Cellular Systems and Infrastructure-Based

Wireless Networks

• The coverage of these slides is based

on Rappaports book
• Chapter 3
• Version 1.0
• Please send comments and
corrections to
• Dr. Ali Muqaibel
• Ver 2.212

Chapter 3 Chapter 15
 Basics of cellular structure

Handoff

Interference

• Co-channel interference
Topics • Adjacent channel interference

Trunking

Coverage and Capacity

• Cell Splitting
• Sectoring
• Zone Microcells

2
 • Replace the single, high-power transmitter covering
a large area by many low-power transmitters
covering smaller areas (cells).
• The cellular concept provides the solution for the
capacity problem.
• The available frequencies will be re-used.
The Basic • This way, the number of available channels over the
Concept of service area equals the number of licensed channels
multiplied by the number of transmitters.
Cellularity

3
Components of Cellular Systems
• BS: Base station (BS) is the bridge
between the mobile station (MS)
and the MSC that relays signals MS
(control and information - voice or
data) from the (MS) to the MSC and
vice versa. MS
MS
• MSC: Controls a cluster of cells.
Base stations are connected to the
MSC via wireline or microwave MSC

links.
• Mobile station (MS)
• Mobile switching center (MSC)
• Base station (BS)
• PSTN: Public Switch Telephone Network
4
Forward vs. Reverse Links

• Forward Link (Downlink):

The transmission path from the BS to the MS. Forward
(Downlink)

• Reverse Link (Uplink):

The transmission path from the MS to the BS. Reverse
(Uplink)

5
Sending and Receiving
Types/Techniques

• Types:
• Simplex: one-way communication
• Half-Duplex: two-way but one-way
at a time
• Full Duplex: two-way
simultaneously
• Techniques:
• Frequency Division Duplex (FDD)
• Time Division Duplex (TDD)

6
FDD vs. TDD

7
The Implications of Cellularity

• Interference between re-used channels

• Mobility management issues
• Locating the users to initiate calls
• Call handover to maintain active calls.

8
Handling the Interference Issue
• Each transmitter (BS) is allocated a
subset of the total available channels.
• Neighboring cells are allocated
different sets of channels.
• Those sets of frequencies will be re-
used only far enough to minimize
interference level.
• This process is called frequency reuse
or frequency planning.
Illustration of the cellular frequency reuse concept. Cells with the same letter use the
same set of frequencies. A cell cluster is outlined in bold and replicated over the
coverage area. In this example, the cluster size, 𝑁, is equal to seven, and the
frequency reuse factor is 1/7 since each cell contains one-seventh of the total number
of available channels.

9
Channels Assignment Strategies
• Fixed Allocation: As in the previous example
• Simple
• Less efficient (higher blocking probability) if all the channels
allocated to a cell are occupied, a new call is blocked.
• Can be improved by implementing a borrowing strategy
(Controlled by MSC)
• Dynamic Allocation: When a call request is made, the serving
BS requests a channel from MSC.
• Less blockage
• Increased channel utilization
• More processing (channel occupancy, traffic distribution,…)
• MSC allocates a channel only if it is not being used by the cell or
any cells within a minimum distance.

10
The evolution of cellular
networks
• What are the main
characteristics of different
cellular network generations?
• 1G
• 2G
• 3G
• 4G
• 5G
• 6G
• …..
11
Why Hexagon Cells?
• A cell is an area over which a base station (BS) has a “good”
coverage.
• Actual cell footprint is irregular.
• A regular cell shape is needed for systematic design.
• An intuitive regular shape of a cell would be circular.
• Circles do not lend themselves easily for coverage analysis.
They will either overlap or leave uncovered areas.
• The only shapes that can be stacked perfectly are the
square, triangle and Hexagon.
• Hexagon is closest to circle, in addition to some other useful
features.

1
Antenna Positions Practical
Considerations Center-excited cells:
Omnidirectional antennas

• Antennas are placed at the center of

the cells (center excited cells)
• Or at three of the six cell vertices
(edge-excited cells) Edge-excited cells:
Directional antennas
• Practical considerations may not
allow the BS to be placed exactly as
they appear.
• Most system designs allows BS to be
positioned up to one fourth of the
cell radius away from ideal position.
2
2
Cellular Definitions and Notations 𝑅

• Cluster: set of neighboring cells that collectively and

distinctively use the available set of frequencies.
• Co-channel cells: cells in different clusters using same
group of frequencies.
• 𝑁 = the cluster size (in cells)
• 1/𝑁 = frequency re-use factor.
• 𝑅 :cell radius (distance from hexagon center to corner).
• 𝐷 : distance between centers of nearest co-channel cells.

3
 𝑆 = 𝑘𝑁
Cellular System Capacity 𝐶 = 𝑀𝑘𝑁 = 𝑀𝑆

▪ Let the total number of available radio channels = 𝑆,

▪ Let there be 𝑁 cells per cluster and each cell is allocated a group of 𝑘
unique channels, such that 𝑆 = 𝑘𝑁
▪ If the number of clusters in the service is 𝑀, then the total number of
channels (# possible simultaneous calls) 𝐶 = 𝑀𝑘𝑁 = 𝑀𝑆 (𝑀 clusters
in the area and each cluster has N cells and each cell has k unique channels)
▪ 𝐶 can be used as a measure of the system capacity in terms of the
total number of channels.
▪ If cluster size is reduced while keeping same cell size. More clusters
required to cover the area => (large 𝐶 but strong co-channel
interference)
4
Hexagonal Cluster Size
i j N • For hexagon cells, to have a uniform co-
channel distance 𝐷 for all cells in the
1 0 1 system, 𝑁 must obey the relation:
1 1 3
𝑁 = 𝑖2 + 𝑖𝑗 + 𝑗2 (𝑖 ≥ 𝑗 ≥ 0)
2 0 4
2 1 7 • Co-channel cells are identified from (𝑖, 𝑗)
2 2 12

5
Co-channel cells for N=7 (2,1) and N=19 (3,2)
Rule: move 𝑖 cells along any chain of hexagon and then turn 60 degrees counter
clock-wise (from the direction that you were moving) and move 𝑗 cells

6
Example 1
• Determine the number of channels per cell for the following cellular system for 𝑁 = 4, 𝑁 = 7, and 𝑁 = 12:
• A total of 33 𝑀𝐻𝑧 bandwidth is allocated to the system.
• It is divided into 50 𝑘𝐻𝑧 (voice/control) channels.
• One control channel per cell.
• Frequency re-use of control channels is 3 times less than voice channels.
• Solution:
• Total number of channels (voice or control) = 33000𝑘/50𝑘 = 660
• 𝑁 = 4:
• Channels reserved for control=4 × 3 = 12
• Every cell has =(660-12)/4=648/4 = 162 voice channels and one control channel
• 𝑁 = 7
• 21 channels reserved for control
• 639/7 = 91.3. Two cells have 92 + control, five have 91 + control.

7
Handoff (Handover)
• Passing an active call from one
BS to another without
disconnection.
• Basic concerns:
• When handoff is due?
• Where to handoff?
• How to handoff transparently?
• Management and billing

1
 Soft and Hard Handoffs
• Hard Handoff:
• Handoff involves moving a call to another channel and
another BS. (FDMA/TDMA systems)
• MS switches to a new channel after leaving the old one
(The old BS drops the MS before the new one acquires it).

• Soft Handoff:
• A call is moved to a different BS (CDMA).
• MS communicates with two BSs until handover is made.
Handoff Threshold Improper handoff situation

• The criterion for handoff is based on received 𝑃𝑟, ℎ𝑎𝑛𝑑𝑜𝑓𝑓

power level at the MS , i.e., received signal
𝑃𝑟, 𝑚𝑖𝑛𝑖𝑚𝑢𝑚
strength information (RSSI), which is inferred from
the channel measurements.
∆= 𝑃𝑟,ℎ𝑎𝑛𝑑𝑜𝑓𝑓 − 𝑃𝑟,𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑢𝑠𝑎𝑏𝑙𝑒
Proper handoff situation
• Measure
▪ 𝑃𝑟,ℎ𝑎𝑛𝑑𝑜𝑓𝑓 is the handoff threshold
▪ 𝑃𝑟,𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑢𝑠𝑎𝑏𝑙𝑒 is the minimum acceptable level
𝑃𝑟, ℎ𝑎𝑛𝑑𝑜𝑓𝑓
to maintain a call ∆
▪ ∆ threshold 𝑃𝑟, 𝑚𝑖𝑛𝑖𝑚𝑢𝑚
• ∆ should not be too large to avoid unnecessary
handoffs.
• ∆ should not be too small as it will provide very
small time for handoff and call can get dropped.
3
Who does the Measurement for Handoff?

• In 1G cellular systems, signal measurements are done by BS and supervised by MSC, BS

measures RSSI of calls in progress within the cell and reports to MSC
• The handoff process takes around 10 seconds. D is in the range 6 – 12 dB.

• In 2G cellular systems, handoff decisions are mobile assisted (MAHO), each MS measures
received power from surrounding BSs and reports to the serving BS.
• A handoff is initiated if power received for a neighboring BS exceeds the power received
from the serving BS (i) by certain threshold or (ii) over certain period of time.
• MAHO systems are faster. The handoff process takes 1-2 seconds. D is usually less than 6 dB.

Search for handoff in 3G, LTE and 5G

4
 A fraction of total channels is
reserved for handoff requests.
(More efficient with dynamic
channel assignment).
Prioritizing
Handoffs
Queuing of handoff requests
where handoff requests can be
given priority over new calls.

5
Practical Issue 1: False Handoffs
• Problem:
• Sometimes the drop in signal level is momentary (fading) and does not
require handoff.
• Solution:
• Monitor the signal level for some time to detect moving-away pattern.
• Averaging the measurements over some period may be useful as well.

6
Practical Issue 2: High Speed Users
• Problem: The Umbrella Cell Approach
• frequent handoffs.
• Solution: Umbrella cell.
• Large and Small cells co-located.
• High-speed users are served by
umbrella cell, while slow users
are served by the microcells.
• Sophisticated algorithms are used
to evaluate and partition users
according to their speeds

7
Practical Issue 3: Cell Dragging
• Problem:
• The signal stays strong even outside cell borders (e.g., because of the presence of
LoS path and slowly moving user)
• Creates potential interference and management problems.
• Solution:
• Handoff thresholds and coverage parameters must be adjusted. For example, the
values previously seen for 1G and 2G have been carefully calculated.
• Note: Handoff is not required to rescue calls only. It is also required for
proper overall system operation.

8
Intra-system and Inter-system Handoffs
• Intra-system: A handoff between BSs that are controlled by the same MSC.
• Inter-system: A handoff between BSs controlled by different MSCs.
• While Intra-system handoff are crucial in any cellular system, Inter-system are not
because they are not very frequent. It is acceptable that the call be disconnected
while trans-crossing systems (or operators).

9
Exercise: Handoff
1. MAHO stands for ______ 4. What was the typical handoff time in first
a) MSC assisted handoff generation analog cellular systems?
b) Mobile assisted handoff a) 1 second
b) 10 seconds
c) Machine assisted handoff
c) 1 minute
d) Man assisted handoff d) 10 milliseconds
2. Umbrella cell approach is possible by using _________ 5. Soft handoff is also known as _________
a) Antenna of same heights a) MAHO
b) Antenna of different heights b) Hand over
c) Different voice channels c) Break before make
d) Different control channels d) Make before break

3. Cell dragging is a problem occur due to __________

a) Pedestrian users
b) Stationary users
c) High speed mobile systems
d) Base stations having same frequency

10
Interference
Co-channel Interference (from users in other cells operating at the
There are many sources of interference, same frequency)
but mainly two: Adjacent-channel interference (from users within cell).

For voice channels, interference results in cross talk and background noise.

For control channels, interference results in missed or blocked calls.

For data channels, higher probability of error or less data rate.

1
 co-channel resuse ratio 𝑄 = 𝐷/𝑅
For a hexagon, 𝑄 = 3𝑁
Co-Channel Interference (CCI) and Capacity
• Assumption: All BSs transmit at the same power.
• Worst case analysis: CCI caused to a user on the
edge of the cell.
• CCI is a decreasing function of
co-channel resuse ratio 𝑄 = 𝐷/𝑅
1
• For a hexagon, 𝑄 = 3𝑁 2

• Small 𝑁 ➔small 𝑄 ➔ high CCI

• Small 𝑁 ➔large 𝑀 and large 𝑘 ➔ high capacity

Recall: 𝐶 = 𝑀𝑘𝑁 = 𝑀𝑆 (𝑀 clusters in the area and each

cluster has 𝑁 cells and each cell has 𝑘 unique channels)
2
Rough SIR Calculations
𝑆 𝑆
𝑆𝐼𝑅 = = 𝑖 (forward channel)
𝐼 σ 0 𝐼𝑖
𝑖=1
• 𝑆 = Signal power from desired BS
• 𝐼𝑖 = CCI power from the 𝑖𝑡ℎ co-channel BS
• 𝑖0 = the number of interferers
• As a rough calculation
𝑑 −𝑛
𝑃𝑟 (𝑑) = 𝑃0
𝑑0
• 𝑃𝑟 = received power at distance 𝑑 from transmitter.
• 𝑃0 = power at a reference point 𝑑0 from transmitter.
• 𝑛 = path loss exponent

3
 Rough SIR Calcuation
• Substituting for 𝑆 and 𝐼
𝑅−𝑛
𝑆𝐼𝑅 =
σ𝑖𝑖=1
0
𝐷𝑖−𝑛
• Assuming same distance 𝐷 from MS to all interferer
BSs, and considering only first tier CCI (𝑖0 = 6)
𝑛
(𝐷/𝑅)𝑛 3𝑁
𝑆𝐼𝑅 = =
𝑖0 6

4
Example 2: Draft Design
If the minimum 𝑆𝐼𝑅 for acceptable operation of a cellular system is 18 dB. Find the
frequency reuse factor. Assume 𝑛 = 4.
𝑛
(𝐷/𝑅)𝑛 3𝑁
𝑆𝐼𝑅 = =
𝑖0 6
1 𝑛
3𝑁 2 > 101.8
6
𝑁 > 6.5. 3𝑁
𝑛
Calculate 𝑆𝐼𝑅 = 6
Choose 𝑁 = 7. Using 𝑁 = 7, 𝑆𝐼𝑅 = 18.65 𝑑𝐵 > 18 𝑑𝐵
Frequency re-use factor is 1/7. (criteria fulfilled)

5
 Re-call
For a hexagon, 𝑄 = 3𝑁
SIR: Better Approximation
1
𝑆𝐼𝑅 =
2(𝑄 − 1)−𝑛 + 2(𝑄 + 1)−𝑛 + 2𝑄−𝑛
• For 𝑁 = 7, 𝑛 = 4:
• 𝑄 = 4.58, 𝑆𝐼𝑅 = 17 𝑑𝐵
• (Previous solution = 18.66 dB)
• To ensure a design guaranteeing worst case,
• SIR is to be increased,
• take 𝑁 = 9 (𝑖 = 3, 𝑗 = 0).
• 𝑄 = 5.2 , 𝑆𝐼𝑅 = 20 𝑑𝐵
Illustration of the first tier of co-channel cells for a cluster size of 𝑁 = 7. An
• This makes a reduction in capacity by 7/9. approximation of the exact geometry is shown here. When the mobile is at
the cell boundary (point 𝑋), it experiences worst case co-channel interference
on the forward channel. The marked distances between the mobile and
different co-channel cells are based on approximations made for easy
analysis.
6
Channel Assignment in CDMA Systems

• In CDMA systems the cluster size is 𝑁 = 1.

• A single 1.25 𝑀𝐻𝑧 channel carries the
simultaneous transmission of a single control
channel and up to 64 voice channels.
• In some ill-behaved propagation conditions a
simple 𝑓1/𝑓2 planning scheme is used:
• Nearest neighbor cells use a frequency channel
that is different from its closest neighbor in
particular location/direction.

7
 The level of
CCI interference
interference is a
is coming from
function of
other users in
number of
the same cell.
active users.
Interference
in CDMA
Systems The coverage
Power levels
and thresholds
range is have to be
dynamic / time adjusted
varying. according to the
traffic intensity.

8
Adjacent Channel Interference (ACI) receiving filter
response

signal on adjacent signal on adjacent

channel channel

desired signal

• 𝐴𝐶𝐼: Results from imperfect filters which

allow nearby frequencies to leak into the FILTER
interference desired signal
interference

working passband.
• Near-far effect, due to: 𝑓2
• 𝑓1 and 𝑓2 being near in band
𝑓1
• One signal being orders of magnitude stronger
than the other
• It results in 𝑓2 capturing the receiver of 𝑓1 .
𝑓1
f2

𝑓2

9
 Filtering
• High quality filters.

Channel assignment
http://wifinigel.blogspot.com/2013/02/adjacent-channel-interference.html
• Channels assigned to a cell are maximally separated.
• Sequential assignment: Guarantees 𝑁 channel separation.
Solutions to • We can even prevent second tier interference by not
assigning adjacent frequencies to adjacent cells (feasible for
ACI large 𝑁).
• More efficient when 𝑁 is large.

Power Control
• Transmit with the smallest power necessary to maintain
good quality.
▪ Reduces interference
• Prolongs battery life

10
Trunking
A technique where a pool of channels are used to serve larger
number of users on demand.
N(t) all trunks busy

1
2
3
4
5
6
7
1
Trunking and Grade of Service
• Trunking refers to allowing a large number of users to share a small number
of channels.
• Trunking exploits the statistical behavior of users so that a small number of
channels may accommodate a large, random user community.
• Trunking theory was developed by a Danish Math. Erlang.
• Erlangs: One Erlangs represents the amount of traffic carried by a channel
that is completely occupied.
– Ex: A radio channel that is occupied for 30 minutes during an
hour carries 0.5 Erlangs of traffic.
• Grade of Service (GOS): The likelihood that a call is blocked (Erlang B) or
delayed greater than a certain queuing time during (Erlang C) the busiest hour.
• Offered traffic and carried traffic.
Traffic Terminology
• Set-up Time: the time required to allocate a trunked radio channel to
a requesting user.
• Blocking: A call is denied service due to lack of channels.
• Traffic intensity: measure of channel time utilization in Erlangs (𝐸).
Denoted by 𝐴 (total) and 𝐴𝑢 ( per user)
• Holding Time: Average duration of a typical call. Denoted by 𝐻 (in
seconds)
• Request Rate: The average number of call requested per unit time.
Denoted by 𝜆 in 𝑠𝑒𝑐𝑜𝑛𝑑𝑠 −1

3
 𝑁 channel
Offered traffic: 𝐴𝑜𝑓𝑓𝑒𝑟𝑒𝑑 Carried traffic: 𝐴𝑐𝑎𝑟𝑟𝑖𝑒𝑑
Traffic Modeling
Lost Traffic= 𝐴𝑜𝑓𝑓𝑒𝑟𝑒𝑑 − 𝐴𝑐𝑎𝑟𝑟𝑖𝑒𝑑 = 𝑃𝑏 𝐴𝑜𝑓𝑓𝑒𝑟𝑒𝑑
▪ Offered traffic = 𝐴𝑜𝑓𝑓𝑒𝑟𝑒𝑑 𝐴𝑢 = 𝜆 × 𝐻 𝑃𝑏 probability of a call being blocked

▪ For a system containing 𝑈 users, total offered traffic , 𝐴 = 𝑈𝐴𝑢

▪ Carried Traffic = 𝐴𝑐𝑎𝑟𝑟𝑖𝑒𝑑 = {1 − 𝑃𝑏 } × 𝐴𝑜𝑓𝑓𝑒𝑟𝑒𝑑
▪ 𝑃𝑏 probability of a call being blocked.
▪ General Assumptions:
▪ Memoryless arrivals of requests
▪ Calls arrive according to Poisson Distribution
𝜆𝑡 𝑘 −𝜆𝑡
𝑃 𝑁 𝑡 =𝑘 = 𝑒
𝑘!
• Call duration follows an exponential distribution.
• Infinite number of users (or much greater than the number of trunks)
4
Models of Trunking Systems
• Blocked Calls Cleared (BCC): Blocked calls not queued
• Erlang’s B Formula: Probability a call is blocked
𝐴𝐶 𝐴𝑘
𝑃𝑏 = / σ𝐶𝑘=0
𝐶! 𝑘!

• 𝐴 =offered load at busiest hour (BH); 𝐶=# of trunks

• Blocked Calls Delayed (BCD): Blocked calls queued
• 𝜏 is the time to get channel access
• Erlang’s C Formula: Probability a call is queued
𝐴𝐶
Pr[ 𝜏 > 0] = 𝐶
𝐴 + 𝐶! (1 − 𝐴/𝐶) σ𝐶−1
0 𝐴𝑘 /𝑘!
𝐶−𝐴 𝑡
𝑃𝐷 𝜏 > 𝑡 = Pr 𝜏 > 0 × Pr 𝜏 > 𝑡 𝜏 > 0 = Pr 𝜏 > 0 × exp −
𝐻
𝐻
• Average delay of all calls (𝐷)= Pr[𝜏 > 0] × 𝐶−𝐴
5
System Capacity
• System capacity can be defined as:
• The maximum number of channels
• The maximum number of subscribers

6
 𝐶
Erlang B Trunking GOS 𝑃𝑏 =
𝐴𝐶
𝐶!
/ ෍
𝐴𝑘
𝑘!
𝑘=0

7
 𝐶 = 19

Erlang B

𝐺𝑜𝑆 = 0.02

For 𝐺𝑜𝑆 = 0.02

and 𝐶 = 19 ⇒
𝐴 = 12 𝐸

𝐴 = 12

8
 Erlang C

From Erlang C curves,

A (15, %5 delay) = 9 E

From Erlang C curves,

A (35, %10 delay) = ? E

9
Example 3: Turnking
• A cell has 19 channels.
• Find the number of users that can be supported at 2% blocking
probability, if each user makes two calls per hour on the average, and
the average call duration is 3 minutes.
• Answer:
• For 𝐺𝑜𝑆 = 0.02 and 𝐶 = 19 ⇒ 𝐴 = 12 𝐸.
3
• Traffic intensity per user 𝐴𝑢 = 2 × = 0.1𝐸
60
• Number of users supported in a cell (𝑈) is
𝐴
𝑈= = 120
𝐴𝑢 10
Trunk Utilization Load (E) # Trunks U
1 5 0.20
𝑈 = (1 – 𝑃𝑏 )(𝐴/𝐶) 2 7 0.29
4 10 0.40
• Table produced at 1% blockage.
8 15 0.53
• Observations: 10 18 0.56
• The larger the offered load the better
30 42 0.71
the utilization (to some extent)
• Increased efficiency in sharing resources 50 64 0.78
• However, small 𝑁 results in higher CCI. 60 75 0.80
90 106 0.85
100 117 0.85

11
Capacity Trade offs
• Which is favorable from capacity perspective:
• Send 10 E over 16 trunks, or
• Send 20 E over 32 trunks?
• From Erlang-B Table:
• 𝑃𝐵 (10E,16 trunks) = 2%
• 𝑃𝐵 (20E,32 trunks) = 0.5%
• For 32 trunks and 2%, you can carry 23.7 E (18% more users)
• For a given service area, and a given cell area (transmitted power), it is favorable form
capacity prospective to have small cluster size 𝑁 for two reasons:
• Each channel is used more number of times (More number of clusters)
• More channels (trunks) per cell (higher supported traffic).

However, small 𝑁 results in higher CCI.

12
Example 4: Erlang B
• Service Area = 1,250 sq miles.
• N = 7; R = 4 miles
• Total system BW = 40 MHz
• Full duplex channel BW = 50 kHz
• One control channel per cell. No
frequency re-use.
• Traffic per user = Au = 0.03 E
• Required GoS = 0.01 (Erlang B)
• Find:
• Maximum number of users that can
be served at one time.
• Traffic intensity of each cell.
• Maximum number of subscribers
that can be supported within this
GOS?
• Number of subscribers per channel
 Cell Area = 2.598×R2 = 41.57 sq. mi.
 Number of cells = 1250/41.57 = 30
 Number of channels = 40M/50k=800
 Total voice channels = 800 − 30 = 770
 Number of voice ch. per cell = 770/7=110
 Ans. to first question: 110 × 30 = 3,300
 Traffic intensity/cell = A(110, 0.01) = 90 E
 Max. # of subscribers in a cell = 90/0.03 = 3,000
 Maximum number of system subscribers
assuming uniformly distributed over the cells
= 3000 × 30 = 90,000
 Max. % of active users = 3,300/90,000 = 3.6%
of total subscribers.
 # of subscribers / channel = 90,000/770 = 117
1
Example 5: Practice Cellular System Design
• Service Area = 1,300 𝑠𝑞 𝑚𝑖𝑙𝑒𝑠.
• 𝑁 = 7; 𝑅 = 4 𝑚𝑖𝑙𝑒𝑠
• Total system BW = 40 𝑀𝐻𝑧
• Full duplex channel BW = 60 𝑘𝐻𝑧
• Traffic per user = 𝐴𝑢 = 0.03 𝐸
• Required GoS = 0.02 (Erlang B)
• Find:
• Maximum number of users that can be
served at one time.
• Traffic intensity of each cell.
• Maximum number of subscribers that
can be supported within this GOS?
• Maximum number of active users
relative to the total number of
subscribers.
 Cell Area = 2.589×R2 = 41.57 sq. mi.
 Number of cells = 1300/41.57 = 31
 Number of channels = 40 MHz/60KHz =
666.66 (repeated every 7 cells)
 Number of channels per cell = 666.66 /7 = 95
 Ans. to first question: 95 × 31 = 2,945
 Traffic intensity/cell = A(95, 0.02) = 84 E
 Max. # of subscribers in a cell = 84/0.03 = 2800
 Maximum number of system subscribers
assuming uniformly distributed over the cells
= 2800 × 31 = 86800
 Max. % of active users = 2945/86800 = 3.39%
of total subscribers.
2
Example on Erlang C
• 𝑁 = 4; 𝑅 = 1.387 𝑘𝑚
• Total voice channels = 60
• 𝐴𝑢 = 0.029 E; l = 1 call/hour
• GoS = 5% probability delay
• Find:
• How many users/km2 the
system supports?
• Probability a delayed call will
wait more than 10 sec.
• Probability a call will be
delayed more than 10 sec.
 Area of cell = 2.589 𝑅2 = 5 km2
 Number of channels/cell =60/4= 15
 From Erlang C curves,
A (15, %5 delay) = 9 E
 Supported Number of users in a cell = 9/0.029
= 310
 Number of users/km2 =310/5= 62
𝐴𝑢
 For 𝜆 = 1, 𝐻 = = 0.029 ℎ𝑜𝑢𝑟 =104.4 s.
𝜆
𝐶−𝐴 10
 Pr[delay >10|delay]=exp − =
𝐻
56.29%
 Pr[𝑑𝑒𝑙𝑎𝑦 > 10] = 0.05 × 0.56 = 2.8%

𝐶−𝐴 𝑡
Pr 𝜏 > 0 × exp − 3
𝐻
 Cell Splitting: Increase the
number of BSs

Improving
Sectoring: Antenna direction
Coverage and adjustment to reduce CCI.
Capacity

Zone Microcell: Antenna power

adjustment to reduce CCI.

1
1. Cell Splitting
• Subdivide congested cells into
smaller cells (microcells), each
with its own BS, by:
• Reduction in antenna height
• Reduction in transmitted power.
• Capacity is increased by
increasing the number of times
the (channel) frequency is used
(more clusters for the same
service area).

2
 Cell Splitting Practice the example in the book
Example 3.8 , page 89

• Both 𝑅 and 𝐷 are reduced at the same rate. 𝑄 is not

changed. 𝑆𝐼𝑅 not changed.
• Depending on the new designed cell radius, the
transmitted power should be re-adjusted such that:
𝑃𝑟 [at new cell boundary] = 𝑃𝑟 [at old cell boundary]
• 𝑃𝑟 is proportional to 𝑃𝑡 𝑅−𝑛
• If the cell radius is reduced to 𝛼𝑅, the transmitted
power should be reduced by 𝛼 𝑛
That is 𝑃𝑡,2 /𝑃𝑡,1 = 𝛼 𝑛
• Example: 𝑅 reduced by half, 𝑛 = 4 ➔ 𝑃𝑡,2 /𝑃𝑡,1 =1/16
3
 For reducing the radius by half, we are reducing cell
area by 4; multiplying the number of clusters in the
service area by 4.

How much
Capacity
Therefore, capacity (number of channels) is
Enhancement is quadrupled.
Achieved by cell
splitting?

At what cost?
Increasing the numbers of BSs More handoffs (sophistication
(by a factor of 4). and processing)

4
Different-cell-size
Systems

• Not all cells have (or need) to be

split.
• In this case, different-size cells
exists (different transmitter
powers).
• Channel assignments becomes
complicated.

5
2. Sectoring

• Using directional antennas.

• Common sectoring patterns: 1200 and 600
• The channels assigned to a cell are divided 120° Sectoring
between the sectors.

60° Sectoring

6
 Sectoring
• By using directional antennas, we reduce the
number of co-channel interferers.

• For 1200 sectoring, 1st tier interferers reduced from

6 to 2.
𝑛
(𝐷/𝑅)𝑛 3𝑁
𝑆𝐼𝑅 = =
𝑖0 2

• Improvement of 10 log (3) ≈ 5 𝑑𝐵

(using better approx. ≈ 6.4 dB).

• We can translate this improvement into a reduction

in 𝑁, which implies more clusters, i.e. more number
of uses of the same frequency.

With no sectoring
𝑛
(𝐷/𝑅)𝑛 3𝑁
𝑆𝐼𝑅 = =
𝑖0 6 7
Illustration of Same CCI Scenarios
Omnidirectional Directional
6 distant interferers 2 near interferers

C
A B

D
B A
D C
B A B
C D C
A B A
D C
B

8
Example: Sectoring
To achieve minimum SIR of 21 dB, assuming path loss exponent = 4 :

• No Sectoring: With no sectoring

𝑛
• Smallest cluster size is 𝑁 = 12 (𝐷/𝑅)𝑛 3𝑁
𝑆𝐼𝑅 = =
𝑖0 6
• 120o Sectoring: With 120° sectoring
• Smallest cluster size is 𝑁 = 7 (𝐷/𝑅)𝑛 3𝑁
𝑛

𝑆𝐼𝑅 = =
• 12/7 increase in capacity 𝑖0 2

9
How much capacity enhancement is 𝑆𝐼𝑅 =
𝑛
3𝑁
𝑖0
achieved? using better approx. ≈ 6.4 dB

• For 𝑛 = 4, we can reduce 𝑁 by a factor of 100.64 = 2.09. For example,

from 𝑁 = 7 to 𝑁 = 4.
• We enhanced the capacity (number of channels) by 1.75 times, BUT ...
• Capacity (number of users) is not enhanced by the same ratio.
• The number of channels per cell is increased (due to smaller 𝑁)
• The three sectors will divide the cell channels among them.
• The number of channels per sector is less than the number of channels in a cell
before sectorization.
• This results in a decrease in trunking efficiency, which leads to lowering the number
of subscribers to keep the same GoS.

10
 Example
• Given:
• number of available voice channels = 400, 𝑛 = 4, 𝑁 = 7, 𝐴𝑢
= 0.03 𝐸. We require 𝐺𝑜𝑆 = 0.01. What is the capacity improvement
in terms of # of channels and # of users?
• Unsectorized design:
• number of channel/cell = 400/7 = 57
• For 𝐺𝑜𝑆 = 1% 𝑏𝑙𝑜𝑐𝑘𝑖𝑛𝑔, 𝐴 = 44.2 𝐸𝑟𝑙𝑎𝑛𝑔𝑠
• For 𝐴𝑢 = 0.03 𝐸, number of users/cell = 44.2/0.03 = 1473
• For the sectored system, N = 4.
• Number of channel/cell = 400/4=100.
• Number of channels/sector = 33.
• For GoS = 1%, 𝐴 = 22 E. Number of users/sector = 733, per cell =
2199
• Capacity enhancement in terms of
number of users = 2199/1473 = 1.5
• While Capacity enhancement in terms
of number of channels = 7/4 = 1.75
• Note that in this example the sectored
system still provides marginal SIR Over
the un-sectored system.
• If we decide to utilize sectorization for
the sole purpose of increasing SIR,
then the sectored system will result in
reduction in number of subscribers.
11
Cell Sectoring: Pros and Cons
• Advantages:
• Improves the SIR of the system
• A gain of ~ 7 𝑑𝐵 is achieved over omni-directional systems when 120°
sectoring is used
• Problems:
• Increased handoff requirements (from one sector to another sector)
• Multiple antennas are required in a BS
• Decreased trunking efficiency (less # channel per sector)

12
 3. Microcell Zone Concept
• The high-power single cell transmitter is replaced by several
lower power transmitters.
• Each transmitter defines a microcell.
• All microcell transmitters are connected to the cell BS (via fiber,
coax, or microwave).
• A mobile in a cell is served by one zone at any instant of time.
• Cell channels are in a pool for all zones (no sub-trunking).
• Once a channel is assigned to a zone, it is not assigned to any
other zone.
• A mobile user traveling from one zone to another in the cell
retains the same frequency.
• Handoff is not required by MSC when crossing zones; the BS
simply hands the mobile from one transmitter to the other
without changing the frequency.

13
Why Capacity Enhancement is achieved by Microcell Zone?

• Because a frequency is used in one zone only (part of the cell) at any
instant of time at a lower power, the co-channel interference is generally
reduced.
• Put it differently, for the same SIR of the normal system, we can have
smaller cluster size, i.e., more frequency re-use, leading to increased
capacity.
• Unlike sectoring, no sub-trunking is done within the cell. Therefore, the
gain from reducing cluster size is intact.

14
Example: Microcell Zone
Define 𝐷, 𝐷𝑧 , 𝑅, and 𝑅𝑧 for a microcell architecture with 𝑁 = 7. The smaller
hexagons form zones and three hexagons (outlined in bold) together form a
• The minimum required 𝑆𝐼𝑅 for system operation is 18 𝑑𝐵. cell. Six nearest co-channel cells are shown.

• Consider a 7 𝑐𝑒𝑙𝑙 structure.

1
𝑆𝐼𝑅 =
2(𝑄 − 1)−𝑛 + 2(𝑄 + 1)−𝑛 + 2𝑄 −𝑛
• 𝑄 = 𝐷 / 𝑅 = 4.6
• If we maintain 𝐷𝑧 /𝑅𝑧 = 4.6, we maintain 18 𝑑𝐵.
• Let’s approximate circles by hexagons in the following sketch.
• 𝑅𝑧 = 𝑅/2
• 𝐷𝑧 ≈ 𝐷 (3/4)
• For 𝐷𝑧 /𝑅𝑧 = 4.6 → 𝐷/𝑅 ≈ 3 (show)
• 𝑄 = 𝐷/𝑅 = 3𝑁 So, 𝑁 = 3, 𝑁 reduces from 7 to 3.
• Channel expansion= 7/3
• User Expansion = 7/3
• Cost: Smarter BSs.
• Exercise: repeat the analysis for 𝑆𝐼𝑅 24𝑑𝐵.
15