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 KPIs 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
1.2 Physical Channel Structure
1.6 Timing Advance
1.3 Logical Channel / Multiframe Formats
1.7 Cell Reselection
1.9 Power Control
1.8 Handovers
1.5 Paging Procedures
1.4 Review of the BCH Carrier
1.10 Frequency Hopping
1.1 Introduction
P-GSM Physical Channels
Uplink Downlink
890 915 935 960 MHz
Duplex spacing = 45 MHz
Guard Band
100 kHz wide
Channel Numbers (n) (ARFCN)
200 kHz spacing
Range of ARFCN:
1 - 124
1 n
Guard Band
100 kHz wide
Fu(n)
2 3 4
0 1 2 3 4 5 6 7
4.615 ms
timeslot = 0.577 ms
1 frame period
Raw data rate =
33.75kbps per traffic channel
270kbps per carrier channel
DCS - 1800 Spectrum
Uplink Downlink
1710 1785 1805 1880 MHz
Duplex spacing = 95 MHz
Guard Band
100 kHz wide
Channel Numbers (n) (ARFCN)
200 kHz spacing
Range of ARFCN:
512 - 885
1 n
Guard Band
100 kHz wide
Fu(n)
2 3 4
GSM Logical Channels
TCH
Traffic
TCH/H
TCH/F
CCCH
Control
BCH DCCH
FCCH
SCH
BCCH
PCH
RACH
AGCH
CBCH
NCH
SDCCH
SACCH
FACCH
Two types of logical channel are defined; traffic and control channels
Each is further sub-divided as shown:
T T
Traffic Channel Multiframe
The TCH multiframe consists of 26 timeslots.
This multiframe maps the following logical channels:


TCH Multiframe structure:

T T T T T T T T T T T T T T T T I T T T T T
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 24 25 16 17 18 19 20 21 22 23
TCH
SACCH
FACCH
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.
T S
Control Channel Multiframe
The control channel multiframe is formed of 51 timeslots
CCH multiframe maps the following logical channels:
0 1 42-45 46-49 50 32-35 36-39 40 41 22-25 26-29 30 31 12-15 16-19 20 21 2-5 6-9 10 11
RACH
Uplink
Downlink
Downlink:
FCCH
SCH
BCCH
CCCH (combination of PCH and AGCH)
Uplink:
RACH

F = FCCH S = SCH I = Idle
S BCCH F CCCH S CCCH F CCCH S CCCH F CCCH S CCCH F CCCH S CCCH F CCCH I
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:


A non-combined multiframe has 9 CCCH blocks:





A complete paging or access grant message takes four bursts (timeslots),
i.e. one CCCH block.
S BCCH F CCCH S CCCH F CCCH S
SDCCH
0
F S F S F I
SDCCH
1
SDCCH
2
SDCCH
3
SACCH
0
SACCH
1
S BCCH F CCCH S CCCH F CCCH S CCCH F CCCH S CCCH F CCCH S CCCH F CCCH I
I
SACCH
0
SACCH
1
SDCCH
0
SDCCH
1
SDCCH
2
SDCCH
3
SDCCH
4
SDCCH
5
SDCCH
6
SDCCH
7
SACCH
2
SACCH
3
I I
Control Channel Configuration
CCCH_CONF
Number of CCCH
Timeslots
Configuration
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 cells 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
MS1 - Timeslot 1
1 2 3
MS2 - Timeslot 2
1 2 3
1 2 3
time
Timing Advance signal causes mobiles further from base station to transmit
earlier - this compensates for extra propagation delay
time
time
1 2 3
MS1 - Timeslot 1
1 2 3
MS2 - Timeslot 2
1 2 3
1 2 3
time
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
delay
Guard Period 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
Delay
3 timeslots
Downlink
Uplink
Timing
Advance
Actual delay
Uplink
TA value updates are sent to the MS on the SACCH
The Timing Advance technique is known as adaptive frame
alignment
0 1 2 3 4 5 6 7
0 1 2 3 4 5 6 7
BCCH Allocation (BA) Lists
Section 1 GSM Principles
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_LOWER RANGE2_HIGHER Octet 7
RANGE2_HIGHER Octet 8
RANGE3_LOWER Octet 9
RANGE3_LOWER RANGE3_HIGHER Octet 10
RANGE3_HIGHER RANGE4_LOWER Octet 11
RANGE4_LOWER RANGE4_HIGHER Octet 12
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
RxLev / BSIC
RxLev 85Bm
96dBm
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 / BSIC
RxLev
RxQual
Measurement
Reports
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)

Section 1 GSM Principles
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)
Section 1 GSM Principles
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


Section 1 GSM Principles
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
Cells within the same BSS (same BSC)
Cells in different BSSs (different BSCs) but under
the control of the same MSC
Cells under the control of different MSCs

GSM handovers are all hard i.e. mobile only
communicates with one cell at a time during the
process

Internal
External
BSC
MSC
VLR
BSC
MSC
VLR
BSC
Section 1 GSM Principles
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
Section 1 GSM Principles
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 Initiation
Section 1 GSM Principles
Handover Procedure - Command Message
Structure of the message sent to MS by original BSS:
Message Type
Cell Description
Handover Reference
Power Command
Channel Description
Frequency List
or
Mobile Allocation
Includes Frequency Hopping
information if required
Non - Frequency Hopping
Frequency Hopping
MS BSS
Handover Command
Section 1 GSM Principles
Handover Margin
N
o
m
i
n
a
l

c
e
l
l

b
o
u
n
d
a
r
y

BTS 1 BTS 2
Handover to BTS 2 Handover to BTS 1
Mobile remains with
BTS 1 or BTS 2
Hysteresis due to handover margin
Section 1 GSM Principles
Handover Procedure Signalling Example
Signalling for a basic Inter-BSC handover involving only one
MSC (Intra - MSC):
MS MSC BSS 1 BSS 2
Measurement report
Measurement report
Measurement report
Measurement report
Handover Required
Handover Request
Handover Command
Acknowledgement
Handover Command
Handover Access
Handover Detection
Physical Information
Handover Complete
Handover Complete
Clear Command
Clear Complete
Measurement report
Measurement report
Section 1 GSM Principles
GSM Power Control Functions
Section 1 GSM Principles
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
MS Class
Full Rate
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)
Power (mW) = Nominal maximum mean power output (milliwatts)
GSM class 1 deleted under GSM Phase 2 Specification
Source: ETSI GSM 02.06 (Version 4.5.2)
Power (dBm) = Maximum power output in dBm (+watts)
Section 1 GSM Principles
BTS Power Classes
Source: ETSI GSM 05.05 (Version 4.23.1)
Section 1 GSM Principles
GSM 900 DCS 1800
TRX
Power
Class
Maximum
O/P Power
(W)
TRX
Power
Class
Maximum
O/P Power
(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
Source: ETSI GSM 05.05 (Version 4.23.1)
Section 1 GSM Principles
System Class Sensitivity
GSM 900 Small MS -102dBm
GSM 900 Other MS -104dBm
GSM 900 Macro BTS -104dBm
GSM 900 M1 Micro BTS -97dBm
GSM 900 M2 Micro BTS -92dBm
GSM 900 M3 Micro BTS -87dBm
DCS 1800 MS Class 1 or 2 -100 dBm
DCS 1800 MS Class 3 -102 dBm
GSM 1800 M1 Micro BTS -102dBm
GSM 1800 M2 Micro BTS -97dBm
GSM 1800 M3 Micro BTS -92dBm
Adaptive Power Control Process
Section 1 GSM Principles
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


Source: ETSI GSM 05.08 (Version 4.22.1)
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
Section 1 GSM Principles
Discontinuous Reception (DRX)
Section 1 GSM Principles
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
TDMA Frame 1
F1 F1 F1 F1 F1 F1 F1 F1
TDMA Frame 2
F1 F1 F1 F1 F1 F1 F1 F1
TDMA Frame 3
F1 F1 F1 F1 F1 F1 F1 F1
TDMA Frame 1
F1 F1 F1 F1 F1 F1 F1 F1
TDMA Frame 2
F2 F2 F2 F2 F2 F2 F2 F2
TDMA Frame 3
F3 F3 F3 F3 F3 F3 F3 F3
TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7
TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7 TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7
Non-Frequency Hopping Carrier:
Frequency Hopping Carrier:
Section 1 GSM Principles
Frequency Hopping Sequence
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)
Hop Sequence
One TDMA frame is 4.6 ms long
Rate of hopping = 1/ (4.6 x 10
-3
) = 217 hops / second
F1 F2 F3 F4 F1 F2 F3 F4 F1 F2 F3 F4 Sequential:
Hop Sequence
F1 F4 F3 F2 F1 F4 F3 F2 F1 F4 F3 F2 Pseudo-Random:
TDMA Frame
Section 1 GSM Principles
Channel Timeslot Hopping Sequence
TDMA Frame 1
0 1 2 3 4 5 6 7 F1 0 1 2 3 4 5 6 7
TDMA Frame 2 TDMA Frame 3 TDMA Frame 4
0 1 2 3 4 5 6 7
0 1 2 3 4 5 6 7
F2
F3
Time
Section 1 GSM Principles
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
Tx
Rx
Fade position depends on
path difference in terms of
wavelengths
x
y
d
Fade when : (x +y) - d = n /2 where n = odd number
Section 1 GSM Principles
Reasons for Frequency Hopping (2)
Cells are subject to interference
from other cells using the same
carriers at the re-use distance
If the cells hop through a set of
frequencies in different
sequences, the effect of this
interference is reduced
C/I ratio is increased

A1
A1
A3
A3
A2
A2
C1
C3 C2
B1
B3 B2
Section 1 GSM Principles
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)
Section 1 GSM Principles
Example of 4-Frequency Hopping
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
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
Section 1 GSM Principles
Baseband Frequency Hopping
Fixed TRX
Combiner
Antenna
Switch controller
Baseband
Data Signal
Fixed TRX
Fixed TRX
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
Section 1 GSM Principles
Synthesiser Frequency Hopping
Baseband
Data Signal
Synthesiser
TRX
Tuning controller
Antenna
Section 1 GSM Principles
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
Section 1 GSM Principles
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)
Section 1 GSM Principles
CM
MM
RR
LAPDm
TDMA
MS
RR
LAPDm
TDMA
BTSM
LAPD
G.703
BTSM
LAPD
G.703
BSSMAP
DTAP
BSSMAP
DTAP
CM
MM
Layer 1
Layer 2
Layer 3
U
m

A
bis
A
GSM Core
Network
PSTN MSC
BTS
BTS
BSC
GSM Layer 3 Messaging
BTS
BSC MSC
RR
SCCP SCCP
MTP MTP
Section 1 GSM Principles
Layer 3 Message Format
Section 1 GSM Principles
8 7 6 5 4 3 2 1
1 Transaction ID Protocol Discriminator
2 0 Message Type
k Information Element ID (IEI)
k+1 IE Content
k+2 IE Content


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
Message Header
Group
Payload
Group
Octets
Section 2

Network Characteristics
and
Problem Types
Section 2
2.1 Introduction
2.3 BSS Capacity Issues
2.2 BSS Coverage Issues
2.4 Quality of Service (QoS) Issues
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
Causes of BSS Coverage Degradation
Cell Balancing
Antenna Configuration
Equipment Performance
Section 2 Network Characteristics & Problems
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









An unbalanced cell could drop many calls in the fringe region
Cell Balancing
Uplink limit Downlink limit
Unbalanced system
Uplink limit
Downlink limit
Balanced system
Section 2 Network Characteristics & Problems
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 P
out (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: P
inMS
= P
oBS
- L
c
- L
d
- L
fb
+ G
ab
- L
p
+ G
am
- L
fm

Uplink: P
inBS
= P
oMS
- L
fm
+ G
am
- L
p
+ G
dBS
+ G
ab
- L
fb
- L
d

When the mobile is at the extreme boundary of the cell:
P
inMS
= P
refMS
P
inBS
= P
refBS

These are the reference sensitivities of the MS and BTS
The output levels P
oBS


and P
oMS
are the maximum allowed values
If the boundaries for uplink and downlink are the same, the path loss L
p
will be the
same in each direction
Subtracting the uplink equation from the downlink gives the system balance
equation:
P
oBS
= P
oMS
+ L
c
+ G
dBS
+ ( P
refMS
- P
refBS
)

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 (P
oBS
) 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 P
oMS
to maintain balance)
Altering the gain of the base station antenna - G
ab
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
Rotate original antennas through 30
o

Add new sites as shown
New site
Old site
rotated
New cell
Section 2 Network Characteristics & Problems
Effect of Antenna Alignment on Interference
Directional pattern of sectored
antennas reduces response to
interference
Increases C/I significantly
Allows greater frequency re-use, i.e.
smaller cells
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

A
B
Section 2 Network Characteristics & Problems
Antenna Tilt
Down tilt of antennas often used to:
reduce interference
adjust cell size
direct coverage e.g. into a building
Mechanical 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
Omni-directional antenna with electrical down tilt
Section 2 Network Characteristics & Problems
Examples of Antenna Tilt
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
No Tilt
Mechanical
Downtilt
Electrical
Downtilt
Electrical Downtilt +
Mechanical Uptilt
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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, 10 = 3 metres

Better isolation between the two signals can
also be obtained by mounting the antennas
at different heights on the tower
10
Section 2 Network Characteristics & Problems
Space Diversity Antenna Systems
Plan views of two possible tri-sectored site antenna
systems

Tx
Tx
Tx
Rx
Rx
Rx
Rx
Rx
Rx
Transmit antenna is separate
from the receivers
Rx
Rx
Rx
Tx/Rx
Tx/Rx
Tx/Rx
One antenna is used for transmit and
receive, using a duplexer in the BTS to
direct the signal
Section 2 Network Characteristics & Problems
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 45
o


This gives good coverage of vertical
polarisation and strong components of
rotated signals
Dipoles
crossed at 45
o

Section 2 Network Characteristics & Problems
Deciding whether to use Diversity
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.
Uplink limit
Downlink limit
Unbalanced system
Uplink limit Downlink limit
Balanced system
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
Causes of Equipment Performance Degradation
Electronic component degradation
Environmental conditions:
Wind, rain, temperature
Water ingress into connectors
Deterioration of feeders

Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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)
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
Long Call Setup Times
Inadequate SDCCH / TCH capacity available
Locating recipient of Call
Roaming
Inadequate paging capacity at called MT end
Section 2 Network Characteristics & Problems
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
Section 2 Network Characteristics & Problems
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 2 Network Characteristics & Problems
Section 3

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


Section 3 Introduction to Performance Management
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 operators business position


Section 3 Introduction to Performance Management
Performance Management Cycle
Monitor Network
Analyse Data
Yes
Identify
Problems
Implement
Changes
No
Initial Network
Design and
Implementation
Performance Management
Optimisation
QoS
Target
s Met?
Section 3 Introduction to Performance Management
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.
Section 3 Introduction to Performance Management
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

Section 3 Introduction to Performance Management
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
Section 3 Introduction to Performance Management
Performance Measurement Methods
Generally, three methods:

Drive Testing
OMC Statistical Testing
Protocol Analyser Testing
Section 3 Introduction to Performance Management
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

Section 3 Introduction to Performance Management
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
Section 3 Introduction to Performance Management
Drive Test Data Analysis Screens
Examples of analysis
screens in NEPTUNE
Call Success
Handover
Section 3 Introduction to Performance Management
OMC Data Analysis Screens
Example of analysis screen in OPTIMA:
Section 3 Introduction to Performance Management
Section 4

Performance Measurement
Metrics
Section 4
4.2 Key Performance Indicators
4.1 Introduction
4.3 BSS KPI Definitions
Section 4 Performance Metrics
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
KPIS 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.)
Section 4 Performance Metrics
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
Section 4 Performance Metrics
) _ _ ( _ _ _
_ _ _ _
handovers incoming successful Setups Call Completed ly Successful
Drop Fail HO Losses RF TCH
DCR
+
+
=
Call Success Rate (CSR)
Measures the proportion of call setup attempts that are
successfully established and terminated normally (ie. not
dropped)
Section 4 Performance Metrics
) 1 ( DCR CSSR CSR =
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.
Section 4 Performance Metrics
Attempts Handover Total
handovers Successful Total
HOSR
_ _
_ _
=
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)
Section 4 Performance Metrics
Attempts Handover Total
Drop Fail HO Total
HOFR
_ _
_ _ _
=
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.
Section 4 Performance Metrics
Attempts Assignment TCH Total
s Assignment TCH Blocked
Cong TCH
_ _ _
_ _
_ =
SDCCH Channel Blocking (SD_Cong)
Measures the proportion of SDCCH Assignment Attempts
that are blocked due to lack of SDCCH resources.
Section 4 Performance Metrics
Attempts Assignment SDCCH Total
s Assignment SDCCH Blocked
Cong SD
_ _ _
_ _
_ =
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)
Section 4 Performance Metrics
Setups Call Completed ly Successful
Losses RF TCH Total
Loss RF TCH
_ _ _
_ _ _
_ _ =
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.
Section 4 Performance Metrics
Attempts Assignment TCH Total
Failures Assignment TCH
Flr Ass TCH
_ _ _
_ _
_ _ =
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.
Section 4 Performance Metrics
Attempts Assignment mediate Total
s Assignment mediate Successful
Suc Acc SD
_ _ Im _
_ Im _
_ _ =
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.)
Section 4 Performance Metrics
Attempts Setup Call Valid
Failures Setup Call BSS Failures Setup Call Total
CSFR BSS Non
_ _ _
_ _ _ _ _ _
_

=
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 TCHs and Erlang-B formula.
Section 4 Performance Metrics
GOS Erlangs l Theoretica
Carried Erlangs
n Utilizatio
% 2 _ _
_
=
Section 5

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

Section 5 Drive Testing
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)
Section 5 Drive Testing
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:
Connector for phone 1 (MS1)
Connector for handset 2 (MS2)
Charging switch (on/off)
for second phone (MS2)
GPS Receiver
Connector
Charging switch (on/off)
for first phone (MS1)
Cradle for handset 1 Cradle for handset 2
Handset 1 to
laptop connector
Handset 2 to
laptop connector
GPS to Laptop
connector
12V Power Supply
Status Indicator
Section 5 Drive Testing
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


Section 5 Drive Testing
Test Mobile Data
For each measurement taken
the:

BCCH
BSIC
TCH

are identified


Section 5 Drive Testing
Test Mobile Data
For each measurement taken
the
RxLev Full
RxLev Sub
are identified

Full = monitored TCH
Sub = DTX TCH


Section 5 Drive Testing
Test Mobile Data
For each measurement taken
the
RxQual Full
RxQual Sub
are identified

Full = monitored TCH
Sub = DTX TCH


Section 5 Drive Testing
Test Mobile Data
For each measurement taken
the Timing Advance Value (TA)
value is recorded
Section 5 Drive Testing
Test Mobile Data
For each of the best six
neighbouring cells, the following
are recorded:

BCCH n
RxLev n
BSIC n
Section 5 Drive Testing
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.

Section 5 Drive Testing
Neighbour Data
RxLev data can be displayed
for the six best neighbour cells
and compared with the serving
cells RxLev

This can be used to identify
neighbouring cells that do not
currently exist in the BA list of
the serving cell


Section 5 Drive Testing
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
Section 5 Drive Testing
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

Section 5 Drive Testing
Test Mobile Data vs BSC Statistics
A test mobile logging tool such as Aircoms 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 5 Drive Testing
Section 6

Measuring Network Performance
The OMC
Section 6
6.1 Introduction
6.3 Statistical Data
6.2 Role and Function of the OMC
Section 6 The OMC
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

Section 6 The OMC
OMC Functions
Fault and Alarm Management
Configuration and Operations Management
Performance Management
Section 6 The OMC
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%):


Blocking (GoS) (2%-5%):



Traffic Throughput:

Traffic for one subscriber = no of calls per hour per subscriber * MHT
3600
100 =
tiveCalls NumberOfAc
oppedCalls NumberOfDr
lRate DroppedCal
100

=
llAttempts NumberOfCa
s llSuccesse NumberOfCa llAttempts NumberOfCa
GOS
Section 6 The OMC
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 6 The OMC
Section 7

Introduction to Optimisation
Section 7
Section 7 Optimisation Overview
7.2 Requirements for Optimisation
7.4 Network Audit Phase
7.3 Outline Optimisation Process
7.5 Network Performance Review Phase

7.6 Activity Phase
7.1 Introduction
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.


Section 7 Optimisation Overview
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
Section 7 Optimisation Overview
Requirements for Optimisation
Network issues that may generate a requirement for optimisation include:
Perceived reduction in network quality
Indications from network performance monitoring
Subscribers 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
Monitor Network
Analyse Data
Yes
Identify
Problems
Implement
Changes
No
Initial Network
Design and
Implementation
Performance Management
Optimisation
QoS
Target
s Met?
Section 7 Optimisation Overview
Outline Optimisation Process
Initial network design
and roll-out phase
Monitoring/
Network Audit Phase
Optimisation Activity
Phase
Design Review and
growth phase
Section 7 Optimisation Overview
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
Section 7 Optimisation Overview
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


Section 7 Optimisation Overview
Network Audit Process
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?

Section 7 Optimisation Overview
Network Audit Process
OMC Statistics Review:
Call Success Rate
Call Setup Rate
TCH Blocking
TCH Assignment Failure (RF)
SDCCH RF Loss
SDCCH Access performance
Dropped Calls
Handovers


Section 7 Optimisation Overview
Network Audit Process
A Interface Analysis:

Call Setup Failures
Location Update Success Rate
Handover Causes

Section 7 Optimisation Overview
Network Audit Process
Call Trace Analysis:

DL receive level and BTS power
UL receive level and MS transmit power
UL and DL RxQual distributions
Section 7 Optimisation Overview
Network Audit Process
Events and Alarms:
DL receive level and BTS power
UL receive level and MS transmit power
UL and DL RxQual distributions


Section 7 Optimisation Overview
Network Audit Process
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


Section 7 Optimisation Overview
Network Audit Process
Drive Test Metrics - Events :
Dropped call events
Call setup failure events
Handover failure events
Successful handover events

Section 7 Optimisation Overview
Network Audit Process
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)

Section 7 Optimisation Overview
Activity Phase of Optimisation
Start Optimisation Activity
Identify and Fix Hardware Problems
Ongoing
Performance
Measurement
Process
Review Process and Results
Feedback to modify
strategy decisions
BSS Database
Parameters Review:
Settings
Consistency
Change Control
Design Review and Growth Plan
Identify and Fix Neighbour Problems
Identify and Fix Frequency Plan Problems
Network Audit Procedures
Feedback
to
Audit
Procedures
Section 7 Optimisation Overview
Section 8

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


Section 8 Optimisation Activities
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



Section 8 Optimisation Activities
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


Section 8 Optimisation Activities
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


Section 8 Optimisation Activities
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 (MSCBSC BTS)
Only implement during low-traffic periods
Use test BSS where available
Avoid simultaneous multiple parameter changes
Section 8 Optimisation Activities
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
Section 8 Optimisation Activities
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

Section 8 Optimisation Activities
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
Section 8 Optimisation Activities
BCCH and TCH Frequency Plans
Section 8 Optimisation Activities
BCCH TCH Hopping MICRO
Guard Band Guard Band
14 ch
1 ch
26 ch 8 ch
1 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


Section 8 Optimisation Activities
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
Section 8 Optimisation Activities
Examples of Antenna Tilting
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
-35
-30
-25
-20
-15
-10
-5
0
No Tilt
Mechanical
Downtilt
Electrical
Downtilt
Electrical Downtilt +
Mechanical Uptilt
Section 8 Optimisation Activities
Antenna Selection
Factors to consider with Antenna selection:

Vertical and Horizontal Beamwidth
Gain
Front-to-Back Ratio
Null Fill
Downtilt (electrical/mechanical)
Section 8 Optimisation Activities
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 8 Optimisation Activities
Section 9

NSS Characteristics

Section 9
9.1 Introduction
9.3 Overview of SS7 Signalling in the NSS
Section 9 NSS Characteristics
9.2 NSS Topology
GMSC
MSC
PLMN Configuration
HLR
AuC
EIR
MSC
VLR
VLR
BSC
BSC
BSC
BSC
Traffic Links/Link Sets
Signalling Links/Link Sets
Section 9 NSS Characteristics
Big BSC Configuration
Section 9 NSS Characteristics
Transit Switching Layer
TSC
1a
HLR 2
300k
HLR 3
150k
HLR 4
1200k
HLR 5
600k
HLR 1
600k
TSC
3b
TSC
4b
TSC
3a
TSC
1b
TSC
4a
TSC
2b
TSC
2a
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
IN
GPRS
VMS
SMS
HLR/SMS/VMS/IN
Layer
TSC Layer
MSC Layer
Section 9 NSS Characteristics
A Interface Configuration
MSC
BSC
0 Frame Alignment
1 TS1 TS2 TS3 TS4 fm Trunk 1
2 TS1 TS2 TS3 TS4 fm Trunk 2
3 TS1 TS2 TS3 TS4 fm Trunk 3
4 Trunk 1 TS0 Data fm Trunk 1
5 TS5 TS6 TS7 TS8 fm Trunk 1
6 TS5 TS6 TS7 TS8 fm Trunk 2
7 TS5 TS6 TS7 TS8 fm Trunk 3
8 Trunk 2 TS0 Data fm Trunk 2
:
:
:
:
:
:
:
:
:
:
13 TS13 TS14 TS15 n/u fm Trunk 1
14 TS13 TS14 TS15 n/u fm Trunk 2
15 TS13 TS14 TS15 n/u fm Trunk 3
16 C7 Signalling
17 TS17 TS18 TS19 TS20 fm Trunk 1
18 TS17 TS18 TS19 TS20 fm Trunk 2
19 TS17 TS18 TS19 TS20 fm Trunk 3
:
:
:
:
:
:
:
:
:
:
30 TS29 TS30 TS31 n/u fm Trunk 2
31 TS29 TS30 TS31 n/u fm Trunk 3
TRUNK 1
TRUNK 2
TRUNK 3
TRUNK 1
TRUNK 2
TRUNK 3
2Mbps E1 Link
Section 9 NSS Characteristics
TRAU Configurations
BTS Site
BSC Site MSC Site
TRAU
BTS Site
CCU
CCU
BSC Site MSC Site
BTS Site
CCU
CCU
BSC Site MSC Site
A
B
C
A
bis

U
m

A
MSC Node BSC Node CCU Channel Coding Unit
TRAU
TRAU
16kbps
16kbps
16kbps 16kbps
64kbps
64kbps
64kbps
64kbps
16kbps
CCU
CCU
Section 9 NSS Characteristics
GMSC
MSC
PLMN Interfaces
HLR
AuC
EIR
MSC
VLR
VLR
A
C
C
F
F
B
B
H
D
D
E
A
A
A
G
E
E
BSC
BSC
BSC
BSC
Section 9 NSS Characteristics
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)

Section 9 NSS Characteristics
C7 Network
GMSC
MSC
PLMN Interfaces C7 Signalling
HLR
AuC
EIR
MSC
VLR
VLR
A
C
C
F
F
B
B
H
D
D
E
A
A
A
G
E
E
BSC
BSC
BSC
BSC
Section 9 NSS Characteristics
Generic SS7 Network
SP
SP
SP
SP
STP
STP
SP
STP
STP
SP
SP
SP
SP
Section 9 NSS Characteristics
MSC
C7 in the GSM NSS
BSC
HLR
AuC
EIR
BSC
BSC
BSC
MSC
VLR
VLR
SP
SP
SP
SP
SP
SP
SP
SP
SP
Section 9 NSS Characteristics
STP
STP
MTP 1-3
A Interface Signalling
SCCP
DTAP
BSSMAP
User Data
BSSAPP
Layer 1-3
Section 9 NSS Characteristics
Over the A Interface, GSM uses C7 signalling with protocol
extensions for GSM-specific requirements.
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



A Interface Signalling Protocols
Section 9 NSS Characteristics
MTP
SCCP
TCAP
MAP
NSS Interface Signalling
HLR
Layers 4-6
MSC MSC VLR
E
C
B
D
Layer 3
Layers 1-2
Layer 7
C7 Protocol Stack
EIR
F
Section 9 NSS Characteristics
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


NSS Interface Signalling Protocols
Section 9 NSS Characteristics
Section 10

NSS Parameter Evaluation
and
Optimisation
Section 10
?.1 Introduction
?.4 Traffic Loading Evaluation
?.2 SS7 Signal Routing Analysis
?.5 Parameter Evaluation
?.6 Database Usage
?.3 SS7 Link Load Evaluation
?.7 NSS Optimisation Process
Section 10 NSS Parameters and Optimisation
NSS Optimisation Process
NSS Data Capture
Phase
Data Analysis Phase
Define Optimisation
Requirements
Define Implementation
Strategy
Optimisation Activity
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
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)


Monitoring Signalling Loads
Signal Load Evaluation Link Load Sharing
Excessive link loads can occur
when:
Original signal load dimensioning was
not correct
Network characteristics have changed


Ideal Situation:
Links should not exceed 200mE
Signalling traffic should be spread
evenly across all links in link set

TSC1 C7 Load C7 Load
Destination
Link
No Rx [mErl] Tx [mErl]
HLR1








0 305 121
1 307 134
2 305 127
3 303 84
4 302 76
5 312 91
6 104 94
7 104 109
8 103 89
9 104 102
Signal Load Evaluation Link Load Symmetry
Ideal Situation:
Individual links should not exceed
200mE
Tx/Rx loads should be as symmetrical
as possible
Benefits:
Simplified routing
Reduces risk of lost MSUs
Reduced risk of signalling loops
More efficient bearer usage



TSC1 C7 Load C7 Load
Destination Link No Rx [mErl] Tx [mErl]
MSC3 0 15 483
1 19 437
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
Total MSU per sec
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
H
L
R
1

(
R
a
h
a
h
a
n
)
H
L
R
2

(
R
a
h
a
h
a
n
)
M
S
C
E
M
A
M
M
S
C
K
J
M
S
C
R
A
M
S
C
S
B

A
l
g
h
a
d
i
r

M
S
C
H
L
R
3

(
B
H
)
M
S
C
B
E
H
M
S
C
F
A
R
M
S
C
L
C
T
H
L
R
5

(
E
m
a
m
)
L
C
T
T
S
C
1
E
M
A
M
T
S
C
1
M
S
C
K
A
N
F
A
T
E
H
M
S
C
Exchange
M
S
U

p
e
r

s
e
c
Tx
Rx
Recommended Value per Node
H
L
R

1

H
L
R

2

M
S
C
1

M
S
C
2

M
S
C
3

M
S
C
4

M
S
C
5

H
L
R
3

M
S
C
6

M
S
C
7

M
S
C
8

H
L
R
4

T
S
C
1

T
S
C
2

M
S
C
9

M
S
C
1
0

Signal Load Evaluation Nodal MSU Capacity
Recommended Nodal MSU Capacity <= 1500 MSUs/second
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 Analysis TMSI Parameters
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 Analysis Authentication Parameters
PARAMETER
REC
Location Update New Visitor Y
IMSI Attach Y
AUTHENTICATION Location Update N
COUNT Periodic Location Update N
Mobile Originating Call N
Mobile Originating SMS N
Mobile Terminating Call N
Mobile Terminating SMS N
Mobile Terminating USSD N
Mobile SS Operation N
Parameter Analysis IMEI Check Parameters
PARAMETER REC
Location Update New Visitor Y
IMSI Attach Y
Location Update Y(10)
Periodic Location Update N
Mobile Originating Call N
IMEI CHECKING ON. Mobile Terminating Call N
Mobile Originating SMS N
Mobile Terminating SMS N
Mobile Terminating USSD N
Mobile SS Operation N
Black List Effect BLOCK
Grey List Effect TRACE
Unknown IMEI Effect BLOCK
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
NSS - Implicit Deregistration
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
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 MSRN Lifetime
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
NSS Call Waiting Time
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 utilisation refers to subscriber record
capacity and usage within a VLR
Utilisation problems include:
Over-capacity in certain VLRs
Uneven spread of records across the
network
Over-capacity can result in:
Call function failures (e.g. setup/handover,
authentication etc)
Lack of growth capability
Sr. No. MSC
VLR
Utilisation
1 MSC-1 65%
2 MSC-2 82%
3 MSC-3 26%
4 MSC-4 100%
5 MSC-5 86%
6 MSC-6 66%
7 MSC-7 71%
8 MSC-8 100%
9 MSC-9 62%
10 MSC-10 7%
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 - VLR Cleaning
NSS Databases HLR Utilisation
HLR utilisation refers to
subscriber record capacity and
usage within a HLR
Utilisation problems include:
Over-capacity in certain HLRs
Uneven spread of records across
the network
Over-capacity can result in:
Call function failures (e.g.
setup/handover, authentication
etc)
Lack of growth capability
Signalling bottlenecks
0
10
20
30
40
50
60
70
80
90
100
HLR1 HLR2 HLR3 HLR4 HLR5
%AGE USAGE OF HLRs
%AGE USAGE
Section 11

Optimising Networks
for
New Services
Section 11
11.1 Introduction
11.3 GPRS Performance Monitoring
11.2 Dimensioning Networks For New Services
Section 11 New Service Optimisation
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
Campbells Theorem
Section 11 New Service Optimisation
Equivalent Erlangs
+
Low
Bandwidth
Equivalent
High
Bandwidth
Equivalent
Combine the two traffic sources together by
converting one to the bandwidth of the other
The trunking efficiency will VARY with the
bandwidth of equivalent Erlang that you choose!
Not suitable for use due to this property
2 Erlangs
of Low
Bandwidth
1 Erlang of
High
Bandwidth
Difference in
capacity
required for
same GoS
Section 11 New Service Optimisation
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.
Section 11 New Service Optimisation
Post Erlang-B
Combine the two traffic
sources together after
calculating required
capacity
The trunking efficiency
variation with magnitude
is not considered -
pessimistic about offered
traffic supported to the
same GoS
Not suitable for use due
to this property
1 Erlang of
Service A
1 Erlang of
Service B
+
1 Erlang and 1
Erlang of of
Service B
Illustration using 2 services of
same bandwidth
Difference
in capacity
required for
same GoS
Accepted correct method
Section 11 New Service Optimisation
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 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
Section 11 New Service Optimisation
Consider 2 services requiring equal resource:
Service 1: uses 1 trunk per connection. 12 Erlangs of traffic.
Service 2: uses 1 trunk 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.
Adding these together gives 31 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.
Post Erlang-B
Section 11 New Service Optimisation
Campbells Theorem
Campbells theorem creates a composite distribution where:







The amplitude (a
i
) 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

( )
c
a C
Capacity
i i

=
c
ffic OfferedTra
o
=

= =
i
i i i
i
i i i
b a
b a
c

o
v
2
C
i
= available capacity
o = mean
u = variance

i
= arrival rate
a
i
= amplitude of service
b
i
= mean holding time
c = capacity factor
i i
b = Traffic Offered Service
Section 11 New Service Optimisation
Campbells Theorem Example
Consider the same 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.

In this case the mean is:


The variance is:



Note: = traffic in Erlangs


= + = = = 30 ) 6 3 ( ) 12 1 ( Erlangs
i i i i
a a b o

= + = = = 66 ) 3 6 ( ) 1 12 ( Erlangs
2 2 2 2
i i i i
a a b v
Section 11 New Service Optimisation
i i
b
Campbells Theorem Example
Capacity Factor c (average trunks per connection):



Equivalent offered traffic:




Trunk capacity for equivalent traffic at 2% GoS = 21
2 . 2
30
66
= = =
o
v
c
63 . 13
2 . 2
30
Traffic Offered Equivalent = = =
c

Section 11 New Service Optimisation

Campbells Theorem Example
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 C
1
=(2.2 x 21) + 1 = 47
Target is Service 2, C
2
=(2.2 x 21) + 3 = 49


Different services will require a different capacity for the
same GoS depending on the amplitude
Section 11 New Service Optimisation
Traffic Analysis Methods Compared
Equivalent Erlangs
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
Campbells theorem
Middle band (47 - 49)
Different capacities required for different services - realistic
Preferred solution for dimensioning, but not ideal...
Section 11 New Service Optimisation
Capacity Dimensioning with Campbells
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
Section 11 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Assuming we have n cells, we can determine the loading
per cell.
n n c
c
n n n
n n n
282
335 . 1
376 mean
traffic offered
335 . 1
376
502
mean
variance
502 2 63 250
variance
376 2 63 250
mean
2
=

= =
= = =
=

+ =
=

+ =
Section 11 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Considering the equation:

C
i
(available capacity) is predefined as 15. a
i
(amplitude)
depends on the service we use as our benchmark or priority
service. Choosing the voice service as the benchmark
service make a
i
equal to 1.



10.5 (rounded to 10) trunks will service 5.08 Erlangs.
c
a C
i i

= Capacity
( )
5 . 10
335 . 1
1 15
=

=
i
C
Section 11 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Each cell requires 10 trunks to service 5.08 Erlangs.
Therefore:

Number of cells required = Equivalent Traffic
Traffic per Cell

=

Cell requirement is established at 56 cells.
Each of the cells will service:
4.46 Erlangs of voice
1.13 Erlangs of HSCSD.
5 . 55
08 . 5
282
=
Section 11 New Service Optimisation
Assessing Cell Loading using Campbells
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
Campbells 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.

Section 11 New Service Optimisation
Assessing Cell Loading using Campbells
Theorem
Using Campbells Theorem:









Hence 20 timeslots required.
( ) 20 1 36 . 1 14 : benchmark as voice Taking
required. trunks 14 B, Erlang From
Erlangs 09 . 8
36 . 1
11
traffic offered Equivalent
36 . 1
11
15
) ( Factor Capacity
15 ) 2 2 ( ) 1 7 ( variance
11 ) 2 2 ( ) 1 7 ( mean
2 2
= +
= =
= =
= + =
= + =
c
Section 11 New Service Optimisation
Evaluating Traffic Requirements
Real time non-controllable load
Spare capacity for which can
be allocated to non real time
applications
Peak traffic
Load
Time
Average circuit
switched traffic
Data may be packet switched, in which case it can be made to fill the
gaps in the demand for voice services.
Section 11 New Service Optimisation
Evaluating Traffic Requirements
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.
Section 11 New Service Optimisation
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.
Section 11 New Service Optimisation
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
Section 11 New Service Optimisation
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
SGSN
MS
GGS
N
BSS
G
b
G
n
U
m
PDP Context
Logical Link Control
Connection
Virtual Tunnel
Data Link Connection
IMSI/NSAPI
TLLI
DLCI
TID
CONNECTION TYPE
CONNECTION
IDENTIFIER
Section 11 New Service Optimisation
GPRS Cell Reselection
Section 11 New Service Optimisation
Cell Reselection per source cell
Cell Reselection Subscribers in READY state
Number of cell
reselections
Number of
Octets
Average Number of Octets Affected
Average Number of Octets Deleted
Average Number of Frames Deleted
Number of Cell Reselections
| 0000100002 | 0000100003 | CI
Frames Lost Through Service Exceptions
Section 11 New Service Optimisation
Number of frames deleted per TLLI
Number of Frames
Deleted
GPRS Average Response Times
Section 11 New Service Optimisation
Average response time after paging, per BVCI
Time in millsecs
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
Section 11 New Service Optimisation
GPRS Attach/Detach Procedures
Section 11 New Service Optimisation
Number of Attach Procedure Messages
Attach Request Attach Accept Attach Complete Attach Reject
Number of
Procedures
Attach Procedures
GPRS Attach/Detach Procedures
Section 11 New Service Optimisation
Number of Attach Reject Messages per Cell
Attach Rejects per Cell
Number of
Attach Rejects
GPRS Attach/Detach Procedures
Section 11 New Service Optimisation
Number of Attach Reject Messages per Subscriber (TLLI)
Number of
Attach Rejects
Attach Rejects per TLLI
GPRS RA Update Procedures
Section 11 New Service Optimisation
Number of RA Update Messages
RA Updates
RA Request RA Accept RA Reject RA Request without response
Number
GPRS RA Update Procedures
Section 11 New Service Optimisation
Number of RA Update Rejects by Cause
Number of
RA Rejects
RA Rejects per Cause
Unknown Causes Network Failures
GPRS PDP Contexts
Section 11 New Service Optimisation
PDP Context Activation by MS
PDP Contexts Activated PDP Contexts Accepted PDP Contexts Rejected
Number PDP
Context
Activations
Number of PDP Context Activation Messages
GPRS Data Volumes and Rates
Section 11 New Service Optimisation
Downlink Data Rate of a Cell
DL LLC Rate
Bytes
Time (minutes)
GPRS Data Volumes and Rates
Section 11 New Service Optimisation
Downlink Data Rate of a Mobile Station
Time (minutes)
Volume (bytes)
DL LLC Global Size
End of Course
Any Questions?

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)


Signalling Load Evaluation
Section 9 NSS Characteristics and Evaluation
Signal Load Evaluation Link Load Sharing
Excessive link loads can occur
when:
Original signal load dimension was not
correct
Network characteristics have changed


Ideal Situation:
Links should not exceed 200mE
Signalling traffic should be spread
evenly across all links in set

Section 9 NSS Characteristics and Evaluation
TSC1 C7 Load C7 Load
Destination
Link
No Rx [mErl] Tx [mErl]
HLR1 0 305 121
1 307 134
2 305 127
3 303 84
4 302 76
5 312 91
6 104 94
7 104 109
8 103 89
9 104 102
Signal Load Evaluation Link Load Symmetry
Ideal Situation:
Individual links should not exceed
200mE
Tx/Rx loads should be as symmetrical
as possible
Benefits:
Simplified Routing
Reduces risk of lost MSUs
Reduced risk of signalling loops
More efficient bearer usage



Section 9 NSS Characteristics and Evaluation
TSC1 C7 Load C7 Load
Destination Link No Rx [mErl] Tx [mErl]
MSC3 0 15 483
1 19 437
Section 9 NSS Characteristics and Evaluation
Total MSU per sec
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
H
L
R
1

(
R
a
h
a
h
a
n
)
H
L
R
2

(
R
a
h
a
h
a
n
)
M
S
C
E
M
A
M
M
S
C
K
J
M
S
C
R
A
M
S
C
S
B

A
l
g
h
a
d
i
r

M
S
C
H
L
R
3

(
B
H
)
M
S
C
B
E
H
M
S
C
F
A
R
M
S
C
L
C
T
H
L
R
5

(
E
m
a
m
)
L
C
T
T
S
C
1
E
M
A
M
T
S
C
1
M
S
C
K
A
N
F
A
T
E
H
M
S
C
Exchange
M
S
U

p
e
r

s
e
c
Tx
Rx
Recommended Value per Node
H
L
R

1

H
L
R

2

M
S
C
1

M
S
C
2

M
S
C
3

M
S
C
4

M
S
C
5

H
L
R
3

M
S
C
6

M
S
C
7

M
S
C
8

H
L
R
4

T
S
C
1

T
S
C
2

M
S
C
9

M
S
C
1
0

Signal Load Evaluation Nodal MSU Capacity
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
Section 9 NSS Characteristics and Evaluation
Using Authentication:
Increases access security on network
Increases signalling load on network

Authentication-Related Parameters:
IMSI Attach (Y)
Location Update (N)
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
Section 9 NSS Characteristics and Evaluation
IMEI Usage:
Increases access security on network
Increases signalling load on network

IMEI-Related Parameters:
IMSI Attach (Y)
Location Update (Y every 10
th
update)
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)
Black (block) Grey (trace)
NSS IMEI-Related Parameters
Section 9 NSS Characteristics and Evaluation
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
NSS - Implicit Deregistration
Section 9 NSS Characteristics and Evaluation
Reduces capacity loading within a VLR
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 erecord in
marked for deletion
Cleaning takes place during non-busy periods
Common settings are typically 24hrs to 3 days
NSS - VLR Cleaning
Section 9 NSS Characteristics and Evaluation
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 MSRN Lifetime
Section 9 NSS Characteristics and Evaluation
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
NSS Call Waiting Time
Section 9 NSS Characteristics and Evaluation
Section 9

Network Subsystem
Optimisation
Section 9
9.1 Introduction
9.3
9.2
Section 9 Network Subsystem Optimisation
Section 10

Optimising Networks
for
New Services
Section 10
10.1 Introduction
10.3 GPRS Performance Monitoring
10.2 Dimensioning Networks For New Services
Section 10 New Service Optimisation
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
Campbells Theorem
Section 10 New Service Optimisation
Equivalent Erlangs
+
Low
Bandwidth
Equivalent
High
Bandwidth
Equivalent
Combine the two traffic sources together by
converting one to the bandwidth of the other
The trunking efficiency will VARY with the
bandwidth of equivalent Erlang that you choose!
Not suitable for use due to this property
2 Erlangs
of Low
Bandwidth
1 Erlang of
High
Bandwidth
Difference in
capacity
required for
same GoS
Section 10 New Service Optimisation
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.
Section 10 New Service Optimisation
Post Erlang-B
Combine the two traffic
sources together after
calculating required
capacity
The trunking efficiency
variation with magnitude
is not considered -
pessimistic about offered
traffic supported to the
same GoS
Not suitable for use due
to this property
1 Erlang of
Service A
1 Erlang of
Service B
+
1 Erlang and 1
Erlang of of
Service B
Illustration using 2 services of
same bandwidth
Difference
in capacity
required for
same GoS
Accepted correct method
Section 10 New Service Optimisation
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 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
Section 10 New Service Optimisation
Consider 2 services requiring equal resource:
Service 1: uses 1 trunk per connection. 12 Erlangs of traffic.
Service 2: uses 1 trunk 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.
Adding these together gives 31 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.
Post Erlang-B
Section 10 New Service Optimisation
Campbells Theorem
Campbells theorem creates a composite distribution where:







The amplitude (a
i
) 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

( )
c
a C
Capacity
i i

=
c
ffic OfferedTra
o
=

= =
i
i i i
i
i i i
b a
b a
c

o
v
2
C
i
= available capacity
o = mean
u = variance

i
= arrival rate
a
i
= amplitude of service
b
i
= mean holding time
c = capacity factor
i i
b = Traffic Offered Service
Section 10 New Service Optimisation
Campbells Theorem Example
Consider the same 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.

In this case the mean is:


The variance is:



Note: = traffic in Erlangs


= + = = = 30 ) 6 3 ( ) 12 1 ( Erlangs
i i i i
a a b o

= + = = = 66 ) 3 6 ( ) 1 12 ( Erlangs
2 2 2 2
i i i i
a a b v
Section 10 New Service Optimisation
i i
b
Campbells Theorem Example
Capacity Factor c (average trunks per connection):



Equivalent offered traffic:




Trunk capacity for equivalent traffic at 2% GoS = 21
2 . 2
30
66
= = =
o
v
c
63 . 13
2 . 2
30
Traffic Offered Equivalent = = =
c

Section 10 New Service Optimisation

Campbells Theorem Example
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 C
1
=(2.2 x 21) + 1 = 47
Target is Service 2, C
2
=(2.2 x 21) + 3 = 49


Different services will require a different capacity for the
same GoS depending oin the amplitude
Section 10 New Service Optimisation
Traffic Analysis Methods Compared
Equivalent Erlangs
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
Campbells theorem
Middle band (47 - 49)
Different capacities required for different services - realistic
Preferred solution for dimensioning, but not ideal...
Section 10 New Service Optimisation
Capacity Dimensioning with Campbells
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
Section 10 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Assuming we have n cells, we can determine the loading
per cell.
n n c
c
n n n
n n n
282
335 . 1
376 mean
traffic offered
335 . 1
376
502
mean
variance
502 2 63 250
variance
376 2 63 250
mean
2
=

= =
= = =
=

+ =
=

+ =
Section 10 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Considering the equation:

C
i
(available capacity) is predefined as 15. a
i
(amplitude)
depends on the service we use as our benchmark or priority
service. Choosing the voice service as the benchmark
service make a
i
equal to 1.



10.5 (rounded to 10) trunks will service 5.08 Erlangs.
c
a C
i i

= Capacity
( )
5 . 10
335 . 1
1 15
=

=
i
C
Section 10 New Service Optimisation
Capacity Dimensioning with Campbells
Theorem
Each cell requires 10 trunks to service 5.08 Erlangs.
Therefore:

Number of cells required = Equivalent Traffic
Traffic per Cell

=

Cell requirement is established at 56 cells.
Each of the cells will service:
4.46 Erlangs of voice
1.13 Erlangs of HSCSD.
5 . 55
08 . 5
282
=
Section 10 New Service Optimisation
Assessing Cell Loading using Campbells
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
Campbells 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.

Section 10 New Service Optimisation
Assessing Cell Loading using Campbells
Theorem
Using Campbells Theorem:









Hence 20 timeslots required.
( ) 20 1 36 . 1 14 : benchmark as voice Taking
required. trunks 14 B, Erlang From
Erlangs 09 . 8
36 . 1
11
traffic offered Equivalent
36 . 1
11
15
) ( Factor Capacity
15 ) 2 2 ( ) 1 7 ( variance
11 ) 2 2 ( ) 1 7 ( mean
2 2
= +
= =
= =
= + =
= + =
c
Section 10 New Service Optimisation
Evaluating Traffic Requirements
Real time non-controllable load
Spare capacity for which can
be allocated to non real time
applications
Peak traffic
Load
Time
Average circuit
switched traffic
Data may be packet switched, in which case it can be made to fill the
gaps in the demand for voice services.
Section 10 New Service Optimisation
Evaluating Traffic Requirements
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.
Section 10 New Service Optimisation
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.
Section 10 New Service Optimisation
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
Section 10 New Service Optimisation
GPRS Transmission Connections
SGSN
MS
GGS
N
BSS
G
b
G
n
U
m
PDP Context
Logical Link Control
Connection
Virtual Tunnel
Data Link Connection
IMSI/NSAPI
TLLI
DLCI
TID
CONNECTION TYPE
CONNECTION
IDENTIFIER
Section 10 New Service Optimisation
GPRS Cell Reselection
Section 10 New Service Optimisation
Cell Reselection per source cell
Cell Reselection Subscribers in READY state
Number of cell
reselections
Number of
Octets
Average Number of Octets Affected
Average Number of Octets Deleted
Average Number of Frames Deleted
Number of Cell Reselections
| 0000100002 | 0000100003 | CI
GPRS Service Exception
Section 10 New Service Optimisation
Number of frames deleted per TLLI
Number of Frames
Deleted
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
Section 10 New Service Optimisation
GPRS Average Response Times
Section 10 New Service Optimisation
Average response time after paging, per BVCI
Time in millsecs
BVCI
GPRS Attach/Detach Procedures
Section 10 New Service Optimisation
Number of Attach Reject Messages per Cell
Attach Rejects per Cell
Number of
Attach Rejects
GPRS Attach/Detach Procedures
Section 10 New Service Optimisation
Number of Attach Reject Messages per Subscriber (TLLI)
Number of
Attach Rejects
Attach Rejects per TLLI
GPRS Attach/Detach Procedures
Section 10 New Service Optimisation
Number of Attach Procedure Messages
Attach Request Attach Accept Attach Complete Attach Reject
Number of
Procedures
Attach Procedures
GPRS RA Update Procedures
Section 10 New Service Optimisation
Number of RA Update Messages
RA Updates
RA Request RA Accept RA Reject RA Request without response
Number
GPRS RA Update Procedures
Section 10 New Service Optimisation
Number of RA Update Rejects by Cause
Number of
RA Rejects
RA Rejects per Cause
Unknown Causes Network Failures
GPRS PDP Contexts
Section 10 New Service Optimisation
PDP Context Activation by MS
PDP Contexts Activated PDP Contexts Accepted PDP Contexts Rejected
Number PDP
Context
Activations
Number of PDP Context Activation Messages
GPRS Data Volumes and Rates
Section 10 New Service Optimisation
Downlink Data Rate of a Cell
DL LLC Rate
Bytes
Time (minutes)
GPRS Data Volumes and Rates
Section 10 New Service Optimisation
Downlink Data Rate of a Mobile Station
Time (minutes)
Volume (bytes)
DL LLC Global Size
Section 9

NSS Characteristics
and
Problem Types
Section 9
9.1 Introduction
?.4 Traffic Loading Evaluation
9.3 SS7 Signalling in the NSS
?.5 Parameter Evaluation
?.6 Database Usage
Section 9 NSS Characteristics and Problem Types
?.3 SS7 Link Load Evaluation
9.2 NSS Topology
Section 10

Network Subsystem (NSS)
Optimisation
Section 10
?.1 Introduction
?.4 Traffic Loading Evaluation
?.2 SS7 Signal Routing Analysis
?.5 Parameter Evaluation
?.6 Database Usage
Section ? NSS Optimisation
?.3 SS7 Link Load Evaluation
?.7 NSS Optimisation Process
NSS Optimisation Process
Section 10 NSS Optimisation
NSS Data Capture
Phase
Data Analysis Phase
Define Optimisation
Requirements
Define Implementation
Strategy
Optimisation Activity
Phase
NSS Data Capture
SS7 signal routing
SS7 link loading
User traffic routing
Trunk group configuration and loading
NSS configuration parameters
Database loading


Section 10 NSS Optimisation
MSC
SS7 Routing
Section ? NSS Optimisation
BSC
HLR
AuC
EIR
BSC
BSC
BSC
MSC
VLR
VLR
A
C
C
F
F
B
B
H
A
A
A
D
D
E
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
Section ? NSS Optimisation
Routing Analysis - Transit Layer Switching
Section ? NSS Optimisation
TSC
1a
HLR 2
300k
HLR 3
150k
HLR 4
1200k
HLR 5
600k
HLR 1
600k
TSC
3b
TSC
4b
TSC
3a
TSC
1b
TSC
4a
TSC
2b
TSC
2a
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
MSC
IN
GPRS
VMS
SMS
HLR/SMS/VMS/IN/
GPRS Layer
TSC Layer
MSC Layer
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
Section ? NSS Optimisation
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

Section ? NSS Optimisation
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

Section ? NSS Optimisation
Link Loading Signal Load Calculation
Signal Load (mE) =

(No. MSUs x 6) + (No. Payload Octets)
8 x length of time (secs)
Section ? NSS Optimisation
Link Loading - Signalling Load Sharing
Section ? NSS Optimisation
TSC1 C7 Load C7 Load
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
Link Loading Symmetrical Load Spreading
Section ? NSS Optimisation
Total MSU per sec
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
H
L
R
1

(
R
a
h
a
h
a
n
)
H
L
R
2

(
R
a
h
a
h
a
n
)
M
S
C
E
M
A
M
M
S
C
K
J
M
S
C
R
A
M
S
C
S
B

A
l
g
h
a
d
i
r

M
S
C
H
L
R
3

(
B
H
)
M
S
C
B
E
H
M
S
C
F
A
R
M
S
C
L
C
T
H
L
R
5

(
E
m
a
m
)
L
C
T
T
S
C
1
E
M
A
M
T
S
C
1
M
S
C
K
A
N
F
A
T
E
H
M
S
C
Exchange
M
S
U

p
e
r

s
e
c
Tx
Rx
Recommended Value per Node
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

Section ? NSS Optimisation
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
Section ? NSS Optimisation
Parameter Analysis VLR Parameters
Section ? NSS Optimisation
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
Section ? NSS Optimisation
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 Analysis TMSI Parameters
Section ? NSS Optimisation
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
Section ? NSS Optimisation
Advantages of Au usage:
Increases security of network
Prevents unauthorised network access
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 Authentication Parameters
Section ? NSS Optimisation
PARAMETER
REC
Triplet Reuse Y
MSRN Life Time (Sec) 20
Country Code Length 2
Location Update New Visitor Y
IMSI Attach Y
AUTHENTICATION Location Update N
COUNT Periodic Location Update N
Mobile Originating Call N
Mobile Originating SMS N
Mobile Terminating Call N
Mobile Terminating SMS N
Mobile Terminating USSD N
Mobile SS Operation N
Parameter Analysis IMEI Parameters
Section ? NSS Optimisation
Advantages of IMEI usage:
Increases security of network
Enables tracing/monitoring of Grey/Black-listed MSs
Prevents unauthorised network access
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
Section ? NSS Optimisation
PARAMETER REC
Location Update New Visitor Y
IMSI Attach Y
Location Update Y(10)
Periodic Location Update N
Mobile Originating Call N
IMEI CHECKING ON. Mobile Terminating Call N
Mobile Originating SMS N
Mobile Terminating SMS N
Mobile Terminating USSD N
Mobile SS Operation N
Black List Effect BLOCK
Grey List Effect TRACE
Unknown IMEI Effect BLOCK
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
Section ? NSS Optimisation
Database Usage - VLR
Section ? NSS Optimisation
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%
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 - VLR Cleaning
Section 9 NSS Characteristics and Evaluation
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
subscribers MS.
Section ? NSS Optimisation
NSS Databases VLR Utilisation
VLR utilisation refers to subscriber record
capacity and usage within a VLR
Utilisation problems include:
Over-capacity in certain VLRs
Uneven spread of records across the
network
Over-capacity can result in:
Call function failures (e.g. setup/handover,
authentication etc)
Lack of growth capability
Section 9 NSS Characteristics and Evaluation
Sr. No. MSC
VLR
Utilisation
1 MSC-1 65%
2 MSC-2 82%
3 MSC-3 26%
4 MSC-4 100%
5 MSC-5 86%
6 MSC-6 66%
7 MSC-7 71%
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
Utilisation problems include:
Over-capacity in certain HLRs
Uneven spread of records across
the network
Over-capacity can result in:
Call function failures (e.g.
setup/handover, authentication
etc)
Lack of growth capability
Signalling bottlenecks
Section 9 NSS Characteristics and Evaluation
0
10
20
30
40
50
60
70
80
90
100
HLR1 HLR2 HLR3 HLR4 HLR5
%AGE USAGE OF HLRs
%AGE USAGE
NSS Traffic Trunk Utilisation
Section 9 NSS Characteristics and Evaluation
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?

Section 10 New Service Optimisation