GSM Network Performance Management and Optimisation

GSM
Network Performance Management

and
Optimisation
Course Objectives
• Understand the types of problems experienced in a GSM network and
why they occur.
• Understand the variety of different tools available to the optimisation
engineer.
• Develop and explain an optimisation process for GSM networks.
• Identify suitable KPI’s which could be used to highlight poorly performing
cells.
• Use statistics to identify performance problems in the BSS and methods
of rectifying these problems.
• How to identify service affecting BSS database and neighbour relation
issues.
• Dimension GSM networks with a view to Multi-Service deployment.
• Understand the characteristics of NSS Topology and Signalling.
• Know how non-BSS issues can affect the network and be identified.
Course Outline
1. Review of GSM Principles

2. Network Characteristics & Problem Types
3. Introduction to Performance Management
4. Performance Management Metrics
5. Measuring Network Performance - Drive Testing
6. Measuring Network Performance – The OMC
7. The Optimisation Process
8. BSS Optimisation Activities
9. NSS Characteristics
10. NSS Parameter Evaluation and Optimisation
11. Dimensioning Network for new Services
Section 1
Review of GSM Principles

Section 1
1.1 Introduction
1.2 Physical Channel Structure
1.3 Logical Channel / Multiframe Formats
1.4 Review of the BCH Carrier
1.5 Paging Procedures
1.6 Timing Advance
1.7 Cell Reselection
1.8 Handovers
1.9 Power Control
1.10 Frequency Hopping

P-GSM Physical Channels
890 915 935 960 MHz
Uplink Downlink
Duplex spacing = 45 MHz

Fu(n)
Range of ARFCN: Guard Band
1 - 124 100 kHz wide
1 2 3 4 n
Guard Band
1 frame period
100 kHz wide Channel Numbers (n) (ARFCN)
4.615 ms
200 kHz spacing
0 1 2 3 4 5 6 7
Raw data rate =
33.75kbps per traffic channel
timeslot = 0.577 ms 270kbps per carrier channel
DCS - 1800 Spectrum
1710 1785 1805 1880 MHz
Uplink Downlink
Duplex spacing = 95 MHz
Fu(n)
Range of ARFCN:
512 - 885
1 2 3 4 n
Guard Band
100 kHz wide
Guard Band
100 kHz wide
Channel Numbers (n) (ARFCN)
200 kHz spacing
GSM Logical Channels
• Two types of logical channel are defined; traffic and control channels
• Each is further sub-divided as shown:
Traffic Control
TCH BCH CCCH DCCH
FCCH PCH SDCCH

TCH/F
SCH RACH SACCH

TCH/H
BCCH AGCH FACCH
CBCH
NCH
Traffic Channel Multiframe
• The TCH multiframe consists of 26 timeslots.
• This multiframe maps the following logical channels:
• TCH
• SACCH
• FACCH
• TCH Multiframe structure:
T T T T T T T T T T T T S T T T T T T T T T T T T I
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
T = TCH S = SACCH I = Idle

FACCH is not allocated slots in the multiframe. It steals TCH slots when required -
indicated by the stealing flags in the normal burst.
Control Channel Multiframe
• The control channel multiframe is formed of 51 timeslots
• CCH multiframe maps the following logical channels:
Downlink: Uplink:
•FCCH •RACH
•SCH
•BCCH
•CCCH (combination of PCH and AGCH)
Downlink F = FCCH S = SCH I = Idle
F S BCCH CCCH F S CCCH CCCH F S CCCH CCCH F S CCCH CCCH F S CCCH CCCH I
0 1 2-5 6-9 10 11 12-15 16-19 20 21 22-25 26-29 30 31 32-35 36-39 40 41 42-45 46-49 50
RACH
Uplink
• Control channel messages generally require 4 timeslots.

Control Channel Configuration
• On the downlink, CCCH consists of paging (PCH) and access grant (AGCH)
messages
• A combined multiframe has only 3 CCCH blocks to allow for SDCCH and
SACCH:
SDCCH SDCCH SDCCH SDCCH SACCH SACCH
F S BCCH CCCH F S CCCH CCCH F S 0 1 F S 2 3 F S 0 1 I
• A non-combined multiframe has 9 CCCH blocks:

F S BCCH CCCH F S CCCH CCCH F S CCCH CCCH F S CCCH CCCH F S CCCH CCCH I
SDCCH SDCCH SDCCH SDCCH SDCCH SDCCH SDCCH SDCCH SACCH SACCH SACCH SACCH
0 1 2 3 4 5 6 7 0 1 2 3 I I I
• A complete paging or access grant message takes four bursts (timeslots),

i.e. one CCCH block.
Control Channel Configuration
CCCH_CONF Number of CCCH Configuration

Timeslots
000 1 TS0 (non-combined)
001 1 TS0 (combined)
010 2 TS0 (non-combined),TS2
100 3 TS0 (non-combined),TS2,TS4
110 4 TS0 (non-combined),TS2,TS4,TS6
BCCH Carrier Characteristics
• Each cell has a designated BCCH carrier carrying the control
physical channel and BCCH logical channel
• All BCCH carriers transmit continuously on full power by filling all
timeslots with traffic, signalling or dummy bursts
• BCCH carriers use maximum frequency re-use distance to reduce
interference
• TS 0 of the BCCH carrier contains the cell’s physical control
channel
• The physical control channel comprises a number of logical
control channels
• Additional physical channels on the BCCH may be required for
extra control channel capacity.
• Remaining physical channels on BCCH carrier carry traffic
channels
Paging Procedures
• Paging locates MS to cell Level for call
routing
• Three paging message types:
• Type 1 - 2 MSs using IMSI/TMSI
• Type 2 - 3 MSs (1xIMSI, 2xTMSI)
• Type 3 - 4 MSs using TMSI only
• Paging message requires 4 bursts (1 CCCH
block)
• Paging messages may be stored at BSS
• Transmitted on PCH
• If DRX is implemented MS listens only to
allocated paging group
Timing Advance
• Signal from MS1 takes longer to arrive at BTS than that from MS2
• Timeslots overlap - collision
1 2 3 1 2 3
MS1 - Timeslot 1 time
1 2 3 1 2 3
MS2 - Timeslot 2 time time
• Timing Advance signal causes mobiles further from base station to transmit
earlier - this compensates for extra propagation delay
1 2 3 1 2 3
MS1 - Timeslot 1 time
1 2 3 1 2 3
MS2 - Timeslot 2 time time
Timing Advance
Timing Advance
• Timing Advance is calculated from delay of data bits in the access burst
received by the base station - long guard period allows space for this delay
Access burst data Guard Period
delay
Access burst data
• TA signal is transmitted on SACCH as a number between 0 and 63 in

units of bit periods (0 and 219 for GSM 400).
• TA value allows for ‘round trip’ from MS to BTS and back to MS
• Each step in TA value corresponds to a MS to BTS distance of 550 metres
• Maximum MS to BTS distance allowed by TA is 35 km (GSM 900)
Timing Advance
• Timing Advance value reduces the 3 timeslot offset
between downlink and uplink
0 1 2 3 4 5 6 7
Downlink
Delay
3 timeslots Uplink
0 1 2 3 4 5 6 7
Timing
Advance
Uplink
0 1 2 3 4 5 6 7
Actual delay
• TA value updates are sent to the MS on the SACCH

• The Timing Advance technique is known as adaptive frame
alignment
Section 1 – GSM Principles
BCCH Allocation (BA) Lists
• Contains a list of BCCH carriers available in a certain

PLMN region
• Passed to MS and stored in the SIM during IMSI Attach
procedure
• MS monitors all carriers in BA list for signal strength.
• In Idle mode this is used to reselect cell whilst moving
• In dedicated mode, the measured values of the six best
neighbouring cells are reported to the serving BSC for
handover purposes
BA Range List
8 7 6 5 4 3 2 1
BA Range IEI Octet 1
Length of BA Range Contents Octet 2
Number of Ranges Octet 3
RANGE1_LOWER Octet 4
RANGE1_LOWER RANGE1_HIGHER Octet 5
RANGE1_HIGHER RANGE2_LOWER Octet 6
RANGE2_HIGHER Octet 8
RANGE3_LOWER Octet 9
RANGE3_HIGHER RANGE4_LOWER Octet 11
RANGE4_HIGHER Octet 13
Octet n
BA ARFCN List
8 7 6 5 4 3 2 1
Cell Channel Description IEI
0 0 0 0 CA CA CA CA
Format ID spare spare ARFCN ARFCN ARFCN ARFCN
124 123 122 121
CA CA CA CA CA CA CA CA
ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN
120 119 118 117 116 115 114 113
CA CA CA CA CA CA CA CA
ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN ARFCN
008 007 006 005 004 003 002 001
CA = 0 – not part of BA list

CA = 1 – included in BA list
BCCH Carrier Measurements
• BCCH carrier measurements are used for:

• Idle Mode:
• Cell Selection/Reselection
• Dedicated Mode
• Handovers
• Power Control
• Call Maintenance
BCCH Carrier Measurements – Idle Mode
85Bm RxLev
96dBm
RxLev / BSIC
BCCH Carrier Measurement – Idle Mode
• Neighbouring Cells:
• MS scans all carriers listed in the BA list and identifies the 6
strongest
• Signal level averaged over at least 5 measurements
• Result stored in RXLEV (n) parameter
• MS must attempt to decode the BCCH channel of 6 best
neighbours at least every 30 seconds
• Serving Cell
• Measurements taken during allocated paging block
• Measurements averaged over 5 consecutive paging blocks or
5 seconds (whichever greater)
BCCH Carrier Measurements – Dedicated Mode
RxLev Measurement
RxQual Reports
RxLev / BSIC
BCCH Carrier Measurement – Dedicated Mode
• MS measures neighbouring cells for:

• BCCH Receive Signal Level (Rx_Lev)
• MS measures serving cell for:
• BCCH Receive Signal Level (Rx_Lev_Full)
• BCCH Receive Signal Quality (Rx_Qual_Full)
• Idle TCH Receive Signal Level (Rx_Lev_Sub)
• Idle TCH Receive Signal Quality (Rx_Qual_Sub)
• MS sends a Measurement Report to the BSC on SACCH
every 4 TCH Multiframes
• BSC measures MS uplink Rx_Qual and Rx_Lev
Cell Selection Procedure
• MS powers-up
• MS starts measuring received power level of the BCCH carrier for all
cells in range
• MS calculates average power level received from each cell:

• Stored in RxLev(n) parameter
• MS calculates a C1 parameter for each measured carrier based on the

RxLev(n) values
• Mobile compares cells which give a positive value of C1 and ‘camps-on’

to the cell with the highest C1 value
Cell Reselection Causes
• C1 indicates that the path loss to the serving

cell has become too high
• Downlink signalling failure
• Serving cell has become barred
• Better cell (in terms C2 value)
• Random access attempts unsuccessful after
"Max retrans" repetitions
Cell Re-selection – GSM Phase 1 Mobiles
For GSM Phase 1 mobiles, cell reselection is achieved by comparing current cell C1
with neighbouring C1 cell measurements:
• Between cells within a Location Area:
C1 (new) > C1 (old)

(for more than 5 seconds)
• Between cells on a Location Area boundary:
C1 (new) > C1 (old) + OFFSET

(for more than 5 seconds)
Cell Re-selection – GSM Phase 2 Mobiles
• GSM Phase 2 introduced a separate cell re-selection

parameter, C2
• Intended to:
• Prevent multiple handovers for fast-moving mobiles
• Ensure MS camps on to cell with greatest chance of successful
communications
• The C2 calculated is:
C2 = C1 + OFFSET – (TEMPORARY_OFFSET x H(PENALTY_TIME –T)

C2 Cell Selection Parameters
C2 = C1 + OFFSET – (TEMPORARY_OFFSET x H(PENALTY_TIME –T)
• Offset:
• optional parameter to encourage or discourage cell selection
• H(Penalty_Time-T):
• when a cell is added to list of strongest cells, a negative ‘Temporary Offset’
offset is applied for a ‘Penalty time’:
• If timer expires, offset is removed making cell more attractive
• If cell is removed from list, timer is reset
• Used to prevent fast-moving MSs from selecting the cell
• Temporary Offset:
• Value of the negative offset applied
Handover Processes
There are four different processes for handing over
within a GSM system, each requiring differing
procedures:
• Channels (time slots) in the same cell BSC
Internal
• Cells within the same BSS (same BSC)
BSC
• Cells in different BSSs (different BSCs) but under

External the control of the same MSC VLR
•
MSC
Cells under the control of different MSCs
GSM handovers are all ‘hard’ – i.e. mobile only BSC

communicates with one cell at a time during the
process VLR
MSC
Handover Types
• Handovers can be initiated by either MS or MSC
• Handover decisions are based on the following parameters
(in priority order):
• UL/DL Signal Quality
• UL/DL Signal Level
• Interference
• Power Budget
• Distance of MS from BTS
• Can be up- or down-link specific
• Each parameter has operator-defined threshold parameters
• Handover decisions can be based on one or a combination
of these parameters
Handover Initiation
Measure serving and neighbouring

cells’ performance
Compare measurements with

enabled thresholds
Initiate handover procedure
Determine best candidate

neighbouring cell
Select cell and perform handover

Handover Procedure - Command Message
Structure of the message sent to MS by original BSS:
MS BSS
Message Type
Cell Description Handover Command
Handover Reference
Power Command
Includes Frequency Hopping

Channel Description information if required
Frequency List Non - Frequency Hopping
or
Mobile Allocation Frequency Hopping
Handover Margin
Handover to BTS 1 Handover to BTS 2
Mobile remains with

BTS 1 BTS 1 or BTS 2 BTS 2
Nominal cell boundary
Hysteresis due to handover margin

Handover Procedure – Signalling Example

Signalling for a basic Inter-BSC handover involving only one
MSC (Intra - MSC):
MS BSS 1 BSS 2 MSC
Measurement report
Measurement report
Handover Required
Measurement report
Handover Request
Measurement report
Acknowledgement
Handover Command
Handover Command
Handover Access
Handover Detection
Physical Information
Handover Complete
Handover Complete
Clear Command
Measurement report
Clear Complete
Measurement report
GSM Power Control Functions

• Prevent unnecessary power emissions to:
• Increase life of battery-powered devices
• Reduce network interference
• Equalise power levels received at BTSs
• Adjustments to cell coverage
• Methods Include:
• Adaptive Power Control
• Discontinuous Transmission (DTX)
• Discontinuous Reception (RTX)
Mobile Station Power Classes

Full Rate
MS Class
Power mW dBm
GSM class 2 960 39 (8W)
GSM class 3 600 37 (5W)
GSM class 4 240 33 (2W)
GSM class 5 96 29 (0.8W)
DCS class 1 120 30 (1W)
DCS class 2 30 24 (.25W)
DCS class 3 480 36 (4W)
GSM class 1 – deleted under GSM Phase 2 Specification

Power (mW) = Nominal maximum mean power output (milliwatts)
Power (dBm) = Maximum power output in dBm (+watts)
Source: ETSI GSM 02.06 (Version 4.5.2)
BTS Power Classes
GSM 900 DCS 1800

TRX Maximum TRX Maximum
Power O/P Power Power O/P Power
Class (W) Class (W)
1 320 1 20
2 160 2 10
3 80 3 5
4 40 4 2.5
5 20
6 10
7 5
8 22.57

Receiver Sensitivity Levels
System Class Sensitivity

GSM 900 Small MS -102dBm
GSM 900 Other MS -104dBm
GSM 900 Macro BTS -104dBm
GSM 900 M1 Micro BTS -97dBm
DCS 1800 MS Class 1 or 2 -100 dBm
DCS 1800 MS Class 3 -102 dBm

Adaptive Power Control Process
• Compulsory in MS, optional in BTS
• 32 power levels separated by 2dBm
• Power changes are commanded using:
• Reduction: POW_RED_STEP_SIZE (2, 4 dB steps)
• Increase: POW_INC_STEP_SIZE (2, 4, 6 dB steps)
• Commands issued on SACCH
• One 2dB step change every 60mS

Discontinuous Transmission (DTX)

• In a conversation, a person generally only speaks for
about 30% to 40% of the time
• DTX makes use of this by reducing transmission when no
voice signal is detected
• Uses a Voice Activity Detection (VAD) unit
• Advantages:
• Reduces interference
• Prolongs battery life of mobile
Discontinuous Reception (DRX)
• Allows MS to power down parts of its circuitry in idle

mode
• MSs within a Location Area divided into paging groups
• MS only listens paging requests within its own group

• Increases battery life of MS
Frequency Hopping Concept

• Non-Frequency Hopping Carrier:
TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7
F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1 F1
TDMA Frame 1 TDMA Frame 2 TDMA Frame 3
• Frequency Hopping Carrier:

TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7
F1 F1 F1 F1 F1 F1 F1 F1 F2 F2 F2 F2 F2 F2 F2 F2 F3 F3 F3 F3 F3 F3 F3 F3
TDMA Frame 1 TDMA Frame 2 TDMA Frame 3

Frequency Hopping Sequence

• One TDMA frame is 4.6 ms long
• Rate of hopping = 1/ (4.6 x 10-3) = 217 hops / second
• The frequency changes follow either a sequential or pseudo-random pattern

• GSM defines 1 sequential pattern and 63 pseudo-random patterns
• Each pattern is defined by a Hop Sequence Number (HSN)
Sequential: F1 F2 F3 F4 F1 F2 F3 F4 F1 F2 F3 F4
Hop Sequence
TDMA Frame
Pseudo-Random: F1 F4 F3 F2 F1 F4 F3 F2 F1 F4 F3 F2
Hop Sequence
Channel Timeslot Hopping Sequence
TDMA Frame 1 TDMA Frame 2 TDMA Frame 3 TDMA Frame 4
F1 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
F2 0 1 2 3 4 5 6 7
F3 0 1 2 3 4 5 6 7
Time
Reasons for Frequency Hopping (1)

• GSM radio signals are affected by multi-path interference, causing fading
• Changing frequency (wavelength) moves the position of the fade
• Frequency hopping cycles through many fade positions
• This reduces the effect of the fades when the mobile is moving slowly
Fade position depends on

y path difference in terms of
wavelengths
x Rx
d Fade when : (x +y) - d = n l/2 where n = odd number

Tx
Reasons for Frequency Hopping (2)
• Cells are subject to interference A1

from other cells using the same
carriers at the re-use distance A3 A2 B1
• If the cells hop through a set of
frequencies in different C1 B3 B2
sequences, the effect of this
interference is reduced C3 C2 A1
• C/I ratio is increased A3 A2

Frequency Hopping at the BTS
• BCCH carrier will not hop - mobiles must be able to access

this for neighbour cell power level measurements
• Only TRXs used for traffic channels will hop through set
sequences
• The set of carrier frequencies assigned to the sequence

(Mobile allocation – MA) will normally be from current cell
allocation
• Hopping sequence for each TRX must be different or have a

different Mobile Allocation Index Offset (MAIO)
Example of 4-Frequency Hopping
Transmitter Hop 1 Hop 2 Hop3 Hop 4
TRX 1 F1 F1 F1 F1
TRX 2 F2 F3 F4 F2
TRX 3 F3 F4 F2 F3
TRX 4 F4 F2 F3 F4
• BCCH carrier remains on single frequency
• TCH carriers must start at different points in sequence (MAIO) to avoid

co-channel (C/I) interference
• Above example uses same HSN for each TRX but different MAIOs
Baseband Frequency Hopping
Fixed TRX
Antenna
Baseband
Fixed TRX Combiner
Data Signal
Fixed TRX
Switch controller
• The baseband signal is fed to one of several TRXs in turn by a switch

• The TRX outputs must be combined to be fed to the antenna
• The combiner must be able to handle a wide bandwidth of signals
• This can be achieved using either:
• hybrid combiners - several stages causing large loss
• cavity filters - one associated with each TRX - maximum loss ~ 5 dB
Synthesiser Frequency Hopping
Baseband Synthesiser
TRX
Antenna
Data Signal
Tuning controller
Frequency Hopping at the MS
• All mobiles must be capable of SFH in case it enters a

cell in which it is implemented
• SFH is implemented to allow time to continue to take
measurements from adjacent cells
• On connection/handover, the MS needs to know:
• Frequencies used for hopping (Mobile Allocation)
• Hop Sequence Number (HSN)
• Start frequency (Mobile Allocation Index Offset - MAIO)
• The MS uplink HSN is the same as the TRX downlink
HSN but offset by 45MHz
Layer 3 Messaging
• Management messages passed at network

layer 3, using layer 2 and 1 services for
transport.
• Layer 3 messages divided in 3 groups:

• Radio Resource (RR) Management
• Mobility Management (MM)
• Connection Management (CM)
GSM Layer 3 Messaging

Um Abis A
BTS
GSM Core
BSC MSC PSTN
Network
BTS
CM CM
MM MM
Layer 3
DTAP DTAP
RR
BSSMAP BSSMAP
RR
RR BTSM BTSM SCCP SCCP
LAPDm LAPDm LAPD LAPD Layer 2

MTP’ MTP’
TDMA TDMA G.703 G.703 Layer 1
MS BTS BSC MSC

Layer 3 Message Format

8 7 6 5 4 3 2 1
1 Transaction ID Protocol Discriminator

Message Header
2 0 Message Type Group
k Information Element ID (IEI)
k+1 IE Content
k+2 IE Content
Octets
Payload
Group
n Information Element ID (IEI)

n+1 IE Content
n+2 IE Content
• Protocol Discriminator: 0 0 1 1 - Call Control

0 1 0 1 - Mobility Management
0 1 1 0 - Radio Resource Management
1 0 0 1 - SMS messages
• A Length Indicator (LI) may be used with variable length IEs
Section 2
Network Characteristics
and
Problem Types
Section 2 – Network Characteristics & Problems
Section 2
2.1 Introduction
2.2 BSS Coverage Issues
2.3 BSS Capacity Issues
2.4 Quality of Service (QoS) Issues

Main Planning Constraints
The general requirements when designing a mobile

network are to maximise:
• Coverage
• Capacity
• Quality of Service (QoS)
• Cost-effectiveness (ROI)
BSS and Non-BSS Issues

• Network problem types can be divided into two distinct areas:
• Those arising at the BSS (BTS - BSC)
• Those arising in the transmission and the NSS
• Although often resulting in the same effect, problems in each

area require a different approach:
• BSS issues typically relate to frequencies, radio resource
dimensioning and/or maintenance of BSS database parameters
• Non-BSS issues may relate to transmission links availability, call set-
ups, location updates and paging attempts
Causes of BSS Coverage Degradation
• Cell Balancing
• Antenna Configuration
• Equipment Performance
Cell Balancing
• Power budget calculations show the maximum distance of the mobile from
the base station at which uplink and downlink can be maintained
• In a balanced system, the boundary for uplink and downlink must be the
same
Downlink limit Uplink limit Downlink limit Uplink limit
Unbalanced system Balanced system
• An unbalanced cell could drop many calls in the fringe region

Conditions for Cell Balance
• The conditions for cell balance depend on the

asymmetries between the uplink and downlink power
budgets
• These asymmetries include:
• MS and BTS Pout (max) not the same
• MS receiver less sensitive than BTS
• Diversity reception at BTS but not at MS
• Downlink-only combiner loss
System Balance Equation
• Power budget equations:
Downlink: PinMS = PoBS - Lc - Ld - Lfb + Gab - Lp + Gam - Lfm
Uplink: PinBS = PoMS - Lfm + Gam - Lp + GdBS + Gab - Lfb - Ld
• When the mobile is at the extreme boundary of the cell:

PinMS = PrefMS
PinBS = PrefBS
These are the reference sensitivities of the MS and BTS
The output levels PoBS and PoMS are the maximum allowed values
• If the boundaries for uplink and downlink are the same, the path loss Lp will be the
same in each direction
• Subtracting the uplink equation from the downlink gives the system balance
equation:
PoBS = PoMS + Lc + GdBS + ( PrefMS - PrefBS )
Field Implications of System Balance
• When changing cell size to alter coverage, consider whether the
change will affect the system balance, for example:
• Increasing BTS Tx power (PoBS ) to increase coverage would upset the
balance
• Ways of altering coverage without affecting balance include:

• Decreasing BTS Tx power - the BSS can force the MS to use dynamic
output power control (adjusting PoMS to maintain balance)
• Altering the gain of the base station antenna - Gab is a symmetrical term in
the power budgets
• Antenna down tilting changes coverage area without affecting balance
Antenna Configuration
• The following antenna configuration issues can

affect coverage performance:
• Antenna Alignment
• Tilting Configuration
• Use of Antenna Diversity

Antenna Alignment
Capacity can be increased by cell splitting where antenna re-
alignment is necessary
New cell
Rotate original antennas through 30o

Add new sites as shown
Old site
New site rotated
Effect of Antenna Alignment on Interference
• Directional pattern of sectored

antennas reduces response to
interference A
• Increases C/I significantly
• Allows greater frequency re-use, i.e.
smaller cells B
• If cells A and B use the same carrier:

• B will cause co-channel interference in A
• A will cause very little co-channel
interference in B
• Interference is no longer mutual
Antenna Tilt
• Down tilt of antennas often used to:

• reduce interference
• adjust cell size
• direct coverage e.g. into a building
• Mechanical tilt: Omni-directional antenna with electrical down tilt
• set by operator
• distorts azimuth (H plane) radiation
pattern
• Electrical tilt:
• set by manufacturer
• reduces radiation H plane pattern
equally in all directions, without
distortion
Examples of Antenna Tilt

0 0
-5 -5
No Tilt -10 Mechanical -10

-15 -15
-20 Downtilt -20
-25 -25
-30 -30
-35 -35
0
Electrical Downtilt + 0
Electrical -5 -5
-10 Mechanical Uptilt -10
Downtilt -15 -15
-20 -20
-25 -25
-30 -30
-35
-35
Diversity Reception
• Diversity reception is a way to improve the quality and
strength of the signal arriving at the base station, by
receiving it in several independent ways
• Two forms of diversity reception often employed are:

• Space diversity
• Polarisation diversity
• Frequency hopping is sometimes referred to as frequency

diversity
Space Diversity
• Two receiving antennas are used at the base
station
• If they are far apart, the received signals will
be independent of each other
• If one has suffered fading, the other may
not
• A suitable distance is generally about 10
wavelengths
• GSM 900, 10l = 3 metres
• Better isolation between the two signals can

also be obtained by mounting the antennas
10 l
at different heights on the tower
Space Diversity Antenna Systems

• Plan views of two possible tri-sectored site antenna
systems
Transmit antenna is separate One antenna is used for transmit and
from the receivers receive, using a duplexer in the BTS to
direct the signal
Rx Tx/Rx
Tx Rx
Rx Rx
Tx Rx
Rx Tx/Rx
Tx Rx
Rx Tx/Rx
Rx
Polarisation Diversity
• As the radio signal undergoes multiple
reflections and scattering, the plane of
polarisation is rotated randomly
• This can be used to provide diversity

reception by designing antennas with
dipoles crossed to receive different
components of the polarisation
• The preferred method is to cross the

dipoles at 45o
Dipoles
• This gives good coverage of vertical crossed at 45o
polarisation and strong components of
rotated signals
Deciding whether to use Diversity
Downlink limit Uplink limit Downlink limit Uplink limit
Unbalanced system Balanced system
• Diversity is used when necessary to balance the system.

• It helps the uplink but not the downlink.
• Diversity allows the BTS to operate at higher power whilst
maintaining link balance.
• Hence it allows greater coverage to be achieved.
Equipment Performance
• Effects of changes in equipment parameters can

result in:
• Frequency drift - resulting in increased interference
problems
• Output power variance – leading to coverage and
interference problems
• Can be addressed through an effective, proactive

equipment maintenance program
Causes of Equipment Performance Degradation
• Electronic component degradation

• Environmental conditions:
• Wind, rain, temperature
• Water ingress into connectors

• Deterioration of feeders
Losses in Feeders and Connectors

• Long feeder cables from antenna to base station
equipment can cause considerable power loss
• Typical loss in co-axial cable: 3 -10 dB per 100m
• Loss increases with frequency:
• 1800 MHz can have 4 - 10 dB greater loss than 900 MHz
• Loss depends on quality of cable:
• Cheap cable may give 20 dB per 100m
• Very expensive cable can have:
• 1 dB / 100m for 900 MHz
• 3 dB / 100m for 1800 MHz
• Connectors between duplexers,

combiners, couplers and so on
should produce no more than
0.2 dB loss
Issues Affecting Capacity
Issues that may affect network capacity include:

• Change in subscriber densities
• Initial dimensioning based on inaccurate initial demographic data
• Creation of new high-density subscriber areas
• Change in subscriber usage patterns

• Average number of calls
• Average call duration
• Predicted usage of existing services
• Introduction of new services

• Predicted take-up of new services
• Usage patterns for new service
• Capacity requirements for new service (e.g. HSCSD vs GPRS)
Issues Affecting BSS QoS
• QoS issues are generally identified through:

• Indications from key network performance parameters
• Customer complaints
• Issues affecting network QoS include:

• Frequency Plan
• Call set-up failures
• Long call set-up times
• Calls being dropped
• Poor speech quality
Adverse Influences on Frequency Plan

• Site Design:
• Physical parameters such as inconsistent antenna heights will
adversely affect the frequency re-use efficiency
• Terrain and Topography:
• Hilly terrain can produce unexpected coverage effects such as
‘splashes’ into other cells which can upset handovers as well as
producing interference
• Good site design and antenna positioning is critical to minimise such
effects
• External Interference:
• May be caused by unauthorised use of spectrum, such as 900MHz
cordless telephones used outside the USA
• Another source of interference is from operators in neighbouring
countries


• Frequency Plan
Call Setup Failures
• Faulty mobile or network equipment
• Congestion (TCH and/or SDCCH)
• Inadequate network coverage

• Frequency-related problems (fading/interference) etc
• Incorrect cell parameter configuration



• Frequency Plan
Long Call Setup Times
• Inadequate SDCCH / TCH capacity available
• Locating recipient of Call
• Roaming
• Inadequate paging capacity at called MT end



• Frequency Plan
Dropped Call Failures
• Faulty mobile or network equipment
• Handover failure
• Inadequate network coverage

• Frequency-related problems (fading/interference) etc
• Incorrect cell parameter configuration
• Emergency call pre-emption (Phase 2+ option)

Section 3
Introduction to
Performance Management
Section 3
3.1 Introduction
3.2 Purpose of Performance Management
3.3 The Performance Management Cycle
3.4 Initial Network Design and Implementation
3.5 Network Monitoring Phase
3.6 Measuring Network Performance
3.7 Data Analysis Phase

Section 3 – Introduction to Performance Management
What is Performance Management?
‘Performance Management is a process

by which the network is continuously
monitored to ensure optimum
performance and to identify & rectify
problems areas.’
Purpose of Performance Management

• Performance Management:
• Ongoing process to monitor network performance
• Sustains network quality throughout its lifecycle by proactive maintenance
• Reduces risk of network degradation
• Improve network performance through optimisation techniques
• Supports business interests of the network operator
• Poor maintenance procedures lead to network performance degradation

resulting in:
• Reactive fault rectification
• More complaints from customers
• Customer churn
• Reduced profit margins
• Weakening of operator’s business position
Performance Management Cycle
Initial Network
Design and
Optimisation Implementation
Monitor Network
Implement
Changes
Analyse Data
Identify
Problems
Yes QoS No
Target
s Met?
Initial Network Design and Implementation
• Part of network design and implementation process

involves:
• Defining quality of service (QoS) levels.
• Defining KPIs for each reporting level
• Defining custom network performance reports
• Network performance KPIs will be measured against

the QoS targets.
• QoS targets may require adjustment in light of
updated demographic/geographical planning data
and/or introduction of new services.
Network Monitoring Phase

• Monitoring current network performance at various
levels:
• Network, Regional MSC, BSS, NSS transmission levels
• Performance data sources include:

• Drive Tests
• Statistical Measurements:
• MSC, Databases (HLR/VLR), BSS Data (BTS/BSC)
• Customer Complaints
• Field Engineer Reports
• Some sources generate information in the form of

performance metrics
Performance Metrics
• Technical functionality:
• Failed establishment, dropped calls, failed handovers etc
• Traffic volume:
• Number of Subscribers, offered traffic, erlangs per cell or
per subscriber, switch processor load etc
• System availability:
• Channel % availability, switch outage times, channel
outage times etc
• System efficiency:
• % utilisation of resources, are GoS targets being met? %
blocking of cells and core network channels etc
Performance Measurement Methods

Generally, three methods:
•Drive Testing
•OMC Statistical Testing
•Protocol Analyser Testing

Comparison of Measurement Methods

• OMC
• Central position in network – network-wide overview
• Capability exists within standard network software
• Continuous Monitoring capability
• Speedier response to network problems
• Drive Testing:
• Can only provide data for limited geographical region
• Can only provide a ‘snap-shot’ of network characteristics
• More accurate local picture
• Can identify specific faults
• Manpower and equipment resource intensive
• Expensive
• Protocol Analyser Testing
• Can test uplink and down link more effectively than test Mobile.
• Can test transmission links
• Expensive – hence cost-limited in number
Analysis of Data
Data can be analysed under several headings:
• Call success
• evaluating the outcome of call attempt in terms of set-up time, clear
down success, assignment success etc
• Statistical distributions
• RxLev, RxQual
• Handover analysis
• showing success rate of attempted handovers
• Neighbours
• comparing neighbour cells found by signal level measurements with
the neighbour list in the site database
• Coverage Analysis
• Analysing the coverage threshold levels using serving cell/neighbour
cell comparison to identify problem areas
• Quality
• gives a comparison of signal quality from serving and neighbouring
cells
Drive Test Data Analysis Screens
• Examples of analysis
screens in NEPTUNE
Call Success Handover

OMC Data Analysis Screens

• Example of analysis screen in OPTIMA:
Section 4
Performance Measurement
Metrics
Section 4 – Performance Metrics
Section 4
4.1 Introduction
4.2 Key Performance Indicators
4.3 BSS KPI Definitions

Key Performance Indicators
• KPIs are established by the operator as a benchmark
against which to judge the performance of a network
• Which KPIs the operator decides to use will depend on:

• the nature of the network
• operator assessment of priorities for their business
• Counters implemented by vendors
• KPIs are generally based on three sets of data:

• Network statistics from the OMC
• Drive test statistics using test mobile equipment
• Protocol Analyser statistics
General Key Performance Indicators
KPI’S vary from operator to operator but the following

are generally common to all KPI statistics:
• TCH Usage
• SDCCH Usage
• Handover Statistics
• Connection Establishment
Speech Quality Measurements
• Speech QoS represented by:

• Received Signal Level – RxLev
• Received Signal Quality – RxQual
• Factors affecting these values include:

• BTS power output
• MS power output
• Current BTS loading
KPI Definitions
Call Setup Success Rate (CSSR)
• Measures proportion of call setup attempts resulting in a

successful TCH assignment, excluding those setup
attempts for reasons that do not require a TCH (eg.
Location Update, IMSI Detach etc.)
CSSR = Successfully Completed Call Setups

Total Valid Call Setup Attempts
Dropped Call Rate (DCR)
• Measures the proportion of successfully established calls

that are terminated abnormally
TCH _ RF _ Losses  HO _ Fail _ Drop

DCR 
Successfully _ Completed _ Call _ Setups( successful _ incoming _ handovers)
Call Success Rate (CSR)
• Measures the proportion of call setup attempts that are

successfully established and terminated normally (ie. not
dropped)
CSR  CSSR(1  DCR )

Handover Success Rate (HOSR)
• Measures the proportion of total handover attempts

resulting in a successfully completed handover, including
inter-cell, Intra-BSS and Inter-BSS handovers.
Total _ Successful _ handovers

HOSR 
Total _ Handover _ Attempts
Handover Failure Rate (HOFR)
• Measures the proportion of handover attempts that result

in handover failure with dropped call (ie. excluding
handover attempts that fail and recover to the source cell)
Total _ HO _ Fail _ Drop

HOFR 
Total _ Handover _ Attempts
Traffic Channel Blocking (TCH_Cong)
• Measures the proportion of attempts to allocate a TCH that

are blocked due to lack of TCH resources. This should
exclude blocked TCH assignment attempts that
subsequently succeed due to queuing procedures.
Blocked _ TCH _ Assignments

TCH _ Cong 
Total _ TCH _ Assignment _ Attempts
SDCCH Channel Blocking (SD_Cong)
• Measures the proportion of SDCCH Assignment Attempts

that are blocked due to lack of SDCCH resources.
Blocked _ SDCCH _ Assignments

SD _ Cong 
Total _ SDCCH _ Assignment _ Attempts
Dropped Call – RF Failure (TCH_RF_Loss)
• Measures the proportion of successfully established calls

that drop due to TCH RF loss (as distinct from
HO_Fail_Drop)
Total _ TCH _ RF _ Losses

TCH _ RF _ Loss 
Successfully _ Completed _ Call _ Setups
TCH Assignment Failure Rate (TCH_Ass_Flr)
• Measures the proportion of TCH assignment attempts that

fail after TCH allocation and before completion of the call
setup procedure.
TCH _ Assignment _ Failures

TCH _ Ass _ Flr 
Total _ TCH _ Assignment _ Attempts
SDCCH Access Success Rate (SD_Acc_Suc)
• Measures the proportion of Immediate Assignment

Attempts that result in successful assignment to SDCCH,
for all call setup reasons including location update, IMSI
detach etc.
Successful _ Im mediate _ Assignments

SD _ Acc _ Suc 
Total _ Im mediate _ Assignment _ Attempts
Non-BSS Call Setup Failure Rate (Non-BSS_CSFR)
• Measures the proportion of Call Setup Attempts that fail

due to Non-BSS reasons (eg. MSC rejections, subscriber
error, etc.)
Total _ Call _ Setup _ Failures  BSS _ Call _ Setup _ Failures

Non  BSS _ CSFR 
Valid _ Call _ Setup _ Attempts
Utilisation Factor
• The ratio of actual traffic carried to the theoretical traffic

that can be carried for a 2% Grade of Service, based on
the number of available TCH’s and Erlang-B formula.
Erlangs _ Carried
Utilizatio n 
Theoretical _ Erlangs _ 2%GOS
Section 5
Measuring Network Performance

Drive Testing
Section 5 – Drive Testing
Section 5
5.1 Introduction
5.2 Benefits and Limitations of Drive Testing
5.3 Drive Test Equipment
5.4 Test Mobile Data

Drive Test Benefits and Limitations
• Benefits:
• Replicates subscriber conditions (QoS perspective)
• Able to provide comparative performance between different operators
• Focus on specific parameter set or geographical region
• Limitations:
• Difficult to replicate subscriber usage patterns
• Area access restriction may limit realistic testing
• Difficult to achieve network-wide snapshot
(limited time and/or geographical area)
• Primarily down-link analysis only
• Resource intensive and expensive
Ascom QVoice
Test Mobile Box
Drive Testing Equipment

• System consists of:
• Test mobiles - either one or two for benchmarking against another
network
• GPS and differential receivers to provide location information
• Logging box to interface the measurement equipment to a laptop
computer
• Computer running logging and analysis software (Neptune)
The Logging Box

• A logging box can be used to interface the test mobile
outputs and GPS signals to the laptop computer system.
• The Aircom logging box is designed for use with Neptune:
12V Power Supply
Status Indicator
Connector for phone 1 (MS1) Handset 1 to

laptop connector
Charging switch (on/off)
for first phone (MS1)
Connector for handset 2 (MS2) Handset 2 to

laptop connector
Charging switch (on/off)
for second phone (MS2)
GPS Receiver GPS to Laptop

Connector connector
Cradle for handset 1 Cradle for handset 2
• Then use of logging boxes is diminishing as PC software

and hardware becomes more sophisticated and adaptable
Test Mobile Data

• Typical data collected is shown
here on the All Data screen in
Neptune:
• Serving cell information - BCCH,

BSIC, TCH, RxLev and RxQual,
Timing Advance, Tx Power
• Neighbour cell information - BCCH

and RxLev for six best neighbour
cells
Test Mobile Data
• For each measurement taken

the:
• BCCH
• BSIC
• TCH
are identified
Test Mobile Data

the
• RxLev Full
• RxLev Sub
are identified
Full = monitored TCH

Sub = DTX TCH
Test Mobile Data

the
• RxQual Full
• RxQual Sub
are identified
Full = monitored TCH

Sub = DTX TCH
Test Mobile Data

the Timing Advance Value (TA)
value is recorded
Test Mobile Data
• For each of the best six

neighbouring cells, the following
are recorded:
• BCCH n
• RxLev n
• BSIC n
Neighbour Lists
• Many handover problems, dropped calls and so on are due to incorrect
neighbour lists.
• Each time a change is made to a network, the neighbour relationships
should be rigorously updated.
• Failure to maintain neighbour lists leads to problems such as:
• Unwanted legacy neighbours
• Neighbour lists reaching the maximum allowable (typically 32)
• Neighbours defined on co-channel frequencies
• Missing neighbours
• Unintended one-way neighbour definitions
• Depending on the vendor, the OMC may support automated neighbour
detection via A-bis interface traces.
Neighbour Data
• RxLev data can be displayed
for the six best neighbour cells
and compared with the serving
cell’s RxLev
• This can be used to identify

neighbouring cells that do not
currently exist in the BA list of
the serving cell
Layer 3 Messages
• Neptune displays all layer 3
messages recorded by the test
mobile using the following fields:
• Time of message
• Direction
• UL = Uplink
• DL = Downlink
• Message Category
• RR = Radio Resource
• MM = Mobility Management
• CC = Call Control
• Message Type
• More detailed description
Graphical Display
• Data can be displayed
in a variety of graphical
forms
• The display here
shows:
• RxLev for the serving cell
• RXLev for six best
neighbours
• RxQual for the serving
cell
• Frame Erasure Rate
(FER) of the current
measurement
Test Mobile Data vs BSC Statistics

• A test mobile logging tool such as Aircom’s Neptune allows data to be
collected by drive testing a live network
• Advantages of test mobile data over BSC statistics:

• BSCs provide counters with the number of dropped calls but do not
indicate why or where
• BSCs do not collate information on poor downlink quality
• BSCs cannot give information on areas without network access from
either poor signal or quality
• BSCs do not store detailed information on calls
• Test mobiles are the only solution for diagnosing localised network
performance issues
Section 6
Measuring Network Performance

The OMC
Section 6 – The OMC
Section 6
6.1 Introduction
6.2 Role and Function of the OMC
6.3 Statistical Data
6.4 OMC Communications

Role of the OMC
• Perform operations and maintenance (O&M) activities

on GSM PLMN Elements
• Provide central network overview
• Often regionally-based under NMC hierarchy
• Manually ordering or remotely configuring
disconnections or restoration of network elements
• Manage the traffic loads at various points in the
network
OMC Functions
• Fault and Alarm Management
• Configuration and Operations Management
• Performance Management
Statistical Data
The network is capable of generating a huge amount of statistical

data. There is therefore a danger that information overload could
obscure some fundamental problems. Hence, it is important to
categorise the data in terms of importance and relevance. To
achieve this, performance counters are divided into groups by
network element and frequency of sampling.
Main QoS Parameters
• Dropped Call Rate (< 2%):

NumberOfDroppedCalls
DroppedCallRate   100
NumberOfActiveCalls
• Blocking (GoS) (2%-5%):

NumberOfCallAttempts  NumberOfCallSuccesse s
GOS   100
NumberOfCallAttempts
• Traffic Throughput:
Traffic for one subscriber = no of calls per hour per subscriber * MHT
3600
Periodic Counters
• Daily:
• Cell Level
• Rapid Fault Identification
• Includes Call setup rate, TCH/SDCCH congestion, TCH drop call
rate
• Weekly:
• Cell Level
• Performance Trends
• Includes traffic trends, cell retention, cell accessibility
• Monthly:
• BSC Level
• Performance Trends
• Includes processor loading, BSS call setup, handover success rates
Section 7
Introduction to Optimisation
Section 7 – Optimisation Overview
Section 7
7.1 Introduction
7.2 Requirements for Optimisation
7.3 Outline Optimisation Process
7.4 Network Audit Phase
7.5 Network Performance Review Phase
7.6 Activity Phase

Network Optimisation
No prescribed methodology
Often network architecture dependant
Often vendor equipment-dependant
Often engineer-experience dependant
An overview course
Insight into one approach to processes
Optimisation is an art as much as a science
A ‘tool-box’ approach
Dictionary Definition of Optimisation
‘Determining the best compromise

between potentially conflicting
requirements in order to plan and
implement an activity with maximum
efficiency.’
Section 7– Optimisation Overview
A Mobile Network Definition of Optimisation
‘The identification and rectification of

performance affecting problems within
the constraints of an existing network
infrastructure.’
Reason for Optimisation

• Goal:
• Ensure optimum network performance within QoS constraints
• Process:
• Identify network problems through analysis of KPIs
• Implement corrective actions and/or procedures to rectify problems
• Monitor results of modifications
• Purpose:
• Maintain/improve QoS
• Reduce churn rate
• Attract new customers
• Maximise revenue-generating service
• Maximize efficiency of network functional elements
Requirements for Optimisation

• Network issues that may generate a requirement for optimisation include:
• Perceived reduction in network quality
• Indications from network performance monitoring
• Subscriber’s experience of using the network
• Maximising the use of existing infrastructure
• operator wants to ensure best return on investment
• Introduction of new services
• maximise existing resources to accommodate new services such as GPRS
• Change in original design parameters
• Flawed original design information
• Original design information has change
• Regardless of the initiating event, the optimisation procedures remain
generally the same
Section 7– Optimisation Overview
Performance Management Cycle
Initial Network
Design and
Optimisation Implementation
Monitor Network
Implement
Changes
Analyse Data
Identify
Problems
Yes QoS No
Target
s Met?
Outline Optimisation Process
Initial network design

and roll-out phase
Monitoring/
Network Audit Phase
Optimisation Activity
Phase
Design Review and

growth phase
Network Audit Phase of Optimisation

Decide on KPIs,
Measurement Strategy and
Tools
Measure Performance,
Establish Benchmark
Performance Review to
Identify Major Performance
Affecting Issues
Decide on Strategy,
Establish Action Plan Feedback from
Optimisation
Activities
Start Optimisation Activity

Network Audit Process
• Performance Review Processes:

• Network performance statistics (OMC)
• A-Interface performance statistical analysis
• Call Trace Analysis
• Alarms and events
• GSM performance drive test
• GPRS Performance drive test
• Competitive drive test benchmarking

• For each category the following structure can be

applied:
• Objective. What are we trying to measure?
• Description. Why measure it and its relevance to

performance?
• Report Form. How should the results be presented?
• Interpretation. What conclusions can be drawn from

results?
• Recommendations. How can the problem be resolved?

• OMC Statistics Review:

• Call Success Rate
• Call Setup Rate
• TCH Blocking
• TCH Assignment Failure (RF)
• SDCCH RF Loss
• SDCCH Access performance
• Dropped Calls
• Handovers
• A Interface Analysis:
• Call Setup Failures
• Location Update Success Rate
• Handover Causes
• Call Trace Analysis:

• DL receive level and BTS power
• UL receive level and MS transmit power
• UL and DL RxQual distributions

• Events and Alarms:

• DL receive level and BTS power
• UL receive level and MS transmit power
• UL and DL RxQual distributions

• Drive Test Metrics - Route :

• Coverage route plot
• Quality route plot
• Frame erasure (FER) route plot
• MS transmit power route plot
• Speech Quality Index (SQI) route plot

• Drive Test Metrics - Events :

• Dropped call events
• Call setup failure events
• Handover failure events
• Successful handover events


• GPRS-Specific Drive Test Metrics:
• RLC throughput
• LLC throughput
• RLC BlkER
• RLC retransmission rate
• Coding scheme usage (CS1-4)
• Allocated timeslots
• Packet loss rate
• Latency/Jitter
• PDP Context activation failure
• PDP context loss (GPRS call drop)
Activity Phase of Optimisation

Network Audit Procedures Feedback
to
Audit
Procedures
Start Optimisation Activity
BSS Database
Parameters Review:
Ongoing Identify and Fix Hardware Problems Settings
Performance Consistency
Measurement Change Control
Process
Identify and Fix Neighbour Problems
Identify and Fix Frequency Plan Problems
Feedback to modify
strategy decisions
Review Process and Results
Design Review and Growth Plan

Section 8
Optimisation Activities
Section 8 – Optimisation Activities
Section 8
8.1 Introduction
8.2 BSS Database Parameter Review
8.3 Identifying and Fixing Hardware Problems
8.4 Identifying and Fixing Neighbour Problems
8.5 Identifying and Fixing Frequency Plan Problems

BSS Configuration Parameter Review

• Review current settings as related to performance
measurement results
• Recommend changes to improve/optimise performance of
specific features.
• Review includes:
• Handover parameters, timers, thresholds and margins
• Power control thresholds
• Averaging mechanisms for handover and power control
• Call setup parameters to maximise resource utilisation
• C1/C2 cell reselection parameters
• Any vendor-specific advance traffic management algorithms
BSS Database Parameter Consistency
• Within a network, different site types are defined (e.g. urban

micro, rural macro etc) by a standard template
• Each site type database will comprise a default parameter
set
• Each site may modify default set to suit local conditions
• Consistency of the default parameter sets should be
checked across BSS types
• Change control management processes should be reviewed
to ensure procedural consistency
BSS Configuration Parameter Sets

• Each BSS operates in accordance with its software
configuration
• All BSS are configured with certain standard default
parameters
• Each BSS will have a subset of BSS-specific parameters
• Parameters defined by:
• ETSI GSM Recommendations (GSM 12.04)
• Proprietary Parameter Implementation
• Potentially hundreds of configuration parameters
• Many parameters are inter-dependant
• Often vendor-specific abbreviations/acronyms used for same parameter
BSS Configuration Parameter Types

• Identifiers:
• CI, LAI, GCI, BSIC etc
• Channel Configuration:
• TCH channels, Signalling channel configuration (e.g. CCCH)
• Timers:
• Location Updates, C2 calculations etc
• Thresholds:
• RxLev, RxQual for handover decisions etc
• Offsets:
• Hysteresis for handovers etc
• Control Features:
• SFH, DTX, DRX etc
Adjusting BSS Configuration Parameters
• Effected from:
• PC connected directly to hardware
• Remotely from OMC/NMC
• Can be individually addressed or broadcast

• May require hardware reset to effect change
• Be aware of hierarchical changes (MSCBSC BTS)
• Only implement during low-traffic periods
• Use test BSS where available
• Avoid simultaneous multiple parameter changes
Identify Hardware Problems
• Problems identified through analysis of performance

measurement counters from:
• OMC Statistics
• Driver \Test data
• A-bis and A interface logs
• Identify ‘worst performing cells’ according to established KPI

benchmarks
Section 8– Optimisation Activities
Typical Hardware Problems

• Misaligned Antennas
• Increased interference, coverage degradation
• Wrongly/poorly connected feeders
• Reduced power output, reduced coverage, cell imbalance, distortion
• Poor transmission line performance (water ingress, corrosion,
physical damage etc
• High VSWR, high Insertion losses, reduced power output, reduced
coverage, cell imbalance, distortion
• Poorly calibrated TRXs
• Inconsistent TRX performance
• Incorrectly configured combiners/duplexers etc
• Reduced power output, reduced coverage, cell imbalance, distortion
Section 8– Optimisation Activities
Optimising Neighbour Lists
• Effects of poor maintenance:

• Unwanted legacy neighbours
• Oversized neighbour lists
• Co-channel neighbour definitions
• Missing neighbours
• Unintentional 1-way neighbour definitions
• Optimise by:
• Analyse neighbour performance form statistics.
• Utilise automated neighbour detection
• Identify inconsistent neighbour profiles
• Modify appropriate neighbour lists
Frequency Optimisation Activities
• Identify problem areas:

• Interference issues (internal and external)
• Coverage issues ( excessive overlaps, coverage gaps, high sites etc)
• Recommend appropriate frequency optimisation techniques
• BCCH and TCH Frequency Plans
• Frequency Diversity (space, polarity)
• Frequency Hopping
• DTX
• Antenna Down tilting
• Advanced Optimisation Techniques
BCCH and TCH Frequency Plans

• BCCH is typically planned with low frequency re-use (long re-use
distance) in order to ensure high quality.
• The BCCH plan should take into account: site design, terrain and
topography and subscriber distribution. A good BCCH plan can generally
be achieved with 14 - 15 carriers.
• The TCH plan requires the same considerations as BCCH, but may also
employ:
• Frequency hopping - synthesiser or baseband
• Multiple Re-use Patterns - giving tighter re-use on lower TCH layers
• Concentric multi-layer cell arrangements in which BCCH is only required on
one band
BCCH and TCH Frequency Plans
Guard Band Guard Band
1 ch 1 ch
BCCH TCH Hopping MICRO
14 ch 26 ch 8 ch
High Site Problems
• Initial roll-out tends to concentrate on higher sites

• High sites potentially cause excessive interference with
subsequent lower site roll-out
• A Typical process for replacement could include:
• From performance parameters, identify those cell interference-
contributing the most interference to the most cells
• Develop plan to lower antennas or decommission these sites. This
may require additional lower sites to cover any coverage gaps
• Prioritise decommissioning and integration of new sites to minimise
disruption to services
Antenna Down-Tilting
• An option for adjusting cell coverage

• e.g. down tilting may direct coverage deeper into a building
• Antenna tilt may be:
• mechanical – operator set – affects directional coverage
• electrical – manufacturer set – affects omni-directional coverage
• Omni antennas may have electrical tilt but not mechanical
• New technologies allow for remote electrical tilting
Examples of Antenna Tilting

0 0
-5 -5
No Tilt -10 Mechanical -10

-15 -15
-20 Downtilt -20
-25 -25
-30 -30
-35 -35
0
Electrical Downtilt + 0
Electrical -5 -5
-10 Mechanical Uptilt -10
Downtilt -15 -15
-20 -20
-25 -25
-30 -30
-35
-35
Antenna Selection
• Factors to consider with Antenna selection:
• Vertical and Horizontal Beamwidth

• Gain
• Front-to-Back Ratio
• Null Fill
• Downtilt (electrical/mechanical)
Antenna Configuration Options
• Combining:
• Economical with antenna elements
• Reduced coverage due to insertion loss
• Diversity Choice:
• Horizontal space diversity
• Vertical space diversity
• Polarisation diversity
Section 9
NSS Characteristics
Section 9 – NSS Characteristics
Section 9
9.1 Introduction
9.2 NSS Topology
9.3 Overview of SS7 Signalling in the NSS

PLMN Configuration Traffic Links/Link Sets

Signalling Links/Link Sets
VLR
BSC
AuC
MSC
BSC
HLR
EIR
BSC
MSC
BSC GMSC
VLR
‘Big BSC’ Configuration

Transit Switching Layer

SMS HLR/SMS/VMS/IN
HLR 5 VMS Layer
HLR 3 HLR 4
1200k 600k
HLR 2 150k IN
300k
HLR 1 GPRS
600k
TSC TSC TSC Layer

1a TSC TSC 4a
TSC TSC TSC TSC
1b 2b 2a 3b 4b 3a
MSC MSC
MSC MSC MSC
MSC MSC MSC Layer
MSC MSC MSC MSC
MSC MSC MSC MSC
MSC MSC MSC MSC
MSC MSC MSC
MSC
MSC
A Interface Configuration
2Mbps – E1 Link
0 Frame Alignment
1 TS1 TS2 TS3 TS4 fm Trunk 1
4 Trunk 1 TS0 Data fm Trunk 1
TRUNK 1 5 TS5 TS6 TS7 TS8 fm Trunk 1 TRUNK 1
8 Trunk 2 TS0 Data fm Trunk 2
: : : : :
: : : : :
TRUNK 2 TRUNK 2
13 TS13 TS14 TS15 n/u fm Trunk 1
BSC 14 TS13 TS14 TS15 n/u fm Trunk 2 MSC

16 C7 Signalling
TRUNK 3 TRUNK 3
: : : : :
: : : : :
TRAU Configurations
Um Abis A
BTS Site BSC Site MSC Site
CCU
TRAU A
CCU
16kbps 64kbps 64kbps

CCU
TRAU B
CCU

CCU
TRAU C
CCU
CCU Channel Coding Unit MSC Node BSC Node

PLMN Interfaces
VLR
B
BSC
A AuC
MSC D
H
A C
F G
BSC
E HLR
EIR E C
F
D
BSC A
MSC
E
A B
BSC GMSC
VLR
Network Signalling Requirements
• BSC to MS – GSM-specific Signalling

• BSC to NSS – C7 Signalling:
• User Data Connections:
• ISDN User Part (ISUP)
• Inter-NSS Node Signalling:
• C7 Mobile Application Part (MAP)
• A Interface Signalling:
• C7 BSS Application Part (BSSAP)
• Physical Layers
• Message Transfer Part (MTP)
PLMN Interfaces – C7 Signalling
VLR
B
BSC
A AuC
MSC D
H
A C
F G
BSC
E HLR
C7 Network EIR E C
F
D
BSC A
MSC
E
A B
BSC GMSC
VLR
Generic SS7 Network
SP
SP
STP SP
STP
SP
SP SP
SP
SP STP
SP
STP
C7 in the GSM NSS
SP VLR
BSC SP
AuC
STP SP
MSC
BSC SP
SP HLR
BSC SP
STP SP
EIR
MSC
SP
BSC SP
VLR
A Interface Signalling
Over the A Interface, GSM uses C7 signalling with protocol

extensions for GSM-specific requirements.
User Data DTAP

BSSAPP
BSSMAP
Layer 1-3 SCCP
MTP 1-3
A Interface Signalling Protocols

• BSS Mobile Application Part (BSSMAP)
• All messages exchanged between BSC and MSC processed by
the BSC
• Direct Transfer Application Part (DTAP)
• Message protocol for direct transfer of MM/CM Layer 3 messaging
between the MSC and the MS
• Transparent to the BSS
• Signalling Connection Control Part (SCCP)
• Logical signalling connection between MSC and MS to support
Layer 3 (MM/CM) message transfers
• Uses reduced set of C7 signalling instructions
• Message Transfer Part Layers 1-3 (MTP 1-3)
• Responsible for routing and transport of signalling messages
NSS Interface Signalling
D
F
E
C B
Layer 7 HLR MSC MSC VLR EIR
MAP
Layers 4-6 TCAP
Layer 3 SCCP
Layers 1-2 MTP
C7 Protocol Stack
NSS Interface Signalling Protocols

• Mobile Application Part (MAP)
• Mobile-specific extension of C7 standard.
• All messages exchanged between NSS Elements (MSC,HLR,VLR)
• Transaction Capabilities Application Part (TCAP)
• A connectionless transport service provided by the SCCP
• Provides core functionality to support roaming
• Allows its users to access other users via worldwide C7/SS7 network
• Signalling Connection Control Part (SCCP)
• Logical signalling connection between MSC and MS to support Layer 3
(MM/CM) message transfers
• Uses reduced set of C7 signalling instructions
• Routes messages via STPs
• Message Transfer Part (MTP)
• C7 protocols responsible for routing and transport of signalling messages
Section 10
NSS Parameter Evaluation

and
Optimisation
Section 10 – NSS Parameters and Optimisation
Section 10
?.1 Introduction
?.7 NSS Optimisation Process
?.2 SS7 Signal Routing Analysis
?.3 SS7 Link Load Evaluation
?.4 Traffic Loading Evaluation
?.5 Parameter Evaluation
?.6 Database Usage

NSS Optimisation Process
NSS Data Capture
Phase
Data Analysis Phase
Define Optimisation
Requirements
Define Implementation
Strategy
Phase
NSS Data Capture Phase
• Available Data:
• C7 signal routing
• C7 link loading
• User traffic routing
• Trunk group configuration and loading
• NSS configuration parameters
• Database loading
NSS Data Analysis Phase
• STEP 1 –
• Organise collected data into functional areas
• STEP 2 –
• Validate organised data
• STEP 3 –
• Evaluate validated data
NSS Optimisation Phase
• STEP 1 –
• Review areas of concern identified as part of
the Validation Phase
• STEP 2 –
• Identify optimisation techniques suitable for
improving poorly performing areas
NSS Implementation Phase
• Plan the implementation of optimisation

procedures
• Ensure sufficient time/manpower/equipment
resources are available of the right time
• De-conflict potentially conflicting procedures
• Identify monitoring procedures/periods required
to assess performance changes
NSS Optimisation Activity Phase
• Implement optimisation procedures:

• as identified during the Optimisation Phase
• in accordance with the plan designed during the

Implementation Phase
• monitored as agreed during the Implementation

Phase
Routing Evaluation
• Check for:
• Main and alternate routing
• Presence of routing circles
• Consistent implementation of routing strategy
• Effective use of transit switches (where used)

Possible Routing Problems
• Lack of uniformally implemented routing strategy:

• Number of primary/secondary/tertiary routes
• Inconsistent Routing parameters
• Excessive signal loading on routes
• Lack of alternate routing to key network elements

• Signalling errors due to routing circles
• Too many routes defined – uneven spreading of load
Monitoring Signalling Loads
• C7 Signalling Links:
• Signalling links are logically independent of voice traffic links
• Links are full-duplex
• Periodicity of Monitoring:
• Carried out at least monthly (preferably weekly)
• Carried out on significant network configuration change
• Recommended Capacity:
• ITU Q.706 specifies maximum of 200mE per link in each direction
• May be exceeded temporarily but is not recommended
• Calculation:
Signalling Load [ mErl] = (Number of MSU * 6 + Number of octets of payload)

( 8 * length of time in seconds)
Signal Load Evaluation – Link Load Sharing
• Excessive link loads can occur

TSC1 C7 Load C7 Load
Link
when: Destination No Rx [mErl] Tx [mErl]
• Original signal load dimensioning was

not correct 0 305 121
1 307 134
• Network characteristics have changed
2 305 127
• Ideal Situation:
3 303 84
4 302 76
• Links should not exceed 200mE HLR1
5 312 91
• Signalling traffic should be spread 6 104 94
evenly across all links in link set
7 104 109
8 103 89
9 104 102
Signal Load Evaluation – Link Load Symmetry
• Individual links should not exceed
200mE
• Tx/Rx loads should be as symmetrical
as possible Destination Link No Rx [mErl] Tx [mErl]
• Benefits:
MSC3 0 15 483
1 19 437
• Simplified routing
• Reduces risk of lost MSUs
• Reduced risk of signalling loops
• More efficient bearer usage
The table above serves as an example of unbalanced signal loading between a
TSC and an MSC in a live network. The Rx side of the two signalling links shown
is relatively unused whereas the Tx side is overloaded (exceeding the 200mE
threshold). Inefficient usage of C7 links is wasteful of C7 hardware resources.
This ideal symmetrical configuration, together with ensuring maximum

recommended thresholds are not exceeded, has the following benefits:
•simplified routing
•reduces the risk of lost MSU
•reduces the risk of signalling loops being created.
•More efficient bearer utilisation due to equal load Tx/Rx volume
spreading
MSU per sec
H
LR
1
H (
LR Ra
2 ha
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
(R ha
ah n )
M aHLR h 1
S an
CE )
HLR
M 2
M AM
SMSC1
C
M KJ
SMSC2
CR
Al M A
gh SMSC3
ad CS
ir B
H M MSC4
LR SC
3 MSC5
(
M BH)
S
CHLR3
M BE
S H
CMSC6
FA
M
Exchange
H R
LR S CMSC7
5 LC
(E T
mMSC8
LC a m
EM TTHLR4)
S
Total MSU per sec
AM C1
TTSC1
M SC
S 1
FA CTSC2 K
TE AN
HMSC9
M
S
C
MSC10
Recommended Nodal MSU Capacity <= 1500 MSUs/second
Tx
Rx
Signal Load Evaluation – Nodal MSU Capacity
Recommended Value per Node

Traffic Load Measurement Requirement
• Recommended twice-daily 1-hour measurement

periods
• Data gathered from MSCs at regular intervals to
assess capacity on each link
• Trunk capacities assessed:
• Between MSCs within the PLMN
• From PLMN to PSTN
• From PSTN to PLMN
Traffic Load Data Capture
• Data captured should include:
• Number of trunk available on each traffic link
• Level of traffic (erlangs) on each link during

the busy hour
• Answer Bit Ratio (ABR) measured on each

link during the busy hour
Traffic Load Analysis
• Analysis of the measured data should include:
• Calculation of number of trunk required to support

traffic on link
• Any difference between trunks required and trunks

available
• Utilisation Factor for each trunk

Recommended Traffic Parameters
• Maximum erlangs per trunk = 0.7e
• Maximum utilisation factor per trunk = 75%
• GOS = 1%
NSS Traffic – Trunk Utilisation Example
Addit.
NSS Call NSS Total Trunks Trunk Trunks

Trunk Gp Time Trunks Attempts ABR% Erlangs Reqd Utilisation Required
1 11-12 1455 128,394 46% 1177.2 1207 83% 154
2 11-12 180 6,783 51% 120.9 139 77% 5
3 11-12 210 12,324 32% 162.7 183 87% 34
4 11-12 1375 104,514 58% 1281.3 1311 95% 373
5 11-12 1527 98,192 58% 1250.2 1280 84% 180
GoS = 1% Max Erlangs/trunk = 0.7 Only trunk utilisation > 75% shown
Network Parameter Analysis
• Parameters control equipment operation and

performance
• Initial parameters determined at network planning
stage
• Parameter sets should remain common where
possible to facilitate efficient network management
• Parameters may require modification to suit local
conditions.
• Periodic operation and performance monitoring
necessary to ensure parameters remain valid
Parameter Analysis – VLR Parameters
SUBSCRIBER PROFILE
TMSI Usage
Authentication Usage
Implicit Detach Usage
Emergency Call
Time Implicit IMSI Detach Time Limit
Limits Call Waiting Time
VLR Cleaning Start Time (for Siemens If capacity above 80%)
Minimum Amount of Triplets to be kept on record (MINT)
SUBSCRIBER SERVICES SUPPORT
Calling Line Identification Presentation (CLIP)
Call Forwarding
Call Waiting (CW)
Call Hold (HOLD)
Advice of charge charging (AOCC)
Call Barring
TELESERVICE SUPPORT
Short Message Mobile Terminated (T21)/Originated (T22)
Fax gr.3 and Alternate Speech (T61)/ Automatic fax group 3 (T62)
Parameter Analysis – TMSI Parameters
• Advantages of TMSI usage:

• Increases paging channel (PCH) capacity
• Reduces VLR/HLR signalling requirement
• Provides greater security than using IMSI only
• TMSI should be unique within a LA
• New TMSI issued on change of LA
• TMSI allocation/update occurrences set by Operator
• Too many TMSI update occurrences increases signalling
capacity requirements
• Use of TMSI should be uniform across network
PARAMETER REC
Location Update New Visitor Y

IMSI Attach Y
Location Update Y
TMSI Periodic Location Update N
ALLOCATION Mobile Originating Call N
COUNTERS Mobile Originating SMS N
Mobile Terminating Call N
Mobile Terminating SMS N
Mobile Terminating USSD N
Mobile SS Operation N
Parameter Analysis – Authentication Parameters
• Advantages of Authentication usage:

• Increases security of network
• Prevents unauthorised network access
• Implementation of Authentication is an Operator decision
• Can be implemented internally to network and/or for
roaming subscribers
• Excessive Authentication increases signalling capacity
requirements
• Use of Authentication should be uniform across network
PARAMETER REC
IMSI Attach Y
AUTHENTICATION Location Update N
COUNT Periodic Location Update N
Mobile Originating Call N
Mobile Originating SMS N
Parameter Analysis – IMEI Check Parameters
PARAMETER REC
IMSI Attach Y
Location Update Y(10)
Periodic Location Update N
IMEI CHECKING ON…. Mobile Terminating Call N
Black List Effect BLOCK
Grey List Effect TRACE
Unknown IMEI Effect BLOCK
NSS - Implicit Deregistration
• Reduces signalling load towards the BSS by

reducing unnecessary paging
• Deregisters MS after period of inactivity
• Compares timestamp value in subscriber profile
with inactivity timer.
• Timer is reset when MSC-MS interaction takes
place
• If timer value > timestamp MS is deregistered
and paging ceases
• Generally at least 2x periodic update timer value
Database Usage – VLR Cleaning
• Procedure to reduce loading within VLRs by

removing redundant subscriber records
• Each subscriber record has an associated timer.
• If timer expired before next contact with subscriber,
the subscriber record in marked for deletion
• VLR cleaning takes place at periodic intervals –
normally during periods of reduced activity.
• All subscriber records marked for deletion a removed
during VLR cleaning process
NSS – MSRN Lifetime
• MSRN allocated by VLR for incoming call routing

• Temporary Assignment
• After call is connected, MRSN released back to
VLR for future use.
• MSRN ‘Lifetime’ refers to the time between MSRN
allocation and being releasing back to the network
• Typical life times are:
• Local Subscribers: 20 secs
• Visiting Roaming subscribers: 90 secs
NSS – Call Waiting Time
• Time between call offered and call accepted

• Network resources assigned during call waiting
time
• If call not accepted within call waiting period,
network resources are released
• Typical call waiting time: 25 secs
Database Usage - VLR
• VLR capacity should be evenly spread where possible
• 75% capacity limit is recommended to allow for peaks
and future expansion
• Options for reducing VLR loading include:
• Redistribution of BSCs amongst MSCs
• Redesigning LA boundaries
• Too large - excessive paging requirement
• Too small - excessive location update requirement
• Modify parameter settings to reduce signalling occurrences
• VLR Cleaning
• Implicit Deregistration
NSS Databases – VLR Utilisation
VLR
• VLR utilisation refers to subscriber record Sr. No. MSC Utilisation
capacity and usage within a VLR
1 MSC-1 65%
• Utilisation problems include: 2 MSC-2 82%
• Over-capacity in certain VLRs 3 MSC-3 26%
• Uneven spread of records across the 4 MSC-4 100%
network
5 MSC-5 86%
• Over-capacity can result in: 6 MSC-6 66%
• Call function failures (e.g. setup/handover, 7 MSC-7 71%
authentication etc)
8 MSC-8 100%
• Lack of growth capability
9 MSC-9 62%
10 MSC-10 7%
NSS - VLR Cleaning
• Reduces capacity loading within a VLR (Nokia proprietary)
• Deletes subscriber records that are inactive for a specified
period of time
• Compares timestamp value in subscriber profile with
inactivity timer.
• Timer is reset when sub scriber record is accessed
• If timer value > timestamp, subscriber record is marked for
deletion
• ‘Cleaning’ takes place during non-busy periods
• Common settings are typically 24hrs to 3 days
NSS Databases – HLR Utilisation
• HLR utilisation refers to
subscriber record capacity and
usage within a HLR
%AGE USAGE OF HLRs
• Utilisation problems include:
100
• Over-capacity in certain HLRs 90
• Uneven spread of records across 80
70
the network
60
• Over-capacity can result in: 50

40
%AGE USAGE
• Call function failures (e.g. 30

20
setup/handover, authentication
10
etc) 0
HLR1 HLR2 HLR3 HLR4 HLR5
• Signalling bottlenecks
Section 11
Optimising Networks
for
New Services
Section 11 – New Service Optimisation
Section 11
11.1 Introduction
11.2 Dimensioning Networks For New Services
11.3 GPRS Performance Monitoring

Dimensioning a Multi-Service System
• The Erlang B formula relies on the variance of the

demand equalling the mean (a Poisson distribution).
• If a particular service requires more than one “trunk”
per connection, the demand is effectively linearly
scaled and the variance no longer equals the mean.
• Methods to investigate:
• Equivalent Erlangs
• Post Erlang-B
• Campbell’s Theorem
Equivalent Erlangs
• Combine the two traffic sources together by Difference in

converting one to the bandwidth of the other capacity
• The trunking efficiency will VARY with the required for

same GoS
bandwidth of equivalent Erlang that you choose!
• Not suitable for use due to this property
Low
Bandwidth
Equivalent
2 Erlangs 1 Erlang of
of Low High
Bandwidth Bandwidth High
Bandwidth
Equivalent
Equivalent Erlangs Example
• Consider 2 services sharing the same resource:

• Service 1: uses 1 trunk per connection. 12 Erlangs of traffic.
• Service 2, uses 3 trunks per connection. 6 Erlangs of traffic.
• We could regard the above as equivalent to 30 Erlangs of

service 1:
• 30 Erlangs require 39 trunks for a 2% Blocking Probability
• Alternatively, we could regard the above as equivalent to 10
Erlangs of service 2.
• 10 Erlangs require 17 trunks, (equivalent to 51 “service 1 trunks”) for a
2% blocking probability
• Prediction varies depending on what approach you choose.
Post Erlang-B Difference

in capacity
required for
same GoS
Illustration using 2 services of
same bandwidth
• Combine the two traffic
sources together after
calculating required 1 Erlang of
capacity Service A +
• The trunking efficiency
variation with magnitude
is not considered - 1 Erlang of
pessimistic about offered Service B
traffic supported to the
same GoS
• Not suitable for use due
to this property Accepted correct method
1 Erlang and 1
Erlang of of
Service B
Post Erlang-B
• Service 2: uses 3 trunks per connection. 6 Erlangs of traffic.
• We could calculate the requirement separately

• Service 1: 12 Erlangs require 19 trunks for a 2% Blocking Probability
• Service 2: 6 Erlangs require 12 trunks (equivalent to 36 “service 1
trunks”).
• Adding these together gives 55 trunks.
• This method is known to over-estimate the number of trunks
required as can be demonstrated by considering services
requiring an equal number of trunks.
Post Erlang-B
• Consider 2 services requiring equal resource:
• Service 2: 6 Erlangs require 12 trunks.
• The accepted method of treating the above would be to regard
it as a total of 18 Erlangs that would require 26 trunks.
• Post Erlang-B overestimates the requirement.
Section 11– New Service Optimisation
Campbell’s Theorem
• Campbell’s theorem creates a composite distribution where:
Capacity 
Ci  ai  OfferedTraffic 
 Ci = available capacity
 = mean
c c u = variance
i = arrival rate
  i i bi
 a 2
ai = amplitude of service
c  i
Service Offered Traffic  γi bi
 a b
bi = mean holding time
 i i i c = capacity factor
i
• The amplitude (ai) used in the capacity is the amplitude of the target service
• Once the equivalent offered traffic and capacity are derived, GoS can be
derived with Erlang-B
• Required capacity can be calculated if offered traffic and GoS target is
known
Campbell’s Theorem Example

• Consider the same 2 services sharing the same
resource:
• In this case the mean is:

   ibi ai   Erlangs  ai  (112)  (3 6)  30
• The variance is:

   ibi ai2   Erlangs  ai2  (12 12 )  (6  32 )  66
Note:  i bi = traffic in Erlangs

• Capacity Factor c (average trunks per connection):

 66
c   2.2
 30
• Equivalent offered traffic:
α 30
Equivalent Offered Traffic    13.63
c 2.2
• Trunk capacity for equivalent traffic at 2% GoS = 21


• Actual trunk requirement is trunk capacity x capacity
factor:
= 21 x 2.2 = 46.2 or 46 trunks
• Required Capacity is modified depending upon target

service for GoS (in service 1 Erlangs):
• Target is Service 1 C1=(2.2 x 21) + 1 = 47
• Target is Service 2, C2=(2.2 x 21) + 3 = 49
• Different services will require a different capacity for the

same GoS depending on the amplitude
Traffic Analysis Methods Compared
• Optimistic if you use the smallest amplitude of trunk (39)
• Pessimistic if you use the largest amplitude of trunk (51)
• Post Erlang-B
• Pessimistic (55)
• Trunking efficiency improvement with magnitude ignored
• Campbell’s theorem
• Middle band (47 - 49)
• Different capacities required for different services - realistic
• Preferred solution for dimensioning, but not ideal...
Capacity Dimensioning with Campbell’s

Theorem
• Consider the following service definition and traffic

forecast.
Service Amplitude Forecast

Voice 1 250 E
HSCSD 2 63 E
• Based on a theoretical availability of 15 voice trunks per

cell and using voice as the ‘benchmark’ service and with
2% blocking, determine the number of cells required to
serve the above traffic levels and the traffic offered per cell
for each service

Theorem
• Assuming we have n cells, we can determine the loading

per cell.
250 63  2 376
mean   
n n n
250 63  2 2 502
variance   
n n n
variance 502
c   1.335
mean 376
mean 376 282
offered traffic   
c 1.335  n n

Theorem
• Ci  ai
Considering the equation: Capacity 
c
• Ci (available capacity) is predefined as 15. ai (amplitude)

depends on the service we use as our “benchmark” or priority
service. Choosing the voice service as the “benchmark”
service make ai equal to 1.
Ci 
15  1
 10.5
1.335
• 10.5 (rounded to 10) trunks will service 5.08 Erlangs.


Theorem
• Each cell requires 10 trunks to service 5.08 Erlangs.

• Therefore:
Number of cells required = Equivalent Traffic
Traffic per Cell
282
=  55.5
5.08
• Cell requirement is established at 56 cells.
• Each of the cells will service:
• 4.46 Erlangs of voice
• 1.13 Erlangs of HSCSD.
Assessing Cell Loading using Campbell’s

Theorem
• After placing sites on the coverage map and

spreading the traffic, the next stage is to assess the
cell loading (timeslots required).
• If mixed services are used, it is necessary to use

Campbell’s Theorem to assess the required number
of timeslots to satisfy the likely demand.
• Consider the case where a particular cell captures 7

Erlangs of voice and 2 Erlangs of HSCSD traffic that
requires 2 timeslots per connection.

Theorem
• Using Campbell’s Theorem:

mean  (7 1)  (2  2)  11
variance  (7  12 )  (2  2 2 )  15
15
Capacity Factor (c)   1.36
11
11
Equivalent offered traffic   8.09 Erlangs
1.36
From Erlang B, 14 trunks required.

Taking voice as benchmark : 14  1.36   1  20
• Hence 20 timeslots required.

Evaluating Traffic Requirements
Data may be packet switched, in which case it can be made to “fill the
gaps” in the demand for voice services.
Load
Peak traffic
Spare capacity for which can
be allocated to non real time
applications Average circuit
switched traffic
Real time non-controllable load
Time
• To evaluate the required cell capacity:
• First assume that the packet data can be scheduled to fill the spare real time
capacity.
• When all the spare real time capacity has been exhausted we must convert the
remaining capacity to an equivalent data capacity.
• One GSM timeslot can carry 13 kb/s of data.

Traffic Requirement Example
• A cell captures 2 Erlangs of voice traffic and is assigned a GSM carrier

with 7 timeslots. The grade of service is 2%.
Estimate the amount of data traffic that can be handled.
How much of this must be packet data and how much can be circuit
switched?
• Solution:
Timeslots available on average = 7 - 2 = 5
This represents a total data rate of 5 x 13 = 65 kb/s
2 Erlangs voice traffic requires 6 trunks (timeslots) for 2% blocking.
1 timeslot can be dedicated (circuit switched) for data, i.e. 13 kb/s

Remaining data must be packet switched = 65 - 13 = 52 kb/s.
Examples of GPRS-Specific Parameters
• Cell Reselection
• Service Exceptions
• Average Response Times
• Suspend/Resume Procedures
• Attach/Detach Procedures
• Routing Area (RA) Update Procedures
• PDP Context Activations
• Data Volumes and Rates
GPRS Service Exceptions
• Radio Link Signal Strength falls below

threshold (link failure)
• Link quality falls below threshold (link failure)
• Cell Reselection initiated

GPRS Transmission Connections

Um Gb Gn
GGS
BSS SGSN
N
MS
IMSI/NSAPI
PDP Context
Logical Link Control TLLI

Connection
DLCI
Data Link Connection
TID
Virtual Tunnel
CONNECTION
CONNECTION TYPE
IDENTIFIER
GPRS Cell Reselection
Cell Reselection – Subscribers in READY state

Number of Number of cell
Octets reselections
Average Number of Octets Affected

Average Number of Octets Deleted
Average Number of Frames Deleted
Number of Cell Reselections
| 0000100002 | 0000100003 | CI
Cell Reselection per source cell

Frames Lost Through Service Exceptions
Number of Frames
Deleted
Number of frames deleted per TLLI

GPRS Average Response Times
Time in millsecs
BVCI
Average response time after paging, per BVCI

Suspend / Resume Procedures
• Suspend Procedure:
• Enables a GPRS Class B MS to suspend GPRS mode in
order to make a circuit-switched connection
• Resume Procedure:
• Enables a GPRS Class B MS to resume GPRS mode having
made a circuit switched connection
• Monitoring:
• Procedure can be monitored and failures can be displayed
by TLLI, cell or cause
GPRS Attach/Detach Procedures
Number of Attach Procedures

Procedures
Attach Request Attach Accept Attach Complete Attach Reject
Number of Attach Procedure Messages

Number of Attach Rejects per Cell

Attach Rejects
Number of Attach Reject Messages per Cell


Number of Attach Rejects per TLLI
Attach Rejects
Number of Attach Reject Messages per Subscriber (TLLI)

GPRS RA Update Procedures
RA Updates
Number
RA Request RA Accept RA Reject RA Request without response
Number of RA Update Messages

Number of
RA Rejects per Cause
RA Rejects
Unknown Causes Network Failures
Number of RA Update Rejects by Cause

GPRS PDP Contexts

Number PDP PDP Context Activation by MS
Context
Activations
PDP Contexts Activated PDP Contexts Accepted PDP Contexts Rejected
Number of PDP Context Activation Messages

GPRS Data Volumes and Rates
DL LLC Rate
Bytes
Time (minutes)
Downlink Data Rate of a Cell

DL LLC Global Size

Volume (bytes)
Time (minutes)
Downlink Data Rate of a Mobile Station

End of Course
Any Questions?
Section 9 – NSS Characteristics and Evaluation
Signalling Load Evaluation

• C7 Signalling Links:
• Signalling links are logically independent of voice traffic links
• Links are full-duplex
• Periodicity of Monitoring:
• Carried out at least monthly (preferably weekly)
• Carried out on significant network configuration change
• Evaluation:
• ITU Q.706 specifies maximum of 200mE per link in each direction
• May be exceeded temporarily
• Calculation:
Signalling Load [ mErl] = (Number of MSU * 6 + Number of octets of payload)

( 8 * length of time in seconds)
Signal Load Evaluation – Link Load Sharing
• Excessive link loads can occur

Link
when: Destination No Rx [mErl] Tx [mErl]
• Original signal load dimension was not

correct HLR1 0 305 121
1 307 134
• Network characteristics have changed
2 305 127
3 303 84
4 302 76
• Links should not exceed 200mE
5 312 91
• Signalling traffic should be spread 6 104 94
evenly across all links in set
7 104 109
8 103 89
9 104 102
Signal Load Evaluation – Link Load Symmetry
• Individual links should not exceed
200mE
• Tx/Rx loads should be as symmetrical
as possible Destination Link No Rx [mErl] Tx [mErl]
• Benefits:
MSC3 0 15 483
1 19 437
• Simplified Routing
• Reduces risk of lost MSUs
• Reduced risk of signalling loops
• More efficient bearer usage
MSU per sec
H
LR
1
H (
LR Ra
2 ha
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
(R ha
ah n )
M aHLR h 1
S an
CE )
HLR
M 2
M AM
SMSC1
C
M KJ
SMSC2
CR
Al M A
gh SMSC3
ad CS
ir B
H M MSC4
LR SC
3 MSC5
(
M BH)
S
CHLR3
M BE
S H
CMSC6
FA
M
Exchange
H R
LR S CMSC7
5 LC
(E T
mMSC8
LC a m
EM TTHLR4)
S
Total MSU per sec
AM C1
TTSC1
M SC
S 1
FA CTSC2 K
TE AN
HMSC9
M
S
C
MSC10
Tx
Rx
Signal Load Evaluation – Nodal MSU Capacity

NSS Authentication-Related Parameters

• Advantages of Using TMSIs:
• Less air interface signalling loading (paging)
• Less NSS HLR-VLR loading (triplets)
• TMSI-Related Parameters:
• IMSI Attach (Y)
• Location Update (Y)
• Location Update - New Visitor (Y)
• Location Update – Periodic (N)
• MO Call Update (N)
• MO SMS Update (N)
• MT Call Update (N)
• MT SMS Update (N)
• MT USSD Update (N)
• Mobile SS Operation Update (N)
NSS Authentication-Related Parameters

• Using Authentication:
• Increases access security on network
• Increases signalling load on network
• Authentication-Related Parameters:
• IMSI Attach (Y)
• Location Update (N)
NSS IMEI-Related Parameters

• IMEI Usage:
• Increases access security on network
• Increases signalling load on network
• IMEI-Related Parameters:
• IMSI Attach (Y)
• Location Update (Y – every 10th update)
• Black (block) – Grey (trace)
NSS - Implicit Deregistration
• Reduces signalling load towards the BSS by

reducing unnecessary paging
• Deregisters MS after period of inactivity
• Timer is reset when MSC-MS interaction takes
place
• If timer value > timestamp MS is deregistered
and paging ceases
• Generally at least 2x periodic update timer value
NSS - VLR Cleaning

• Reduces capacity loading within a VLR
• Deletes subscriber records that are inactive for a
specified period of time
• Timer is reset when sub scriber record is
accessed
• If timer value > timestamp subscriber erecord in
marked for deletion
NSS – MSRN Lifetime

• MSRN allocated by VLR for incoming call routing
• Temporary Assignment
• After call is connected, MRSN released back to
VLR for future use.
• MRSN ‘Lifetime’ refers to the time between
MSRN allocation and releasing back to the
network
• Typical life times are:
• Local Subscribers: 20 secs
• Visiting Roaming subscriberts: 90 secs
NSS – Call Waiting Time
• Time between call offered and call accepted

• Network resources assigned during call waiting
time
• If call not accepted within call waiting period,
network resources are released
• Typical call waiting time: 25 secs
Section 9
Network Subsystem
Optimisation
Section 9 – Network Subsystem Optimisation
Section 9
9.1 Introduction
9.2
9.3
Section 10
Optimising Networks
for
New Services
Section 10
10.1 Introduction
10.2 Dimensioning Networks For New Services
10.3 GPRS Performance Monitoring

Dimensioning a Multi-Service System
• The Erlang B formula relies on the variance of the

demand equalling the mean (a Poisson distribution).
• If a particular service requires more than one “trunk”
per connection, the demand is effectively linearly
scaled and the variance no longer equals the mean.
• Methods to investigate:
• Post Erlang-B
• Campbell’s Theorem
Equivalent Erlangs
• Combine the two traffic sources together by Difference in

converting one to the bandwidth of the other capacity
• The trunking efficiency will VARY with the required for

same GoS
bandwidth of equivalent Erlang that you choose!
• Not suitable for use due to this property
Low
Bandwidth
Equivalent
2 Erlangs 1 Erlang of
of Low High
Bandwidth Bandwidth High
Bandwidth
Equivalent
Equivalent Erlangs Example

• We could regard the above as equivalent to 30 Erlangs of

service 1:
• 30 Erlangs require 39 trunks for a 2% Blocking Probability
• Alternatively, we could regard the above as equivalent to 10
Erlangs of service 2.
• 10 Erlangs require 17 trunks, (equivalent to 51 “service 1 trunks”) for a
2% blocking probability
• Prediction varies depending on what approach you choose.
Post Erlang-B Difference

in capacity
required for
same GoS
Illustration using 2 services of
same bandwidth
• Combine the two traffic
sources together after
calculating required 1 Erlang of
capacity Service A +
• The trunking efficiency
variation with magnitude
is not considered - 1 Erlang of
pessimistic about offered Service B
traffic supported to the
same GoS
• Not suitable for use due
to this property Accepted correct method
1 Erlang and 1
Erlang of of
Service B
Post Erlang-B
• Service 2: uses 3 trunks per connection. 6 Erlangs of traffic.

• Service 2: 6 Erlangs require 12 trunks (equivalent to 36 “service 1
trunks”).
• This method is known to over-estimate the number of trunks
required as can be demonstrated by considering services
requiring an equal number of trunks.
Post Erlang-B
• Consider 2 services requiring equal resource:
• Service 2: 6 Erlangs require 12 trunks.
• The accepted method of treating the above would be to regard
it as a total of 18 Erlangs that would require 26 trunks.
• Post Erlang-B overestimates the requirement.
Campbell’s Theorem
• Campbell’s theorem creates a composite distribution where:
Capacity 
Ci  ai  OfferedTraffic 
 Ci = available capacity
 = mean
c c u = variance
i = arrival rate
  i i bi
 a 2
ai = amplitude of service
c  i
Service Offered Traffic  γi bi
 a b
bi = mean holding time
 i i i c = capacity factor
i
• The amplitude (ai) used in the capacity is the amplitude of the target service
• Once the equivalent offered traffic and capacity are derived, GoS can be
derived with Erlang-B
• Required capacity can be calculated if offered traffic and GoS target is
known

• Consider the same 2 services sharing the same
resource:
• In this case the mean is:

   ibi ai   Erlangs  ai  (112)  (3 6)  30
• The variance is:

   ibi ai2   Erlangs  ai2  (12 12 )  (6  32 )  66
Note:  i bi = traffic in Erlangs

• Capacity Factor c (average trunks per connection):

 66
c   2.2
 30
• Equivalent offered traffic:
α 30
Equivalent Offered Traffic    13.63
c 2.2
• Trunk capacity for equivalent traffic at 2% GoS = 21


• Actual trunk requirement is trunk capacity x capacity
factor:
= 21 x 2.2 = 46.2 or 46 trunks
• Required Capacity is modified depending upon target

service for GoS (in service 1 Erlangs):
• Target is Service 1 C1=(2.2 x 21) + 1 = 47
• Target is Service 2, C2=(2.2 x 21) + 3 = 49
• Different services will require a different capacity for the

same GoS depending oin the amplitude
Traffic Analysis Methods Compared
• Optimistic if you use the smallest amplitude of trunk (39)
• Pessimistic if you use the largest amplitude of trunk (51)
• Post Erlang-B
• Pessimistic (55)
• Trunking efficiency improvement with magnitude ignored
• Campbell’s theorem
• Middle band (47 - 49)
• Different capacities required for different services - realistic
• Preferred solution for dimensioning, but not ideal...

Theorem
• Consider the following service definition and traffic

forecast.
Service Amplitude Forecast

Voice 1 250 E
HSCSD 2 63 E
• Based on a theoretical availability of 15 voice trunks per

cell and using voice as the ‘benchmark’ service, determine
the number of cells required to serve the above traffic
levels and the traffic offered per cell for each service

Theorem
• Assuming we have n cells, we can determine the loading

per cell.
250 63  2 376
mean   
n n n
250 63  2 2 502
variance   
n n n
variance 502
c   1.335
mean 376
mean 376 282
offered traffic   
c 1.335  n n

Theorem
• Ci  ai
Considering the equation: Capacity 
c
• Ci (available capacity) is predefined as 15. ai (amplitude)

depends on the service we use as our “benchmark” or priority
service. Choosing the voice service as the “benchmark”
service make ai equal to 1.
Ci 
15  1
 10.5
1.335
• 10.5 (rounded to 10) trunks will service 5.08 Erlangs.


Theorem
• Each cell requires 10 trunks to service 5.08 Erlangs.

• Therefore:
Number of cells required = Equivalent Traffic
Traffic per Cell
282
=  55.5
5.08
• Cell requirement is established at 56 cells.
• Each of the cells will service:
• 4.46 Erlangs of voice
• 1.13 Erlangs of HSCSD.

Theorem
• After placing sites on the coverage map and

spreading the traffic, the next stage is to assess the
cell loading (timeslots required).
• If mixed services are used, it is necessary to use

Campbell’s Theorem to assess the required number
of timeslots to satisfy the likely demand.
• Consider the case where a particular cell captures 7

Erlangs of voice and 2 Erlangs of HSCSD traffic that
requires 2 timeslots per connection.

Theorem
• Using Campbell’s Theorem:

mean  (7 1)  (2  2)  11
variance  (7  12 )  (2  2 2 )  15
15
Capacity Factor (c)   1.36
11
11
Equivalent offered traffic   8.09 Erlangs
1.36
From Erlang B, 14 trunks required.

Taking voice as benchmark : 14  1.36   1  20
• Hence 20 timeslots required.

Data may be packet switched, in which case it can be made to “fill the
gaps” in the demand for voice services.
Load
Peak traffic
Spare capacity for which can
be allocated to non real time
applications Average circuit
switched traffic
Real time non-controllable load
Time
• To evaluate the required cell capacity:
• First assume that the packet data can be scheduled to fill the spare real time
capacity.
• When all the spare real time capacity has been exhausted we must convert the
remaining capacity to an equivalent data capacity.
• One GSM timeslot can carry 13 kb/s of data.

Traffic Requirement Example
• A cell captures 2 Erlangs of voice traffic and is assigned a GSM carrier

with 7 timeslots. The grade of service is 2%.
Estimate the amount of data traffic that can be handled.
How much of this must be packet data and how much can be circuit
switched?
• Solution:
Timeslots available on average = 7 - 2 = 5
This represents a total data rate of 5 x 13 = 65 kb/s
2 Erlangs voice traffic requires 6 trunks (timeslots) for 2% blocking.
1 timeslot can be dedicated (circuit switched) for data, i.e. 13 kb/s

Remaining data must be packet switched = 65 - 13 = 52 kb/s.
Examples of GPRS-Specific Parameters
• Cell Reselection
• Service Exceptions
• Average Response Times
• Suspend/Resume Procedures
• Attach/Detach Procedures
• Routing Area (RA) Update Procedures
• PDP Context Activations
• Data Volumes and Rates
GPRS Transmission Connections

Um Gb Gn
GGS
BSS SGSN
N
MS
IMSI/NSAPI
PDP Context
Logical Link Control TLLI

Connection
DLCI
Data Link Connection
TID
Virtual Tunnel
CONNECTION
CONNECTION TYPE
IDENTIFIER
GPRS Cell Reselection
Cell Reselection – Subscribers in READY state

Number of Number of cell
Octets reselections
Average Number of Octets Affected

Average Number of Octets Deleted
Average Number of Frames Deleted
Number of Cell Reselections
| 0000100002 | 0000100003 | CI
Cell Reselection per source cell

GPRS Service Exception
Number of Frames
Deleted
Number of frames deleted per TLLI

Suspend / Resume Procedures
• Suspend Procedure:
• Enables a GPRS Class B MS to suspend GPRS mode in
order to make a circuit-switched connection
• Resume Procedure:
• Enables a GPRS Class B MS to resume GPRS mode having
made a circuit switched connection
• Monitoring:
• Procedure can be monitored and failures can be displayed
by TLLI, cell or cause
GPRS Average Response Times
Time in millsecs
BVCI
Average response time after paging, per BVCI

Number of Attach Rejects per Cell

Attach Rejects
Number of Attach Reject Messages per Cell


Number of Attach Rejects per TLLI
Attach Rejects
Number of Attach Reject Messages per Subscriber (TLLI)

Number of Attach Procedures

Procedures
Attach Request Attach Accept Attach Complete Attach Reject
Number of Attach Procedure Messages

RA Updates
Number
RA Request RA Accept RA Reject RA Request without response
Number of RA Update Messages

Number of
RA Rejects per Cause
RA Rejects
Unknown Causes Network Failures
Number of RA Update Rejects by Cause

GPRS PDP Contexts

Number PDP PDP Context Activation by MS
Context
Activations
PDP Contexts Activated PDP Contexts Accepted PDP Contexts Rejected
Number of PDP Context Activation Messages

DL LLC Rate
Bytes
Time (minutes)
Downlink Data Rate of a Cell

DL LLC Global Size

Volume (bytes)
Time (minutes)
Downlink Data Rate of a Mobile Station

Section 9
NSS Characteristics
and
Problem Types
Section 9 – NSS Characteristics and Problem Types
Section 9
9.1 Introduction
9.2 NSS Topology
9.3 SS7 Signalling in the NSS
?.6 Database Usage

Section 10
Network Subsystem (NSS)

Optimisation
Section ? – NSS Optimisation
Section 10
?.1 Introduction
?.7 NSS Optimisation Process
?.2 SS7 Signal Routing Analysis
?.6 Database Usage

Section 10 – NSS Optimisation
NSS Optimisation Process

NSS Data Capture
Phase
Data Analysis Phase
Define Optimisation
Requirements
Define Implementation
Strategy
Phase
Section 10 – NSS Optimisation
NSS Data Capture
• SS7 signal routing

• SS7 link loading
• User traffic routing
• Trunk group configuration and loading
• NSS configuration parameters
• Database loading
SS7 Routing
VLR
B
BSC
A AuC
MSC D
H
A C
BSC
E F HLR
BSC A
F EIR D
MSC
A
B
BSC
VLR
Routing Analysis – Strategy Consistency
• Strategy for signal routing should be consistent

in order to:
• Optimise signalling link resources
• Ease engineering by applying a consistent approach
• Spread traffic loads evenly across the network
• Increase reliability
• Simplify future growth planning
Routing Analysis - Transit Layer Switching

SMS HLR/SMS/VMS/IN/
HLR 5 VMS GPRS Layer
HLR 3 HLR 4
1200k 600k
HLR 2 150k IN
300k
HLR 1 GPRS
600k
TSC TSC TSC Layer

1a TSC TSC 4a
TSC TSC TSC TSC
1b 2b 2a 3b 4b 3a
MSC MSC
MSC MSC MSC
MSC MSC MSC Layer
MSC MSC MSC MSC
MSC MSC MSC MSC
MSC MSC MSC MSC
MSC MSC MSC
MSC
MSC
Routing Analysis – Transit Layer Switches
• Transit Layer switching provides an additional

layer of switching that makes signal routing
more efficient
• Signalling traffic should be evenly distributed
across TSCs in TSC layer
• If TSCs are used, MSCs should not be used
as STPs as this reduces MSC and SS7
performance
Routing Analysis – Alternate Routing
• Each signalling destination should be reachable by

a minimum of a main and alternate route,
particularly HLR connections
• Too few routes can cause loss of signalling
messages
• Too many routes is inefficient and overly-complex to
manage
SS7 Link Load Evaluation
• Monitoring of signalling link loading should be

an ongoing process
• ITU-T Q706 defines a recommended
signalling traffic capacity of 200mE per link.
• This limit can be exceeded for short periods
Link Loading – Signal Load Calculation
• Signal Load (mE) =
(No. MSUs x 6) + (No. Payload Octets)

8 x length of time (secs)
Link Loading - Signalling Load Sharing

Destination Link No Rx [mErl] Tx [mErl]
HLR4 0 305 121
1 307 134
2 304 113
3 305 127
4 303 84
5 303 99
6 302 76
7 302 91
8 104 94
9 104 109
10 103 89
11 104 102
12 104 58
13 103 72
14 103 51
15 104 64
MSU per sec
H
LR
1
H (
LR Ra
2 ha
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
(R ha
ah n )
M aha
S
CE n )
M
M AM
S
C
M KJ
S
CR
Al M A
gh S C
ad S
ir B
H M
LR SC
3
(B
M H
S )
C
M BE
S H
CF
Exchange
M AR
H S
LR C
5 LC
(E T
m
LC a m
EM TT )
S
Total MSU per sec
AM C1
T
M SC
S 1
FA CK
TE AN
H
M
S
C
Tx
Rx
Link Loading – Symmetrical Load Spreading

Traffic Load Evaluation
• Data gathered from MSCs at regular intervals

to assess capacity on each link
• Trunk capacities between MCs and between
PLMN-PSTN assessed
Network Parameter Analysis
• Parameters control equipment operation and

performance
• Initial parameters determined at network planning
stage
• Parameter sets should remain common where
possible to facilitate efficient network management
• Parameters may require modification to suit local
conditions.
• Periodic operation and performance monitoring
necessary to ensure parameters remain valid
Parameter Analysis – VLR Parameters

SUBSCRIBER PROFILE
TMSI Usage
Authentication Usage
Implicit Detach Usage
Emergency Call
Time Implicit IMSI Detach Time Limit
Limits Call Waiting Time
VLR Cleaning Start Time (for Siemens If capacity above 80%)
Minimum Amount of Triplets to be kept on record (MINT)
SUBSCRIBER SERVICES SUPPORT
Calling Line Identification Presentation (CLIP)
Call Forwarding
Call Waiting (CW)
Call Hold (HOLD)
Advice of charge charging (AOCC)
Call Barring
TELESERVICE SUPPORT
Short Message Mobile Terminated (T21)/Originated (T22)
Fax gr.3 and Alternate Speech (T61)/ Automatic fax group 3 (T62)
• Advantages of TMSI usage:

• Increases paging channel (PCH) capacity
• Reduces VLR/HLR signalling requirement
• Provides greater security than using IMSI only
• TMSI should be unique within a LA
• New TMSI issued on change of LA
• TMSI allocation/update occurrences set by Operator
• Too many TMSI update occurrences increases signalling
capacity requirements
• Use of TMSI should be uniform across network
PARAMETER REC

IMSI Attach Y
Location Update Y
TMSI Periodic Location Update N
ALLOCATION Mobile Originating Call N
COUNTERS Mobile Originating SMS N
• Advantages of Au usage:
• Implementation of Au is an Operator decision
• Can be implemented internally and/or for roaming
subscribers
• Excessive Au increases signalling capacity requirements
• Use of Au should be uniform across network

PARAMETER REC
Triplet Reuse Y
MSRN Life Time (Sec) 20
Country Code Length 2
IMSI Attach Y
AUTHENTICATION Location Update N
COUNT Periodic Location Update N
Parameter Analysis – IMEI Parameters
• Advantages of IMEI usage:

• Enables tracing/monitoring of Grey/Black-listed MSs
• Implementation of Au is an Operator decision
• Can be implemented internally and/or for roaming
subscribers
• Excessive Au increases signalling capacity requirements
• Use of Au should be uniform across network
Parameter Analysis – IMEI Check Parameters

PARAMETER REC
IMSI Attach Y
Location Update Y(10)
Periodic Location Update N
IMEI CHECKING ON…. Mobile Terminating Call N
Black List Effect BLOCK
Grey List Effect TRACE
Unknown IMEI Effect BLOCK

• VLR capacity should be evenly spread where possible
• 75% capacity limit is recommended to allow for peaks
and future expansion
• Options for reducing VLR loading include:
• Redistribution of BSCs amongst MSCs
• Redesigning LA boundaries
• Too large - excessive paging requirement
• Too small - excessive location update requirement
• Modify parameter settings to reduce signalling occurrences
• VLR Cleaning
• Implicit Deregistration
MSC VLR Utilization

MSC 1 65%
MSC 2 82%
MSC 3 26%
MSC 4 100%
MSC 5 86%
MSC 6 66%
MSC 7 71%
MSC 8 100%
MSC 8 62%
MSC 9 7%
MSC 10 74%
NSS - VLR Cleaning

• Reduces capacity loading within a VLR (Nokia proprietary)
• Deletes subscriber records that are inactive for a specified
period of time
• Compares timestamp value in subscriber profile with
inactivity timer.
• Timer is reset when sub scriber record is accessed
• If timer value > timestamp, subscriber record is marked for
deletion
Database Usage – Implicit Deregistration
• Procedure to reduce loading within VLRs by reducing

paging requirement
• Each subscriber record has an associated Implicit
Deregistration timestamp.
• The timestamp is reset each time an Implicit Detach
procedure takes place.
• If no contact has been made when the timer
threshold is reached, the MSC ceases paging the
subscriber’s MS.
NSS Databases – VLR Utilisation
VLR
• VLR utilisation refers to subscriber record Sr. No. MSC Utilisation
capacity and usage within a VLR
1 MSC-1 65%
• Utilisation problems include: 2 MSC-2 82%
• Over-capacity in certain VLRs 3 MSC-3 26%
• Uneven spread of records across the 4 MSC-4 100%
network
5 MSC-5 86%
• Over-capacity can result in: 6 MSC-6 66%
• Call function failures (e.g. setup/handover, 7 MSC-7 71%
authentication etc)
8 MSC-8 100%
9 MSC-9 62%
10 MSC-10 7%
NSS Databases – HLR Utilisation

• HLR utilisation refers to
subscriber record capacity and
usage within a HLR
%AGE USAGE OF HLRs
• Utilisation problems include:
100
• Over-capacity in certain HLRs 90
• Uneven spread of records across 80
70
the network
60
• Over-capacity can result in: 50

40
%AGE USAGE
• Call function failures (e.g. 30

20
setup/handover, authentication
10
etc) 0
HLR1 HLR2 HLR3 HLR4 HLR5
• Signalling bottlenecks
NSS Traffic – Trunk Utilisation

Addit.
NSS Call NSS Total Trunks Trunk Trunks

Trunk Gp Time Trunks Attempts ABR% Erlangs Reqd Utilisation Required
1 11-12 1455 128,394 46% 1177.2 1207 83% 154
2 11-12 180 6,783 51% 120.9 139 77% 5
3 11-12 210 12,324 32% 162.7 183 87% 34
4 11-12 1375 104,514 58% 1281.3 1311 95% 373
5 11-12 1527 98,192 58% 1250.2 1280 84% 180
GoS = 1% Max Erlangs/trunk = 0.7 Only trunk utilisation > 75% shown
End of Course
Any Questions?

GSM Network Performance Management and Optimisation

Uploaded by

Copyright:

Available Formats

GSM Network Performance Management and Optimisation

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

GSM Network Performance Management and Optimisation

Uploaded by

Copyright:

Available Formats

GSM

Network Performance Management

1. Review of GSM Principles

Review of GSM Principles

1.2 Physical Channel Structure

1.3 Logical Channel / Multiframe Formats

1.4 Review of the BCH Carrier

1.5 Paging Procedures

1.6 Timing Advance

1.7 Cell Reselection

1.9 Power Control

1.10 Frequency Hopping

Duplex spacing = 45 MHz

Duplex spacing = 95 MHz

TCH BCH CCCH DCCH

FCCH PCH SDCCH

SCH RACH SACCH

BCCH AGCH FACCH

• TCH Multiframe structure:

T = TCH S = SACCH I = Idle

• Control channel messages generally require 4 timeslots.

• A non-combined multiframe has 9 CCCH blocks:

• A complete paging or access grant message takes four bursts (timeslots),

CCCH_CONF Number of CCCH Configuration

MS2 - Timeslot 2 time time

MS2 - Timeslot 2 time time

Access burst data Guard Period

Access burst data

• TA signal is transmitted on SACCH as a number between 0 and 63 in

• TA value updates are sent to the MS on the SACCH

BCCH Allocation (BA) Lists

• Contains a list of BCCH carriers available in a certain

CA = 0 – not part of BA list

• BCCH carrier measurements are used for:

• MS measures neighbouring cells for:

• MS calculates average power level received from each cell:

• MS calculates a C1 parameter for each measured carrier based on the

• Mobile compares cells which give a positive value of C1 and ‘camps-on’

• C1 indicates that the path loss to the serving

Cell Re-selection – GSM Phase 1 Mobiles

• Between cells within a Location Area:

C1 (new) > C1 (old)

• Between cells on a Location Area boundary:

C1 (new) > C1 (old) + OFFSET

Cell Re-selection – GSM Phase 2 Mobiles

• GSM Phase 2 introduced a separate cell re-selection

C2 = C1 + OFFSET – (TEMPORARY_OFFSET x H(PENALTY_TIME –T)

C2 Cell Selection Parameters

C2 = C1 + OFFSET – (TEMPORARY_OFFSET x H(PENALTY_TIME –T)

• Cells in different BSSs (different BSCs) but under

GSM handovers are all ‘hard’ – i.e. mobile only BSC

Measure serving and neighbouring

Compare measurements with

Initiate handover procedure

Determine best candidate

Select cell and perform handover

Handover Procedure - Command Message

Structure of the message sent to MS by original BSS: