DSP-Based Receiver Narrowband front-end

Architectures Per-channel down-conversion (0, or IF)
- Selectivity in analog part, good IF/BB filters needed.

Location of ADC (A/D-converter) - "Normal" frequency sythesizer needed.
1. Baseband or low IF - No big demands for ADC dynamic range or jitter.
2. IF - Sampling rate requirements are such that it is enough
to attenuate aliasing to the desired band.
3. RF
- Narrowband (baseband or bandpass) ADC (like )
can be utilized.
- Can in general build on real mixing (IF) or I/Q mixing (0
Analog front-end bandwidth - ADC bandwidth
or IF), depending on the characteristics of RF filtering
1. Single channel
2. Few channels
3. Frequency slice RF-stages Mixer IF-stages
4. Service band (e.g., GSM)
BP/LP
ADC
5. Frequency band (like 2 GHz range) filter
The bandwidths of the analog front-end and ADC may or LO

may not go hand in hand. Some examples below.
fIF
Wideband front-end, per-channel down- Wideband front-end, wideband (like slice or

conversion so that desired channel is around a service band) down-conversion so that desired
fixed center frequency (0 or IF) channel is located in a wider frequency range
- Analog front-end simplified in the sense that highly - Analog front-end simplified in the sense that highly
selective IF filters are not needed. selective IF filters are not needed.
- "Normal" frequency synthesizer needed. - Only single or few LO frequencies needed for each
- Selectivity in digital part, with fixed center frequency. service band -> simplified synthesizer; fast frequency
hopping becomes feasible through digital tuning.
- Higher demands for ADC dynamic range and jitter.
- Selectivity in digital part, with tunable center frequency.
- Sampling rate requirements are such that it is enough
to attenuate aliasing to the desired band. - High demands for ADC dynamic range and jitter.
- Narrowband ADC (like ) can still be utilized (in the - Sampling rate requirements are such that no aliasing
sense that high accuracy is only needed at fixed band). into the whole band is allowed (or needs to be tuned).
- Can in general build on real mixing (IF) or I/Q mixing (0 - Wideband ADC (or  with tunable center
or IF), depending on the characteristics of RF filtering. frequency!?) needed.
- Can in general build on real mixing (IF) or I/Q mixing (0
or IF), depending on the characteristics of RF filtering.
RF-stages Mixer IF-stages

RF-stages Mixer IF-stages
BP/LP
filter ADC BP /LP
filter ADC
LO
LO
fIF
BIF
BIF
About the Choice Between Lowpass and About Direct Sampling Architecture
Bandpass Sampling
In high-performance systems, it is necessary to have some
Due to the I/Q gain and phase imbalance problems in selectivity and gain before sampling. The reasons are
practical analog circuitry, the wideband downconversion -  signal aliasing
wideband sampling approach is very difficult to implement  noise aliasing
at 0 (or low) IF. But utilizing a combination of different Sampling is inherently more noisy operation than mixing!
techniques for mitigating these effects, the mentioned
approach is becoming feasible, but mostly on the base- Sampling directly from the antenna signal is usually not
station side. adequate.
On the other hand, wideband IF sampling is very The Ultimate SW Radio Architecture
challenging due to the apperture jitter and other
implementation problems concerning the sampling circuitry
(usually track&hold). a bank of RF filters T/H
Antenna and LNA’s for different A/D
The ADC requirements (apart from the sampling process) frequency bands DSP
concern mainly the spurious-free dynamic range, and are
not so heavily depending on the choice between lowpass The needed technologies are not mature for
or IF sampling. challenging radio system specifications in the
frequency bands used in mobile systems!
However, the useful ADC bandwidth has a great impact,
e.g., on power consumption. So the cost/complexity metrics
However, direct sampling is already an interesting
for per-channel A/D-conversion (usually and architecture in various applications
multichannel A/D-conversion (usually something else than o For example, satellite-based positioning (GPS/Galileo)
are quite different. where the dynamic range requirements are greatly
reduced comparing with wireless communications.
Direct Sampling & Analog Discrete-Time Some Dependencies and Conclusions

Processing Considering sampling and A/D-conversion
Texas Instruments (TI) has introduced so-called digital  highest signal frequency determines the T/H
radio processor (DRP) concept that is based on direct bandwidth and jitter requirements
sampling, together with analog discrete-time processing to  signal bandwidth (after analog RF/IF/baseband
implement main part of the channel selectivity, down- filtering) determines the minimum sampling rate
conversion, and sampling rate reduction. Increasing the degree of bandpass subsampling (
o For example, CIC/running-sum filters can be f c / f s ) leads to
implemented with switched-capacitor techniques with  lower sampling rate
analog processing.
 more selectivity needed before sampling
o Then the ADC is operating at relatively low rate and  more noise aliasing => more gain needed before
has reduced dynamic range requirements compared to sampling
digital direct-sampling approach.  lower processing gain -> more bits from ADC & tighter
Such designs have so far been considered for systems with jitter requirements
low-order modulation and relatively narrow bandwidth or Implementing the receiver selectivity in DSP-part
reduced dynamic range, like GPS, Bluetooth, and leads to
GSM/GPRS.
 simplified analog part
In such architectures, also the sampling process may be  hard requirements for the T/H and ADC dynamic
designed to provide frequency selectivity. Then the idea of range
the sampling process is not anymore just taking Wideband sampling
instantaneous sample values, but to
 has been mostly considered at IF due to I/Q-
- Integrate the signal over a finite-length interval imbalance problems; direct conversion/low-IF
- Weighting the input signal by a proper window during becoming feasible, depending on system specs
the integration interval. Rectangular window results Using IF sampling
in sinc-response, other kind of windows can be  sets hard requirements for the T/H circuitry and jitter of
designed for optimized performance. the sampling clock.
Connection to Advanced Broadband Wireless RF Challenges in Cognitive Radio

System Developments As seen above, the RF implementation challenges can be
greatly relieved in radio environments, where the signal
The latest and future wireless communication systems use levels are well-controlled by the radio resource
increasing bandwidths for data transmission. For example, management functionalities.
3GPP-LTE and WiMAX have the maximum bandwidth of
20 MHz, and the next generation (“IMT-advance”) is A completely different case is the opportunistic spectrum
targeted to bandwidths of up to 100 MHz. use in the cognitive radio context. Here the spectral gaps
Especially, LTE is using frequency-division multiplexing, (so-called white spaces) between licensed users (primary
and the spectrum entering the receiver resembles that of a users) are intended to be used locally by secondary users.
multichannel receiver for more narrowband systems. Some The secondary system has no control of the power levels in
characteristics and comments: adjacent frequency slots, and it should also operate at
- The wide bandwidth makes it possible to utilize fast relatively low power levels, in order not to introduce
frequency hopping and other forms of frequency diversity, interference to the primary users.
to enhance the transmitted data rate.
- In LTE (and other similar systems) the power levels of the The needed flexibility regarding the signal waveforms,
frequency slots of different users are well-controlled (e.g., bandwidths, and centre frequencies can be only achieved
20 dB maximum variation in the power levels). This is in through software radio techniques. In order to efficiently
contrast to, e.g., multichannel GSM receiver, were the utilize the spectral gaps, a cognitive radio receiver should
dynamic range is much bigger. This makes it feasible to
be able to operate in the presence of strong primary signals
implement the needed wideband receivers for such
systems. in adjacent frequency slots.
- Direct conversion architecture is preferred. Actually, for
most of the frequency channels, the low-IF model is valid.
- IQ-imbalance is significant, but not very critical because of
the well-controlled power levels. DSP-based IQ-imbalance
compensation is interesting in case of high-order
modulations.
- In these systems, and in OFDM systems in general, the
frequencies at or close to DC in baseband processing are
commonly not utilized in order to make direct conversion
receiver design easier.
Beamforming and Dynamic Range Multimode Terminal Receivers

Among the various multiantenna techniques, beamforming In flexible multi-mode receivers, the target is to use
on the receiver side helps to reduce the dynamic range of common blocks for different systems as much as possible.
the received signal. This is achieved by directing zeros of
A long-term target is to make the transceiver configurable
the radiation pattern towards strong interfering transmitters.
for any system. However, presently a combination of a few
Naturally, to achieve this benefit in the analog-to-digital predetermined systems is more realistic, e.g.,
interface, the combination of the antenna signals should be GSM/3GPP/WLAN.
done at the analog RF stage.
A realistic approach has the following elements:
- Separate RF stages for different systems.
- Common IF/baseband analog parts; bandwidth
according to the most wideband system.
- Common ADC at IF or baseband; fixed sampling rate.
- Especially in the terminal side: careful choice of IF
frequency & sampling rate to make the down-
conversion simple. Typically, fIF=(2k+1) fs/4.
- Digital channel selection filtering optimized for the
different systems.
I
LNA IF1 filter AGC S&H DSP
ADC 0,1,0,-1
Q
IF2 DSP
LO 1,0,-1,0
DSP for Flexible Receivers Digital Channel Selection & Down-Conversion

Digital Down-Conversion
In advanced SW radio concepts, the selectivity filtering and
down-conversion are moved from analog continuous-time 1. Desired channel centered at fixed IF
part to the discrete-time/DSP part.  Fixed down-conversion
Special choices of fIF and fs make things easy.
 Here efficient multirate filtering techniques become Especially when fIF=(2k+1) fs/4, the signal aliases to fs/4 and
very important. down-conversion is very easy.
2. Wideband sampling case
It also helps to move as much as possible functionality from
 Tunable down-conversion and NCO (numerically controlled
the analog or digital front-end to baseband processing.
oscillator) needed.
 All-digital synchronization concept becomes very Also frequency translation due to sampling rate conversion
interesting in this context. operations can be utilized.
o Free-running local oscillators for demodulation and There are possibilities to move the I/Q mixer and NCO closer
to baseband, which helps to reduce the power consumption.
frequency conversion.
o Free-running sampling clock. Channel Selection Filtering
o Errors are compensated in digital part. - After down-conversion, efficient lowpass decimator structure is
needed.
Some errors due to RF-front-end can be corrected by DSP. - Multiplierles CIC-filters are commonly used in the first
decimation stages, FIR-filters and the last stages. Nth-band IIR
 Dirty RF paradigm filters also an efficient solution.
o I/Q imbalance compensation Adjusting Symbol Rates
o Compensation of intermodulation effects - Different systems use different symbol/chip rates.
- Common sampling clock frequency is preferred.
o etc.
 Decimaton by a fractional factor is needed.
- This can be done at baseband or earlier in the decimation
chain.
[11] B. Brannon, "Aperture uncertainty and ADC system performance," Analog

Further reading Devices, Application Note AN-501, Nov. 1998.
[12] J. Tsui, Digital Techniques for Wideband Receivers. Artech House 1995.
Receiver architectures: [1] - [9]
[13] M.E. Frerking, Digital Signal Processing in Communication Systems.
Direct conversion architecture: [6] - [8]
Chapman & Hall, 1994.
Low-IF architecture: [9]
Analog-digital interface: [10] - [13] [14] L. Larson et al., "Device and circuit approaches for improved wireless
Transmitter issues: [14] communications transmitters," IEEE Personal Communications, pp.18-23, Oct.
Direct digital synthesis: [15] 1999.
Software radio architecture: [12], [13], [16], [17] [15] J. Vankka, K. Halonen, Direct digital synthesizers: Theory, design, and
Advanced sampling & analog discrete-time processing: [18] – [22] applications. Kluwer 2001.
Cognitive radio challenges: [23] [16] T. Hentschel, G. Fettweis, "Software Radio Receivers". Chapter 10 in F.
Swarts (ed.), CDMA Techniques for Third Generation Mobile Systems. Kluwer,
Literature 1999.
[1] U.L. Rohde, J.C. Whitaker, T.T.N. Bucher, Communications Receivers: [17] M. Valkama, J. Pirskanen, and M. Renfors, "Signal Processing Challenges
Principles and Design, 2nd Edition. McGraw-Hill 1997. for Applying Software Radio Principles in Future Wireless Terminals: An
Overview", International Journal of Communication Systems (Wiley), fall 2002.
[2] S. Sheng, R. Brodersen, Low-Power CMOS Wireless Communications - A
Wideband CDMA System Design. Kluwer 1998. [18] K. Muhammad et al., “The First Fully Integrated Quad-Band GSM/GPRS
[3] L.E. Larson (Ed.), RF and Microwave Circuit Design for Wireless Receiver in a 90-nm Digital CMOS Process,” IEEE Journal of Solid-State
Communications. Artech House 1996. (Chapters 2 and 3) Circuits, vol. 41, pp. 1772-1783, August 2006.
[4] C. Chien, Digital Radio Systems on a Chip, A Systems Approach. Kluwer [19] D. Jakonis, K. Folkesson, J. Dabrowski, P. Eriksson, C. Svensson, “A 2.4
2001. GHz RF Sampling Receiver Front-End in 0.18 mm CMOS,” IEEE Journal of
[5] P.-I. Mak, S.-P U, and R. P. Martins, “Transceiver architecture selection: Solid-State Circuits, vol. 40, pp. 1265-1277, June 2005.
Review, State-of-the-Art Survey and Case Study”, IEEE Circuits and Systems [20] S. Andersson, Multiband LNA Design and RF-Sampling Front Ends for
Magazine, vol. 7, 2nd quarter 2007, pp. 6-25. Flexible Wireless Receivers. Doctoral Dissertation, Department of Electrical
[6] B. Razavi, "Challenges in portable RF transceiver design," IEEE Circuits and Engineering, Linköping University, Sweden, 2006.
Devices Magazine, Sept. 1996, pp. 12-25. [21] S. Karvonen, Charge-Domain Sampling of High-Frequency Signals with
[7] B. Razavi, "Design considerations for direct-conversion receivers, " IEEE Embedded Filtering, Doctoral Dissertation, Department of Electrical and
Trans Circuits Syst. -II, vol. 44, pp. 428-435, June 1997. Information Engineering, Oulu University, Finland, Jan. 2006.
[8] A. Loke, F. Ali, "Direct conversion radio for digital mobile phones - Design [22] R. Bagheri et al., “Software-Defined Radio Receiver: Dream to Reality,”
issues, status, and trends," IEEE Trans. Microwave Theory and Techniques, vol IEEE Communications Magazine, pp. 111-118, August 2006.
50, no. 11, pp. 2422-2435, Nov. 2002.
[23] J. Yang, R.W. Brodersen, D. Tse, “Addressing the dynamic range problem in
[9] J. Crols, M.S. Steyaert, "Low-IF topologies for high-performance analog front cognitive radios”, in Proc. ICC 2007 (Glasgow, UK), June 2007.
ends of fully integrated receievrs," IEEE Trans Circuits Syst. -II, vol. 45, pp. 269-
282, March 1998.
[10] R. H. Walden, "Analog-to-digital converter survey and analysis," IEEE J.
Select. Areas Commun., vol. 17, pp. 539-550, Apr. 1999.

Radio Arch SP Short 2pp

Uploaded by

Copyright:

Available Formats

Radio Arch SP Short 2pp

Uploaded by

Document Information

Original Description:

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

Radio Arch SP Short 2pp

Uploaded by

Copyright:

Available Formats

Radio Architectures and Signal Processing - short pp. 1 / 84 Radio Architectures and Signal Processing - short pp.

Short Overview Principal Transmitter Functionalities

Radio transmitters and receivers are essentially I

DIGITAL BB/IF ANALOG BB/IF ANALOG RF

Different radio architectures then mean how the

Principal Receiver Functionalities Focus in this presentation

Analog RF/IF/BB Anothe central theme is related to different

Essential System Characteristics Main Targets in Radio Parts in Wireless

Key Areas of Development Possible Solutions

 MEMS (micro electro-mechanical systems) Supporting multiple radio access technologies in a

 Packaging  Separate highly integrated radios for different

Cognitive Radio Concepts

 Amplification to compensate for transmission losses  Amplifiers

and non-idealities affecting in the different building blocks AGC

ANALOG RF ANALOG IF/BB DIGITAL IF/BB

the frequency axis. Such filters find application in e j

- Real mixer produces the combination of frequency

More details in the separate material ! (earlier

Main Components for Receiver Front-Ends Classical Receiver Architecture:

Majority of all the receivers have been based on the

Filtering Requirements in Superheterodynes Some Drawbacks of the Superheterodyne

 The RF filter should provide sufficient attenuation for the  IF-filter

 LO and IF signals and harmonics and mixtures leaking to

Direct-Conversion Receiver Architecture Direct-Conversion Receiver Architecture

Direct-conversion principle has become quite popular

DC-Offsets in Direct-Conversion Receivers Low-IF Receiver Architecture

The channel filtering and final demodulation can be done after

Low-IF Receiver by Steyert et al. About Analytic Bandpass Signals

Other side of the desired signal spectrum suppressed by a

In this architecture, the image frequencies are suppressed

About Image Rejection as a Function of Gain and

Image Rejection as a Function of Gain and Examples of Mirror-Frequency Crosstalk with

This can be applied to all cases of quadrature mixing where

Frequency-Dependent Imbalance Model Effects of Gain and Phase Imbalance in

Then also the image suppression ratio is frequency-

Selectivity Requirements in System Example: GSM Interference Mask

don't allow a strong adjacent channel signal to be present −30

when a weak desired signal is to be received. (Radio

Signal Power [dBm]

For example, the GSM specifications give:

 The maximum levels for the 3 adjacent channels on both −80

 Maximum levels for more distant signals (>600 kHz from

carrier), blocking signals, in case of a sinusoidal interferer −120

Summary: Essential Non-Idealities and Performance

Essential Non-Idealities and Performance Essential Non-Idealities and Performance

Essential Non-Idealities and Performance Noise Figure

Intermodulation Intercept points - IP2, IP3

OUTPUT LEVEL, dBm

INTERCEPT POINT (IP3)

IP2, IP3 (cont’d) Example of IP3 Calculations

OUTPUT LEVEL, dBm

INTERCEPT POINT (IP3)

Effects of 2nd and 3rd-Order Intermodulation Non-Idealities in Oscillators

 In typical superheterodyne receivers, only 3rd-order 3

intermodulation is a problem, because signals causing 2

2nd-order products on the signal band are attenuated by Q 0

 In wideband superhet with relatively low IF, also the 2nd- -4

order intermodulation may become a problem. I

Example of Phase Noise Calculations Contents