GSM Systems

RF Network Design - Introduction

Slide No.1
 Frequency Bands
GSM-900
The term GSM-900 is used for any GSM system which operates in any 900 MHz
band.

P-GSM-900
P-GSM-900 band is the primary band for GSM-900 Frequency band for primary
GSM-900 (P-GSM-900) : 2 x 25 MHz

890 – 915 MHz for MS to BTS (uplink)

935 – 960 MHz for BTS to MS (downlink)

E-GSM-900
In some countries, GSM-900 is allowed to operate in part or in all of the following
extension band. E-GSM-900 (Extended GSM-900) band includes the primary
band (P-GSM-900) and the extension band :

880 – 890 MHz for MS to BTS (uplink)

925 – 935 MHz for BTS to MS (downlink)
Slide No.2
 Frequency Bands
R-GSM-900
R-GSM-900 (Railway GSM-900) band includes the primary band (P-GSM-900)
and the following extension band:
876 – 890 MHz for MS to BTS (uplink)
921 – 935 MHz for BTS to MS (downlink)

GSM-1800
Frequency band: 2 x 75 MHz
1710 – 1785 MHz for MS to BTs (uplink)
1805 – 1880 MHz for BTS to MS (downlink)

Slide No.3
 Carrier Spacing and Channel Structure
Channel number – the carrier frequency is designated by the absolute radio
frequency channel number (ARFCN). The frequency value of the carrier n in the
lower band is called FL (n) while FU (n) is the corresponding frequency value in
the upper band. Frequencies are in MHz

P-GSM-900:
FL (n) = 890 + 0.2 n with 1 < n < 124
FU (n) = FL (n) + 45

E-GSM-900:
FL (n) = 890 + 0.2 x n with 1 < n < 124
FL (n) = 890 + 0.2 x (n-1024) with 975 < n < 1024
FU (n) = FL (n) + 45

Slide No.4
 Carrier Spacing and Channel Structure
R-GSM-900:
FL (n) = 890 + 0.2 x n with 1 < n < 124
FL (n) = 890 + 0.2 x (n-1024) with 955 < n < 1024
FU (n) = FL (n) + 45

GSM-1800:
FL (n) = 1710.2 + 0.2 x (n-512) with 512 < n < 885
FU (n) = FL (n) + 95

• Carrier spacing is 200 kHz

• 8 time slots per carrier

Slide No.5
 Coverage, Capacity, and Quality
Providing coverage is usually considered as the first and most important activity
of a new cellular operator. For a while, every network is indeed coverage driven.
However, the coverage is not the only thing. It provides the means of service
and should meet certain quality measures.
The starting point is a set of coverage quality requirements.
• To guarantee a good quality in both uplink and downlink direction, the power
levels of BTS and MS should be in balance at the edge of a cell. Main output
results of the power link budgets are:
– Maximum path loss that can be tolerated between the MS and the BTS
– Maximum output power level of the BTS transmitter.
• These values are calculated as a function of design constraints:
– BTS and MS receiver sensitivity levels
– MS output power level
– Antenna gain
– Diversity reception
– Losses in combiners, cables, etc.
Slide No.6
 Coverage, Capacity, and Quality
The cell ranges are derived with propagation loss formulas such as Okumura-
Hata, using inputs of maximum path loss, differences in the operating
environments and the quality targets in different cell ranges.

The traffic capacity requirements have to be combined with the coverage

requirements, by allocating frequencies. This also may have impact on the cell
range.

Slide No.7
 Coverage Planning Strategies
The selection of site configurations, antennas and cables is the core of the
coverage planning strategy. The right choice will provide cost savings and
guarantees smooth network evolution.

Some typical configurations are:

• 3-sector sites for (sub)urban areas
• 2-sector sites for road coverage
• omni sites for rural areas
These are not the ultimate solutions, decisions should be based on a careful
analysis

Cell Range and Coverage Area

For any site configuration, the cell ranges can be determined given the
equipment losses and gains. The site coverage areas can be calculated then
and these will lead to the required number of sites for a given coverage region.
This makes it possible to estimate the cost, e.g. per km2, to be used for strategic
decisions.
Slide No.8
 Methodology
Define design rules and parameters
• Identify design rules to meet coverage and capacity targets efficiently
• Acquire software tools and databases
• Calibrate propagation models from measurements

Set performance targets

• Clear statement of coverage requirements (roll out and quality)
• Forecast traffic demand and distribution
• Test business plan for different roll out scenarios and quality levels

Design nominal plan

• Use computer tool to place sites to meet coverage and capacity targets
• Verify feasibility of meeting service requirements
• Ensure a frequency plan can be made for the design
• Estimate equipment requirements and costs
• Develop implementation and resource plans (including personnel
requirements)
• Radio plan will provide input to fixed network planning
Slide No.9
 Methodology
Implement cell plan
• Identify physical site locations near to nominal or theoretical locations, using
search areas.
• Modify nominal design as theoretical sites are replaced with physical sites
• Modify search areas in accordance with evolving network.
Produce frequency plan
• Fixed cluster configuration, can be done manually.
• Flexible, based on interference matrix using an automatic tool.
Optimising the network
Expand the network
• In line with the roll out requirement
• In line with the forecasted traffic level
• Improve the coverage quality
• Maintain the blocking performance

Slide No.10
 RF Propagation
A radio wave transmitted to and from a moving mobile station is subject to
several effects. These effects will cause loss of signal strength and interference.
The effects are:-

• Distance attenuation
• Shadowing
− Diffraction
• Rayleigh fading
− Reflections
− Inter-symbol interference
− Doppler shift
− Ducting

The most important conventional countermeasures to deal with the problems of

the mobile channels are :-

• The use of fade margins

• Various types of diversity reception
• Installation of supplementary BTSs
Slide No.11
 Practical Attenuation
In practice, the mobile radio link is not set up in free space. The path loss is
more severe that the inverse square law would predict. The slope will be
steeper, rather between –30 and –45 dB/decade, caused by :-

• Obstructions in the propagation path, particularly in the first Fresnel

zone
This is frequently the case because of the low height of the mobile antenna.
Even of line-of-sight conditions apply, the first Fresnel zone is obstructed in
most cases.

• Reflections from the ground and from objects

Reflections combine different phases of the signal on the receiving antenna.
This will cause multipath signal strength variations.

The loss is depended upon the frequency, the antenna design and the terrain.

Slide No.12
 Fade Margin
The concept of a fade margin is to reserve extra signal power to overcome
potential fading.

Assume :
• The mobile radio system needs an signal level of Pr dBm at the receiver
• The maximum likely fade (loss) is calculated to be L(fade) dB

The a received signal level of Pr dBm can be ensured by transmitting enough

power for a normal received signal level of (Pr + L(fade)) dBm

The fade margin is normally equal to the maximum expected fade or to a

smaller value. The value is chosen in such a way that the threshold value is
undershot in only a low percentage of time.

For this purpose, it is necessary to know the probability density function of the
fading.

In RF planning, the impact of Rayleigh fading is taken into account by

implementing an extra fade margin of 8 dB.
Slide No.13
 Multipath Propagation
The radio wave may be reflected, from a hill, a building, a truck, an aeroplane
or a discontinuity in the atmosphere. In some cases, the reflected signal is
significantly attenuated, while in others almost all the radio energy is reflected
and very little absorbed. The result is that not one but many different paths are
followed between the transmitter and receiver. This is known as Multipath
Propagation

Slide No.14
 Multipath Propagation
Reflection and multipath propagation can cause positive and negative effects :-

• Coverage extension
Multipath propagation allows radio signal to reach behind hills and buildings
and into tunnels
The latter effect is known as ducting

• Constructive and destructive interference

The interference due to multipath propagation manifest itself in the following
3 most important ways:-

– Random phase shift creates rapid fluctuations in the signal strength

known as Rayleigh fading
– A delay spread in the received signal causes each symbol to overlap with
adjacent symbols : intersymbol interference
– Random frequency modulation due to different doppler shifts on
different paths
Slide No.15
 Ducting
Ducting may occurs in tunnels, valleys, building canyons, and in the
atmosphere if the boundaries (steep hillsides, atmosphere layers) are good
reflectors for radio waves.

VHF frequencies do not propagate well in long tunnels, but higher frequencies
(>800 MHz) follow the tunnel like a waveguide.

If the coverage in a tunnel needs enhancement, repeater station at the tunnel

entrance radiating into the tunnel may help.

Slide No.16
 Rayleigh Fading
The reflected radio wave will be altered in both phase and amplitude. The
signal may effectively disappear if the reflected wave is 180 degrees out of
phase with the direct path signal. Partial out of phase relationships among
multiple received signals produce smaller reductions in received signal
strength.

Rayleigh fading is dependent on :

• Time
Time dependent fading is applicable for moving mobiles only

The countermeasure against time dependent Rayleigh fading is the use of

bit interleaving in burst building

Slide No.17
 Rayleigh Fading
• Location
The fading effect is a spatial effect. The depth and spacing of the fades is
related to the wavelength. Maximum fades are very deep (down to –40 dB or
less), a few inches apart. In between are many shallower fades. When a
mobile antenna moves through this field, the received signal strength will vary
very rapidly. Sometimes it is possible that a mobile is in a fade of the correct
BTS but not in a fade of any “incorrect” BTS transmitting on the same
frequency.

The countermeasure against location dependent Rayleigh fading is diversity

reception.

• Frequency
Due to the impact of the wavelength, the pattern of the fades is also
dependent on the radio frequency.

The countermeasure against frequency dependent Rayleigh fading is

frequency hopping reception
Slide No.18
 Inter-symbol Interference
The sharp pulse that is transmitted arrives in the receiver as a delayed, smeared
and flattened budge that lasts longer than the original pulse. This effect, called
delay spread is caused by multipath propagation effects.

If the delay spread is large relative to the average symbol duration, the
individual symbols will overlap each other and ISI will occur.

Slide No.19
 Doppler Shift
The movement of the MS relative to the BTS will cause a shift in frequency of
the radio signal, known as doppler shift. This frequency shift varies
considerably as the MS changes direction and/or speed.

Doppler shift introduces random frequency modulation in the radio signals.

Thus Doppler frequency shift Δ f is :-

Δ f = Vr / λ

where Vr is the radial speed component pointing to/from the BTS or a reflection
point. Doppler shift affects all multiple propagation paths, some with positive
shift, others a negative shift at the same instant. The power spectrum of the
received radio signal will be smeared.

Doppler shift effects can be limited by using a well-designed (adaptive)

equaliser in the receiver.

Slide No.20
 Equalisation
To some extent, the general countermeasure against distortion due to multipath
effects is adaptive equalisation :-

• The distortion characteristics of the channel are measured continuously.

It uses the well known 26 bits (or more) TSC training sequence transmitted in
each timeslot burst (once per 0.5ms) to measure the channel characteristics.
The TSC (training sequence codes) are specified in GSM Rec. 05.02.

• The predicted distortions in the received signal are subtracted from the
received signal.

Knowing the channel characteristics, the predicted distortion in the

transmitted pulses are subtracted from the received waveform and the most
likely sequence of data for the distorted received signal is estimated.

The Viterbi algorithm is an example of an adaptive MLSE (maximum

likelihood sequence estimation) solution.
Slide No.21
 Equalisation
The equaliser used for GSM is specified to equalise echos up to 15 μs after the
first signal. This corresponds to 4.5 km in distance. One bit period is 3.69 μs.
Hence, echos with about 4 bit lengths delay can be compensated.

Echos with a delay of > 15 μs cannot be cancelled by the equaliser. These

signals should be considered as co-channel interference for which the required
minimum C/I ratio of 9 dB must be met. This means that the sum of the echos
with delays of >= 15μs should remain >= 9 dB under the sum of the wanted
carrier signal plus the “useful” echos within the 15μs window. The echos
resulting from reflections just outside the ellipse for Δt = 15 μs are mostly the
strongest and will cause most trouble.

Slide No.22
 Fresnel Zone

A fresnel zone is a 3 dimensional body, bounded by ellipsoids that have their

focal points at the transmitter and the receiver antennas. The sum of the
distances from a point (P) on the ellipsoid to the transmitter (T) and to the
receiver (R) is n/2 wavelengths longer than the LOS path (S) :

Distance (P-T) + Distance (P-R) = S + n (λ/2)

For the first fresnel zone, n ≡ 1. The radius of the first Fresnel zone is r(F1). To
keep out of this zone, the distance r from the optical LOS should be :

λd1( S − d1)
r ≥ r ( F1) =
S

The obstacles may be hills, buildings or vegetation.

Slide No.23
 Diffraction
Shadowing does not always mean that no signal is received behind an
obstacle. Radiowaves may bend around obstructions to a certain extent. This
effect is called diffraction. The diffraction effect depends on the wavelength in
relation to the size of obstacle, and is greater the longer the wavelength.

Slide No.24
 Shadow Fading
The effect of shadowing by obstacles is fading of the received signal. The
problems of shadowing are most severe in heavily built-up urban centres.
Shadows as deep as 20dB may occur over very short distances, literally from
one street to another.

The fading effects produced by shadowing are often referred to as slow fading

The radio network planning tool uses a topographical database. The

topographical area is divided in a grid of pixels. Each pixel has an size in the
range of 50m x 50m to 500m x 500m.

One pixel is characterised by :-

• Terrain height
• Clutter type : high/low building, forest, water etc.

Slide No.25
 Shadow Fading
The shadowing problem is approached in 2 ways, depending on the size of the
obstruction :-

• Shadowing, diffraction and reflection by obstructions larger than the database

resolution (e.g. hills) can be predicted by propagation models in
computerised planning tools. The distances between the fading dips are in
the magnitude of hundreds of meters.

• Shadowing by obstructions smaller than the database resolution (e.g.

individual building) can be treated statistically. The distances between the
fading dips are in the magnitude of tens of meters.

Slide No.26
 Shadow Fade Margin
Shadow fade margins must be added to the receiver sensitivities specified in
GSM Rec 05.05, to give the probability of signal being greater than the receiver
sensitivity.

The fade margin depends on :-

• The desired coverage probability

• The propagation slope
• The standard deviation of the log-normal fading

Slide No.27
 Jakes Graphs
A way to find an appropriate fade margin is the method according to Jakes. It
can be used for a wide range of propagation slopes and standard deviations, by
using a set of standard graphs.

The inputs are :-

• The propagation slope, e.g. 40 dB/decade
This means that the signal will decay according to 1/rn where n = 4
• The shadow fading standard deviation σs, e.g. 7 dB
• The required coverage probability, e.g. 90% coverage probability over the
area

Slide No.28
 Jakes Graphs
The output will be the fade margin for a given required area coverage
probability. This can be found as follows :

1. Find the abscissa value σs/n

2. Take required area coverage probability P(area) as the ordinate value
3. The intersection of the 2 values will provide a value for the cell edge
coverage probability P(edge)
4. Find the fade margin for P(edge) in the CDF table for the standard
normal distribution table N(0,1)

Slide No.29
 Slow Fade Margin – Example
According to GSM 03.30, the normal case of urban propagation has a standard
deviation of σs = 7 dB while the propagation path loss slope is –35dB/decade. In
order to find the required fade margin to achieve 90% area coverage, the
following steps are taken :-

1. Determine the σs/n abscissa value :

The propagation slope is 35 dB/decade, then
n = 35/10 = 3.5
Because the σs = 7 dB, the value for σs/n = 7/3.5 = 2

2. In the graph, the P(area) = 90% and σs/n = 2

Intercept at the curve for P(edge) ~ 0.73 = 73%

3. In the normal distribution N(0,1) table, 0.73 corresponds to 0.61 x σs.

Hence, the fade margin = 0.61 x 7 = 4.3 dB

Slide No.30
 Propagation Modeling
• Statistical propagation models
− These calculate a median signal for each pixel. The level within this pixel
varies about the median in a way that can only be analysed statistically.
− Local mean signal levels are distributed around the pixel median with a
log-normal probability distribution.
− Formulas derived from measurements (e.g. Okumura-Hata).
− No obstacles assumed to be close to the BTS antenna.

• Deterministic propagation models

− Take into account individual buildings and use ray tracing techniques.
− Make use of high resolution map data (at least 10m).

Slide No.31
 Noise Levels
There are 2 kinds of noise that play a role in mobile communication :-

• Thermal noise
• Man-made noise (e.g. spurious signals)

The thermal noise depends on the receiver bandwidth B (in Hz) and the
absolute temperature T (Kelvin).

Ni = k T B Watt
Where
k = Boltsmann’s constant = 1.38 x 10-23 J/K

Slide No.32
 Noise Figure
A mobile radio signal, received on the antenna, will be amplified by the front-
end RF amplifier in the radio receiver. After amplification, the S/N ratio will be
worse than at the antenna because the amplifier has added some extra noise
by itself. The noise figure F is the ratio between :

• The total output noise level generated by both the external noise and the
internal noise of the amplifier
• The output noise level due to external (thermal) noise only

A typical noise figure for a GSM receiver is 6 dB. At a temperature of 17

degrees C and a receiver bandwidth of 200 kHz, the received thermal noise is :-

1.38 10-23 x (17 + 273) x 200 x 103 = 8 x 10-16 W = -120 dBm

Slide No.33
 Receiver Sensitivity
With the thermal noise level of –120 dBm and a noise figure F = 6 dB, the noise
floor will be at –114 dBm. The implementation margin being 2 dB and the fade
margin for Rayleigh fading being 8 dB, the reference receiver sensitivity can be
taken as :-

For normal GSM 900 BTS

-120 dBm + 6 dB + 2 dB + 8 dB = -104 dBm

Slide No.34
 Receiver Sensitivity
GSM 900 Receiver Sensitivity
The reference sensitivity levels specified in GSM Rec 05.05 are as follows:-
• -104 dBm (Class 1, 2 and 3 mobile stations and normal BTS)
• -120 dBm (Class 4 and 5 mobile stations)
• -97 dBm (micro-BTS M1)
• -92 dBm (micro-BTS M2)
• -87 dBm (micro-BTS M3)
This already take into account the effect of multipath fading on moving mobiles,
Rayleigh Fading (time domain) and Doppler Effect (frequency domain)

GSM 1800 Receiver Sensitivity

The reference sensitivity levels specified in GSM Rec 05.05 are as follows :-
• -102 dBm (class 3 mobile station or micro-BTS M1)
• -100 dBm (GSM 1800 class 1 and 2 mobile stations)
• -97 dBm (micro-BTS M2)
• -92 dBm (micro-BTS M3)
Slide No.35
 Cellular Architecture
The essential principles of the cellular architectures are :-

• Low power transmitters with antenna heights between 20 – 50 m

• Small coverage zones (cells), typical macro cell radius 3 – 30 km
• Frequency reuse (factor n = 3, 4, 7 ... )
• Cell splitting to increase local capacity
• Micro and pico cells act as patches for hot spots, tunnels and buildings

Balance is to be found between conflicting requirements of :

• Coverage
• Traffic capacity

Slide No.36
 Cell Clustering
Frequency reuse is the core concept of the cellular mobile radio system, given
the fact that the number of allowed frequencies is fixed. A frequency can be
reused simultaneously in different cells, provided that the cells using the same
frequency set are far enough separated so that co-channel interference is kept
at an acceptable level most of the time.
The total frequency spectrum allocation can be divided into K frequency reuse
patterns.
• Theoretically, a large K is desired.
In practice, the total number of allocated frequencies is fixed. When K is too
large, the number of frequencies assigned to each of K cells becomes too
small. Trunking inefficiency will be the result.
• The challenge is to find the smallest K value which can still meet our system
performance requirements. This involves :-
– Estimation of the co-channel interference
– Calculation of the minimum frequency reuse distance D to meet the co-
channel interference criterion
– The practical values for K range up from 3 to 21
Slide No.37
 Cluster Size
Valid values for K are found by setting i and j to positive values in :-
K = i2 + i j + j2
The smallest value for K is 3, found for i = j = 1.

The K value can be found as follows :-

• The starting direction of the i axis is arbitrary
• j is rotated by one cell face (60 degrees) to the left from the i axis
• After finding the first co-channel cell, go back to the starting cell
• Rotate the i axis by one cell face
• Repeat the procedure.

Frequency Reuse Distance

The frequency reuse distance D can be derived from the K value:-

D = R 3K
Slide No.38
 Cell Types
The 2 main cell types are :-

• Omni cells :
– Coverage is in principle a circle, but in reality a rough pattern

• Sector cells :
– 2 sectors (e.g. for highways)
– 3 sectors

Cell Coverage Area

Omni cell (Hexagon) = 2.6 R2

Sector cell (Hexagon) = 1.96 R2

Slide No.39
 Base Station Antenna Problems
Problems that are encountered in the design and installation of cellular
antennas :-

• Dead Spots
Slight unintentional tilts and minor lobes nulls in the radiation pattern may
result in gain loss on some spots

• Isolation
The more spacing between transmitter and receiver antennas. Less the
coupling

• Collinear antenna mounting

Only one antenna can be mounted at the top most point of the site tower.
Several antennas cannot be mounted at the same point

Slide No.40
 Dead Spots
A higher antenna gain is achieved by compressing the beamwidth in the
elevation plane. Unfortunately, with compression, more minor lobes appear in
the radiation pattern. In the desired coverage area, nearby dead spots may exist
due to minor lobe nulls even though the distant coverage is good because of a
high main lobe gain.

Moving a dead spot away from a certain location can be done by :-

• Tilting the antenna beam

• Reduction of antenna height
• Use of a lower gain antenna

Slide No.41
 Isolation
Isolation between transmitter and receiver antennas is required to avoid
receiver desensitisation, which is a reduction in receiver sensitivity. This is
caused by :-
• Receiver in-band noise caused by the co-site transmitter (spurious signals)
• Gain reduction of the low-noise amplifier caused by an strong off-channel
signal

Techniques used for isolation are :-

• Decoupling of the antennas by adequate spacing

• Filtering the transmitter’s out of band channel noise by multicouplers,
duplexers and isolators

Slide No.42
 Isolation
Horizontal Spacing
The isolation A(h) between 2 horizontally separated antennas is given by the
empirical formula :-

A(h) = 31.6 + 20 log d – (Gt + Gr) dB for 900 MHz

A(h) = 37.6 + 20 log d – (Gt + Gr) dB for 1800 MHz

Vertical Spacing
The isolation A(v) in dB is given by :-

A(v) = 47.3 + 40 log d dB for 900 MHz

A(v) = 59.3 + 40 log d dB for 1800 MHz

Slide No.43
 Service Contour
The propagation prediction model provides the signal level in terms of dBm.
This is the median value, e.g. –88 dBm

Given the standard deviation, there is a certain probability (e.g. 95%) that the
signal in a given area will be at least a number of X dB below the median value
of that area. Thus, with a 95% reliability, the signal level can only be
guaranteed top be –102 dBm (or more) which is the receiver sensitivity of the
mobile.

The signal contour for a specified receiver sensitivity must be plotted around
the cell site to define the coverage area. This contour is a statistical boundary.

If the MS travels along the boundary, for 95% of all the locations it is expected
to receive a signal that is above –102 dBm.

Slide No.44
 Cell Structure Planning
A homogeneous cell structure is practically impossible. However it is desirable
to design a cell structure as homogeneous as possible. This will lead to :-

• Reliable coverage
• Simple frequency planning
• Easy calculation of traffic loads
• Reliable handovers

Slide No.45
 Cell Structure Planning
Good cell structures can be planned by keeping the following points in mind :-

• Use as homogeneous a cell structure as possible (no abrupt changes in cell

size, e.g. at the edge of towns)
• Avoid random pointing of antenna direction. The front lobe at any BTS
directional TX antenna should illuminate only the back lobe of its co-channel
counterpart
• Define cell boundaries firmly. Avoid areas with many equally good server,
resulting in many handovers and many interferers
• Sufficient overlapping zones
• Avoid cell boundaries across traffic hot spots
• Keep all antenna heights about the same

Slide No.46
 Cell Structure Planning
Once a BTS is located through site establishment, and good coverage can be
achieved, there is no guarantee that the cell will maintain its original coverage.
Cells are living because :-

• New buildings may be erected within the coverage area

• Existing building may be demolished
• Trees are also a concern, when they grow across the LOS radio path

Slide No.47
 Cell Structure Growth
Network growth can be required for the following reasons :-
• Extension of coverage area
A new coverage area needs to be added
• Capacity increase
The traffic density in an existing cell has grown
• Coverage quality increase
For example, existing outdoor coverage needs to be upgraded to indoor
coverage

Integration of each new BTS or even each TRX has to be carefully planned into
the greater system. In all cases, the existing cells adjacent to the growth area
will be affected in the following aspects :-
• Changes in cell size and shape
• Changes in the BSS parameters
• Updates in neighbour list
• Frequency allocation
• Interference performance
Slide No.48
 Coverage Quality and Capacity Increase
If the number of available channels is fixed, the basic cellular principle required
that capacity increase is achieved by reusing frequencies more often over a
certain coverage area. Hence more sites are needed within the existing area.
This is accomplished by reducing the cell sizes in areas of high demand :-

• This requires the creation of new small cells within the overall cluster pattern
• Frequency reuse must not infringe on rules determining frequency allocation
for the large pattern
• Some coverage quality improvement can be expected as well

Slide No.49
 Coverage Quality and Capacity Increase
Increasing the cell density in a coverage area can be achieved by :-

• Adding more sites in the coverage area

• Cell splitting (sectorisation)

The capacity increases while the number of sites remains the same
– The size of the small cell is dependent on 2 factors:-
• Radio aspect
• Capacity of the system
– Certain channels should be used as barriers

• Cell Splitting

Slide No.50
 Coverage Limited System
In a noise limited cell, there is a limitation due to SNR limitations only. This is also
called coverage limitation

• No interference (C/I is good)

− Co-channel interference
− Adjacent channel interference

• No traffic congestion

Slide No.51
 Coverage Extension

The coverage can be increased by one or a combination of the following

actions :-
• Increase transmitted power. Doubling the power gives a gain of +3dB
• Increase BTS antenna height. Doubling the height may give +6 dB gain
• Use a high gain or a directional antenna at BTS
• Lower the threshold level of a received signal
• Install a masthead amplifier
• Decrease the front-end noise figure F (low noise receiver)
• Use a diversity receiver
• Select proper BTS site locations
• Use enhancers or micro/pico cells to enlarge coverage or to fill in holes
• Engineer the antenna pattern

Slide No.52
 Filling Coverage Holes
In areas where the traffic intensity is low, its is not cost effective to install a BTS.
An enhancer can be use to fill these coverage holes at low investments. Savings
are installation and operational costs.
Two types of enhancers are distinguished :
• Wideband
• Channelised
The enhancer can be considered as a relay, that receives at a low height and
transmit to a higher height and vice versa. Aspects:-
• The antenna pointing to the cell site BTS is directional
• The lower antenna is omni or directional
• Enhancers do not improve the SNR, they have only a relay function
• Repeater gain 10 – 85 dB adjustable
• Typical repeater range 0.5 – 3 km
• Interference aspects may make implementation difficult
• Ring oscillation shall be avoided
• Distance to serving BTS site as small as possible to avoid spread of power
into a large area in the vicinity of BTS and beyond
Slide No.53 • Enhancers may impact the network of another operator
 Interference
The C/I ratio can be increase in a number of different ways :-

• Good frequency management chart

− Grouping the channels into subsets

• Intelligent frequency assignment

− Allocation of specific channels to cell sites and MS, avoiding problems
from co-channel and adjacent channel interference

• Selection of a proper channel

− Among a set of assigned channels to a particular MS
− If the quality of the signal is poor, an intracell handover to another
frequency or time slot should occur

• Frequency hopping
− Effective on uplink and downlink path
− Choose different hopping sequences for co-channel cells, resulting in a
different interferer from hop to hop
Slide No.54
 Interference
• Antenna pattern design
− In some directions a strong signal is required, in other directions no signal
may be needed

• Tilting of antenna patterns

− To confine energy within a small area
− Downward tilt of directional antenna

• Reduction of antenna height

− Reducing interference is as important as radio coverage

• Power reduction of interfering transmitter

− RF power control, adaptive power control to keep transmission power as
low as possible, on a per time slot basis
− DTX, interrupted transmission during gaps in speech

• Choosing cell site location

Slide No.55
 Planning the frequencies
The frequency plan can be made in different ways :-
• Fixed cluster configuration
– For example, cluster of K = 21 cells will use 21 frequencies (at least). This
fixed frequency planning can be done manually. It is simple but not
particularly efficient

• Flexible assignment
– Based on the interference matrix using an automatic tool. In general, this
method can lead to a more efficient frequency use, e.g. 18 frequencies
doing the job instead of the fixed K = 21 frequency cluster size for the
same level of coverage quality.

• A mix of these methods is also possible

– Control channels are always transmitted at maximum power. The basic
idea is to protect the BCCH frequency
– It is a good solution to use first e.g. f1, ..., f21 for the control channels on a
safe K = 21 cell cluster, and then let the other frequencies be at a closer
range, determined by the interference matrix
Slide No.56
 Extension and Frequency Changes
When a network is to be extended, e.g. by increasing the cell density in order to
improve the traffic capacity and the coverage quality, a revised frequency plan is
necessary
• To minimise the re-tuning, the already operational base station should be left
unchanged as much as possible
• The pre-assigned frequencies of the cell cliques that will change significantly
should be abandoned in favour of new frequencies

It is convenient to define a set of frequency group

• Initially, each cell starts with a layer of a particular frequency group
• In a later stage, new frequencies from other layers of the same frequency
group can be added in that cell.
• No interference analysis is required
• It is possible however that frequencies from adjacent layers in differnt groups
can be adjacent channels. This needs to be verified
• If the frequency planning is performed by a computer tool, the frequency
group are of less importance
Slide No.57