Você está na página 1de 180

The Cellular Academy

GSM Radio Network Planning and Optimisation


Coverage and Cell Structure Planning
Capacity and Frequency Planning
Network Optimisation
Advanced Topics

Version 4.0

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

INTRODUCTION

(C) Network Consultants / TNC Ltd 2005

Definitions
(Radio) Network optimisation
Fine tuning of existing network infrastructure
Improvement of the air interface performance in an
operational part of the network towards desired quality
targets without adding new hardware

Network extension
Extension of service into new areas or improvement of
service quality in existing network by means of adding
more network components

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Definitions
Minimise interference
Maximise capacity

Quality

Traffic

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

More Traffic results in more interference and a reduction in quality. One


solution could be to add more cells (blue curve), however this costly. More
likely to use optimisation techniques and / or advanced network solutions
such power control, DTX, frequency hopping to try to maintain quality levels
as traffic increases (red curve) .

(C) Network Consultants / TNC Ltd 2005

Introduction
Why optimise?
Systematic inaccuracies
Tool databases, prediction algorithms ...

Statistical processes involved


Radio propagation, traffic ...

Assumptions made
Human error
Planning errors, data entry, installation errors ...

Equipment faults
Hardware, software

Network growth

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

SYSTEM BASICS

(C) Network Consultants / TNC Ltd 2005

System basics
GSM Air Interface
Physical Layer Specifications
05.01: Physical layer on the radio path: general
description
05.02: Multiplexing and multiple access on the
radio path
05.03: Channel coding
05.04: Modulation
05.05: Radio transmission and reception
05.08: Radio subsystem link control
05.10: Radio subsystem synchronisation

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

The GSM standards are available from the Internet. If they were printed out
they would use and enormous a mount of paper. Many are to do with fixed
networking and with detailed protocols. The ones mainly relevant to the
radio interface are the 05 and 08 series. It is thoroughly recommended that
radio planners and optimisation engineers read the 05 and 08 series
specifications entirely.

(C) Network Consultants / TNC Ltd 2005

System basics
GSM Air Interface
The ARFCN:
Absolute Radio Frequency Channel Number
GSM 900

F-DL(n) = 890 + 0.2*n

E-GSM 900 F-DL(n) = 890 + 0.2*n


GSM 1800

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

1 n 124

F-UL(n) = F-DL(n) + 45

0 n 124

F-UL(n) = F-DL(n) + 45

F-DL(n) = 890 + 0.2*(n - 1024)

975 n 1023

F-DL(n) = 1710.2 + 0.2*(n - 512)

512 n 885

Network Consultants / TNC Ltd 2005

F-UL(n) = F-DL(n) + 95

System basics
GSM Air Interface
Channel types
Defined in GSM 05.02
Physical channels
Frequency and timeslot
HSN and MA and MAIO

Logical channels
Speech (e.g. TCH/FS, TCH/HS)
Data (e.g. TCH/F9.6)
Control (e.g. BCCH, SDCCH, SACCH etc)

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

The repetition of one particular timeslot [0,1,2,3,4,5,6, or 7] forms a physical


channel. For example the repetition of timeslot(0) forms a physical channel,
which in the case of the BCCH carrier is reserved for Control Channel
information.
Each RF carrier has 8 physical channels.
A logical channel is a subset of a physical channel, it is a cycle of date
related to a specific function. The number of timeslots used and their
periodicity is dependent on the type of data mapped onto the logical channel.

(C) Network Consultants / TNC Ltd 2005

System basics
GSM Air Interface
Frequencies, channel allocation
CA: the list of frequencies allocated to a cell
MA: the list of frequencies allocated to a mobile
When frequency hopping

MAI: indexes the list of mobile frequencies


For each TDMA frame

MAIO: offset to the MA Index to avoid two TRX in


one cell using the same frequency
HSN: Hopping sequence number

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

10

10

System basics
GSM Air Interface
GSM Terms / Specifications
Channel bandwidth = 200 KHz
Data rate = 270.8 Kbits per seconds
Bit period = 1/data rate = 1 / 270.8 Kbps = 3.69 s
Timeslot = 156.25 bits = 156.25x3.69s = 0.577ms
Burst: the bits transmitted during a timeslot, varies
dependent on data type
Frame: the consecutive 8 timeslots labelled [07]
Frame duration = 8 TSs = 8 x 0.577ms = 4.615 ms
Physical channel: the repetition of one particular
timeslot [0,1,2,3,4,5,6,7]
Optimisation v 4.0

Network Consultants / TNC Ltd 2005

11

Each RF carrier on BS may transmit a burst of data during a timeslot period


every 0.577 ms. The number of bits in a burst will depend on the type of data
transmitted, however they must be less than the limit of 156.25 bits.
A timeslot has period of 156.25 bits = 156.25 x 3.7 s = 0.577 ms
There are 5 different kind of bursts
Normal burst
Synchronisation burst
Frequency correction burst
Access burst
Dummy burst

(C) Network Consultants / TNC Ltd 2005

11

System basics
GSM Air Interface
Physical channels - bursts
Normal Bursts

3 tail bits
58 encrypted data bits
26 training sequence bits
58 encrypted data bits
3 tail bits
8.25 guard bits
58

26

58

3 8

Normal Burst
Optimisation v 4.0

Network Consultants / TNC Ltd 2005

12

The Normal Burst is used to carry information on the Traffic Channel (TCH)
and on some of the Control Channels (excluding RACH, SCH, FCCH).
The burst is divided into 156.25 bits. The encrypted bits are 57 bits of data /
speech plus one bit stealing flag indicating whether the burst was stolen for
FACCH signalling (for handover commands) or not.
The training sequence is a known 26 bit pattern used by the equalizer to
create a channel model. The tail bits are used by the equalizer algorithm to
detect the start and stop points of the burst. The guard period of 8.25 bits is
used to protect against overlapping messages at the BS.

(C) Network Consultants / TNC Ltd 2005

12

System basics
GSM Air Interface
Physical channels - bursts
Frequency correction burst (FB)
Synchronisation burst (SB)
Dummy burst (DB)

39

64

39

Synchronisation Burst

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

13

The Frequency correction burst is used for frequency synchronisation of the


mobile. The FCCH burst appears to the mobile as an unmodulated carrier
and is used by the mobile to identify the BCCH carrier.
Dummy burst is sent from the BTS on some occasions and carries no
information. It is the same format as a Normal burst.
A Synchronisation (SCH) burst is used for time synchronisation of the
mobile. The SCH burst consists of a 64 bit extended training sequence used
for synchronisation. The remain 78 encrypted bits contain Base Station
Identity Code (BSIC), frame number and time data for the mobile station
(MS).

(C) Network Consultants / TNC Ltd 2005

13

System basics
GSM Air Interface
Logical Channels
Traffic CHannels (TCH)
Speech channels
Full rate traffic channel for speech (TCH/FS)
Half rate traffic channel for speech (TCH/HS)

Data channels

Optimisation v 4.0

Full rate traffic for 9.6 kbit/s data (TCH/F9.6)


Full rate traffic for 4.8 kbit/s data (TCH/F4.8)
Full rate traffic for 2.4 kbit/s data (TCH/F2.4)
Half rate traffic for 4.8 kbit/s data (TCH/H4.8)
Half rate traffic for 2.4 kbit/s data (TCH/H2.4)

Network Consultants / TNC Ltd 2005

14

A large variety of information must be transmitted between the BS and MS.


The information consists of user data and control signalling. The type of
information transmitted dictates the logical channel used. Logical channel
are mapped onto physical channels, for example, digital speech is sent on a
logical channel named TCH. During the transmission the TCH can be
allocated a physical channel, say timeslot (6).
TCHs can carry either encoded speech or user data in both uplink and
downlink directions in a point to point link.
A full rate TCH carries information (speech or data) at gross rate of 22.8
Kbps. The raw data rate for each TCH is 13 Kbps for speech, and 12 Kbps,
6Kbps and 3.6 Kbps for data, however, the actual user data rates are 9.6
Kbps, 4.8 Kbps, and 2.4 Kbps.
A half rate TCH carries information at half of the full rate at a gross rate of
11.4 Kbps.
The actual data rates used for full rate / half rate speech and data are listed
in the slide above.

(C) Network Consultants / TNC Ltd 2005

14

System basics
GSM Air Interface
Logical Channels
Control channels
Broadcast Channels (BCH)
Common control channels (CCCH)
Dedicated Control Channels (DCCH)

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

15

Controls channels are used for signalling and synchronisation information.


They can be unidirectional (Uplink or Downlink only) or bi-directional (Uplink
and Downlink).
The Broadcast Control Channels provide general information about a
particular cell on a per BTS basis.
The Common Control Channels are used to carry signalling information
necessary for call origination for the BS to establish a connection to the MS
and / or for the MS to establish a connection to the BS.
Dedicated Control Channels are used for systems signalling in Idle mode,
and for signalling information during call set up before a TCH is allocated.

(C) Network Consultants / TNC Ltd 2005

15

System basics
GSM Air Interface
Logical Channels: Control channels
Broadcast CHannels (BCH)
Frequency correction channel (FCCH)
Synchronisation channel (SCH)
Broadcast control channel (BCCH)

Common Control CHannels (CCCH)

Optimisation v 4.0

Paging channel (PCH)


Random access channel (RACH)
Access grant channel (AGCH)
Cell Broadcast Channel (CBCH)

Network Consultants / TNC Ltd 2005

16

Broadcast Channels
FCCH provides MS information to synchronise to the BCCH carrier
frequency
SCH provide MS information of BSIC, frame number and timing data
BCCH provide general information about the cell the MS is camped on
Common Control Channels
PCH used by the BTS to page MS in the area
RACH used by MS to request a channel from the BS
AGCH used by the BS to assign a channel to the MS
CBCH use

(C) Network Consultants / TNC Ltd 2005

16

System basics
GSM Air Interface
Control channels
Dedicated Control CHannels

Optimisation v 4.0

Slow associated control channel (SACCH/TF)


Fast associated control channel (FACCH/TF)
Stand alone dedicated control channel (SDCCH/8)
Slow associated control channel (SACCH/C8)
Stand alone dedicated control channel, combined with
CCCH (SDCCH/4)

Network Consultants / TNC Ltd 2005

17

Slow Associated Control Channel (SACCH) is a sub channel of a TCH or an


SDCCH. The SACCH is a continuous data channel used for non urgent
procedures, providing measurement reports (every 480 ms) including signal
strength measurements. The channel is used in both uplink and downlink
directions and is necessary for mobile assisted handover decisions.
The SACCH is used with the SDCCH for authentication, set up information,
measurement reports, and power control information before a TCH is
assigned.
Handover is commanded using the Fast Associated Control Channel
(FACCH), which replaces a TCH. The receiver detects a FACCH by reading
binary information on the TCH called a stealing flag. Hence, handovers steal
time from a TCH and reduce effective data rate to the user. Handovers
should therefore be kept to a minimum.

(C) Network Consultants / TNC Ltd 2005

17

System basics
GSM Air Interface
Mapping Logical to Physical channels
Physical channel has several logical channels
TCH/F + FACCH/F + SACCH/TF
TCH/H+ FACCH/H+ SACCH/TH
FCCH + SCH + BCCH + CCCH + SDCCH/4 + SACCH/C4
Combined

FCCH + SCH + BCCH + CCCH


SDCCH/8 + SACCH/C8
Non-combined

where CCCH = PCH + RACH + AGCH

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

18

18

SCH

BCCH

BCCH

BCCH

BCCH

BCCH

BCCH

BCCH

BCCH

CCCH

CCCH

CCCH

CCCH

CCCH

CCCH

CCCH

CCCH

FCCCH

FCCCH

SCH

CCCH
CCCH
CCCH

Cycle repeated after 102 TDMA frames

Network Consultants / TNC Ltd 2005

CCCH
FCCH
SCH

SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
FCCH
SCH

SDCCH/4(2)
SDCCH/4(2)
SDCCH/4(2)
SDCCH/4(2)

BCCH

BCCH

CCCHBCCH
CCCHBCCH
BCCH

CCCH

CCCHCCCH

CCCH

CCCH

CCCH
CCCH
FCCCH
SCH
CCCH

CCCHCCCH
CCCH

CCCH

FCCCH

CCCHSCH

CCCH

CCCH

CCCHCCCH

CCCH

CCCH

CCCH
CCCH
CCCH

CCCHCCCH
FCCHCCCH
CCCH

CCCH

SCHCCCH

CCCH

CCCH

FCCH
SCH
SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
FCCH
SCH
SDCCH/4(2)
SDCCH/4(2)
SDCCH/4(2)

SDCCH/4(0)
FCCH
SCH
SDCCH/4(0)

SDCCH/4(0)

SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(0)

SDCCH/4(0)
SDCCH/4(0)
SDCCH/4(1)

SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)
SDCCH/4(1)

SDCCH/4(1)
FCCH
SCH

SDCCH/4(1)
SDCCH/4(2)
SDCCH/4(2)
FCCH
SDCCH/4(2)

SDCCH/4(2)

SDCCH/4(2)
SCH

SDCCH/4(3)

SDCCH/4(3)

SDCCH/4(3)
SDCCH/4(3)
SDCCH/4(3)
FCCH
SCH

SDCCH/4(2)
SDCCH/4(3)
SDCCH/4(3)
SDCCH/4(2)
SDCCH/4(3)

FCCH
SDCCH/4(2)
SCH

SACCH/C4/(2)

SDCCH/4(2)
SACCH/C4/(0)

SDCCH/4(3)

SACCH/C4/(2)

SACCH/C4/(0)

SACCH/C4/(2)

SDCCH/4(3)
SACCH/C4/(0)

SDCCH/4(3)

SACCH/C4/(2)

SACCH/C4/(0)
SDCCH/4(3)

SDCCH/4(3)

SACCH/C4/(3)

SACCH/C4/(3)

SDCCH/4(3)

SACCH/C4/(3)

FCCH

IDLE

SACCH/C4/(3)

SACCH/C4/(1)

SDCCH/4(3)
SACCH/C4/(1)
SACCH/C4/(1)

SDCCH/4(3)
SACCH/C4/(1)
FCCHIDLE
SCH
SACCH/C4/(0)

SACCH/C4/(2)

SACCH/C4/(0)

SACCH/C4/(2)

SACCH/C4/(0)

SACCH/C4/(2)

SACCH/C4/(0)

SACCH/C4/(3)

SACCH/C4/(1)

SACCH/C4/(3)

SACCH/C4/(1)

SACCH/C4/(3)

SACCH/C4/(1)

SACCH/C4/(3)

SACCH/C4/(1)

IDLE

IDLE

19

SCH
SACCH/C4/(2)

System basics

CCCH

BCCH
BCCH

FCCH

SCH SCH

FCCH + SCH + BCCH + CCCH + SDCCH/4 +


SACCH/C4

CCCH

SCH

GSM Air Interface

CCCH

FCCH
BCCH

Mapping Logical to Physical channels

Optimisation v 4.0

SCH

CCCH

Cycle repeated after 102 TDMA frames

The logical control channel information is mapped on to the physical channel


of timeslot (0) of the BCCH carrier. The information transmitted follows a
defined pattern known to the MS and is repeated every 102 TDMA frames.

(C) Network Consultants / TNC Ltd 2005

FCCH

FCCH

19

System basics
Measurements
Idle mode
Rx Lev_DL (server)
Rx Lev_DL (neighbour 1)
Rx Lev_DL (neighbour 2)
Rx Lev_DL (neighbour 3)
Rx Lev_DL (neighbour 4)
Rx Lev_DL (neighbour 5)
Rx Lev_DL (neighbour 6)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

20

20

System basics
Measurements
RxLev
0
1
2
:
:
61
62
63

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Signal
< -110dBm
-110 to -109dBm
-109 to -108dBm
:
:
- 49 to - 48dBm
- 48 to - 47dBm
> - 47 dBm

Network Consultants / TNC Ltd 2005

21

21

System basics
Measurements
RxLev SUB and RxQual SUB
FULL measurements based on all TDMA frames
With DTX not all frames transmit
SUB measurements based on 12 frames which
always transmit

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

22

22

System basics
Radio link measurements
Basis for all handover
decisions
Neighbour

k
n lin
Dow
ink
Upl

MS measures
(Downlink):
MS
Signal level
Quality (RXQUAL)
Signal levels of
neighbouring cells (BCCH)
- strongest 6 are reported to BSC
Optimisation v 4.0

BS measures
(Uplink):
Signal level
Quality (RXQUAL)
BS-MS distance
(Interference levels in
idle time slots)

BS

BSC
Collects all data
- BS and MS send
BSC
measurement
reports every 480 ms
Makes handover decisions

Network Consultants / TNC Ltd 2005

23

The radio link measurements are performed on the Slow Associated Control
Channels (SACCH). The SACCH is a sub channel of the TCH or SDCCH
and is used by the BS and MS to send measurement date every 480ms.
Every 480 ms the MS takes 100 samples of signal strength and quality
measurements. The BS sends lists containing signal strength and BCCHs of
neighbour cells.
MS collects (100 / # of neighbours) samples of signal strength of neighbour
cells.

(C) Network Consultants / TNC Ltd 2005

23

System basics
Measurements
Dedicated mode
RxLev_DL (Server)
RxLev_UL (Server)
RxLev_DL (All neighbours - best 6 reported)
RxQual_DL (Server)
RxQual_UL (Server)
Timing Advance

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

24

24

System basics
Measurements
Timing advance
BS Tx

= range / c

2
MS Rx

MS Tx

1
3 TS

2
BS Rx

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

25

MS transmits a message to the BTS exactly 3TS after it detects a received


message from the BTS. The uplink scheme is shifted by 3 timeslots from the
downlink scheme.
As the MS moves, propagation delay changes. The BTS monitors the delay
between the start of a received frame (immediate) and the time the MS
message actually arrives. The BS sends the measured delay as a Timing
Advance (TA) value on the next SACCH or AGCH.
TA updates occur every SACCH multi-frame.

(C) Network Consultants / TNC Ltd 2005

25

System basics
Idle mode procedures
Network selection
At power on, MS first tries to select last
registered PLMN (Public Land Mobile Network)

If no previous registered PLMN, selection is


made in
Automatic mode: uses PLMN list in order of
priority
Manual mode: selection is made by user based
on list made available by MS

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

26

26

System basics
Idle mode procedures
Location area updates

LAC 1

Incoming calls only


page mobile within its
present Location Area.
Mobile updates VLR when
LAC changes
Also when timer dictates

Optimisation v 4.0

LAC 2

Network Consultants / TNC Ltd 2005

27

Cell broadcasts Cell Global Identifier (CGI) on BCCH carrier. The MS reads
the CGI from the cell it camps on.
CGI = MCC + MNC + LAC + CI
MCC = Mobile Country Code
MNC = Mobile Network Code
LAC = Location Area Code
CI = Cell Identifier, within location area
Location Area Identifier (LAI) is a subset of CGI= MCC+MNC+LAC
The MSC controls all cells within one or more LAIs. MS reports to MSC
when it camps on new serving cell with a different LAI. This know as location
updating.

(C) Network Consultants / TNC Ltd 2005

27

System basics
Idle mode procedures

Major road along


boundary?

Location Area Codes

LAC 1

Small location areas result


in more location updates
(SDCCH congestion)
Topographical features must be
taken into account to minimise
location update traffic on SDCCH

LAC 2

LAC boundary
along river?

Large location areas result in


unnecessary paging traffic (PCH capacity)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

28

28

System basics
Idle mode procedures
Monitoring
MS monitors BCCH channel
Contains:
Frequencies of BCCHs of neighbour cells
Base Station Identity Code (BSIC)
Location Area Code

PCH (Paging Channel) is multiplexed on the same


physical channel

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

29

1)

MS scans all (or those specified in SIM card) GSM channels and
calculates average RxLex.

2)

MS tunes to frequency with highest RxLev and listens for FCCH.

3)

MS tunes to FCCH and identifies BCCH carrier frequency,


decodes SCH to get BSIC information and to synchronise time
and frame to serving BS

4)

MS reads the network identity from the BCCH carrier to identify


the serving Public Land Mobile Network (PLMN).

5)

MS will measure RxLev of camped on BS, as well as Neighbour


RxLev (However, measurement is infrequent, every few seconds).

(C) Network Consultants / TNC Ltd 2005

29

System basics
Channels for Call Setup
Mobile originated
BCCH
RACH

AGCH
SDCCH

TCH

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Mobile monitors BCCH


Request for access made on the
Random Access CHannel
Access granted and details of
dedicated channel transmitted
on the Access Grant CHannel
Call set-up, encryption details, etc
exchanged on the Stand alone
Dedicated Control CHannel
Data or speech exchanged on
a Traffic Channel

Network Consultants / TNC Ltd 2005

30

30

System basics
Channels for Call Setup
Mobile terminated
BCCH
PCH
RACH
AGCH
SDCCH

TCH

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Mobile monitors BCCH


Page received on Paging CHannel
Access requested
Access granted
Call set-up, encryption details, etc
exchanged on the Stand alone
Dedicated Control CHannel
Data or speech exchanged on
a Traffic CHannel

Network Consultants / TNC Ltd 2005

31

31

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

32

BSS PARAMETERS

(C) Network Consultants / TNC Ltd 2005

32

BSS Parameters
Idle mode procedures
Cell selection / reselection
Cell selection parameter C1
Enables MS to camp on cell with which high
probability of communications is possible
Based on signal strength only
C1 calculated for several cells

Mobile selects cell / network with highest C1


C1=(Received Signal Level- RxLev_Access_Min) MAX[(Ms_TxPwr_Max_CCH -P),0]
Only cells with positive C1 are considered

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

33

P = Maximum (Hardware limit) transmit power of MS


Ms_TxPwr_Max_CCH = Maximum transmit power allowed to be used by the
mobile to access that BTS

(C) Network Consultants / TNC Ltd 2005

33

BSS Parameters
Idle mode parameters
Cell selection
RxLev_Access_Min
Minimum RxLev at which mobile is allowed to camp
on a cell
Higher means more reliable service once camped
Lower means apparently more coverage
Typically -108 dBm to -100 dBm

C1
Difference between RxLev_DL and RxLev_Access_Min
Mobile camps on Cell with highest C1

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

34

34

BSS Parameters
Idle mode parameters
C1 criterion for cell access:
C1 = A - MAX ( B , 0 ) > 0
RxLev_Access_Min = -100 dBm

Example:
RXLEV_DL = -97 dBm

Ms_TxPwr_Max_CCH = 39 dBm
BCCH

P = 33 dBm (1 W)
A = RXLEV_DL - RxLev_Access_Min
B = Ms_TxPwr_Max_CCH - P
P = Max. (hardware) transmit power of the MS

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

35

35

BSS Parameters
Idle mode parameters
C2 criterion for cell reselection
C2 = C1 + Offset + Temp_offset * H
H=1 (T<Timer)
H=0 (T>Timer)
T = time on measurement list

C1
C2

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

36

C2 varies amount of cell reselection criteria C1, based on an offset value


which may be constant or temporary (based on timer).
In example on left, the indoor cell has a constant negative offset to prevent
people outside building camping on cell.
On example on right, cars on the main road pass by micro cell in less time
than temporary offset timer. If they pull into the service station and are within
range after timer expires, they camp on micro-cell.

(C) Network Consultants / TNC Ltd 2005

36

BSS Parameters
Idle mode parameters
LAC updating
MS updates VLR using SDCCH channel if it detects
change in LAC after cell reselection
Cell_Reselect_Hysteresis = 6 dB
(to avoid ping-pong)

Periodic update every 0.1hrs to 25.5hrs


If no update received, MS detached from network
Assumed to have lost coverage or battery is flat
Otherwise would have detached itself
when switched off

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

37

Avoid situation where Detach timer value is less than periodic update value!!!

(C) Network Consultants / TNC Ltd 2005

37

BSS Parameters
Idle mode parameters
Max. number of retries on the Access Channel
(RACH)
Burst collisions are possible (Slotted Aloha Effect)

)
H (1
RAC
)
H (2
RAC

RA
CH

H
AGC

Default: max_Access_Retry = 2

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

38

If two access request messages arrive during the same timeslot. BTS may
not detect either.
Max_Access_Retry [1,2,3,4,7] defines the maximum number of time the
MS repeats and access attempt before selecting another cell.

(C) Network Consultants / TNC Ltd 2005

38

BSS Parameters
Idle mode parameters: BSICs
Base Station Identity Codes
Used by the MS to
distinguish between
neighbouring cells
on the same frequency
Co-channel cells
must have
different BSICs in
the same locality

Optimisation v 4.0

f9

f9

f7
f1

f11

f1

f5
f11

f9

f5
f3

f13

f3
f13

Server
Neighbours
Other cells

f7

f7
f1

f11

f5
f3

f13

Network Consultants / TNC Ltd 2005

39

The BSIC is transmitted on the SCH on timeslot(0) of the BCCH carrier.


BSIC = [NCC] + [BCC]
NCC= Network Colour Code [0-7]
BCC= Base Station Colour Code [0-7]
Careful use of NCC values required in border areas between different
countries to avoid confusion. Use pre-assigned NCC values.

(C) Network Consultants / TNC Ltd 2005

39

BSS Parameters
Radio link measurement averaging
5
50

51

52

53

54

55

56

57

58

59

Averaged level
60

-5

-10

RxLev Samples

-15

-20

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

40

40

BSS Parameters
Radio link measurement averaging
BSC receives measurement samples from
BS + MS
Every SACCH-Multiframe (480 ms, 104 TDMAFrames for TCH)

Types of average
Sliding Window averaging:
Block averaging
Recursive filtering

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

41

41

BSS Parameters
Radio link measurement averaging
Sliding window averaging
1 to 32 measurements included
More measurements: more delay
32 measurements takes 15.4 seconds
0

1 2 3 4 5 6 7 . . . 30 31 32

33

One averaged value

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

42

42

BSS Parameters
Radio link measurement averaging
Example with window_length = 3
Raw measurements

Averaged results

Delay is about half the window length


Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

43

43

BSS Parameters
Radio link measurement averaging
Different window lengths for
Uplink
Downlink
Power control
Handover (different types)
Bit Error Rate (RxQual)
Signal (RxLev)
Timing advance

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

44

44

BSS Parameters
Level-triggered Power Control (PC)
Transmit minimum power necessary
Reduces interference
Prolongs MS battery life
Protects receiver from RF blocking when very
close

Used at MS and at BS
But not on downlink BCCH carrier
(Constant power since used for mobile
measurements)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

45

45

BSS Parameters
Level-triggered Power Control
Aim to keep received signal between limits
PC_Upper_Threshold_Lev
PC_Lower_Threshold_Lev

PC_Lower_Threshold_Lev is some dB
above the minimum sensitivity threshold

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

46

There should be a margin allowed between the pc_lower_threshold_Lev


and the receiver minimum sensitivity threshold. This margin is necessary to
prevent dropped calls when user make a sudden change such as moving
from outdoor to indoor environment.
A typical value for this margin is about 15dB.

(C) Network Consultants / TNC Ltd 2005

46

BSS Parameters
Level-triggered Power Control
PC_Upper_Threshold_Lev ( _UL or _DL)
Decrease power if signal goes above this
threshold
PC_Interval

PC_Upper_Threshold_Lev
Averaged_RxLev
Pwr_Red_Step
Receiver sensitivity

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

47

47

BSS Parameters
Level-triggered Power Control
PC_Lower_Threshold_Lev ( _UL or _DL)
Increase power if signal drops below this
threshold
PC_Interval

Averaged_RxLev
PC_Lower_Threshold_Lev
Pwr_Inc_Step

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Receiver sensitivity

48

48

BSS Parameters
Level-triggered Power Control
The difference between the two thresholds is
a hysterysis
PC_Interval

PC_Upper_Threshold_Lev

MS/BS_Tx_Pwr

PC_Lower_Threshold_Lev
30dBm
26dBm
22dBm

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

49

49

BSS Parameters
Level-triggered Power Control
Common default settings:
PC_Lower_Threshold_Lev_DL = 87 dBm
Lower Threshold 15 dB above mobile sensitivity
102 dBm + 15 dB = -87 dBm
Allows for sudden signal loss due to in-building

PC_Upper_Threshold_Lev_DL = 77 dBm
Difference large enough (10 dB) to avoid rapid
consecutive power increases and decreases

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

50

50

BSS Parameters
Level-triggered Power Control
Common default settings:
PC_Lower_Threshold_Lev_UL = 92 dBm
Lower Threshold 15 dB above base station sensitivity
(including diversity gain)
104 dBm 3 dB + 15 dB = 92 dBm

PC_Upper_Threshold_Lev_UL = 82 dBm
10 dB above lower threshold

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

51

51

BSS Parameters
Level-triggered Power Control
Common default settings:
Pwr_red_step = 2 dB
Pwr_inc_step = 4 dB
Power increase is more urgent

PC_Interval = 2 SACCH multi-frames = 960 ms


Reasonably fast response
Already delayed by measurement averaging

PC_Averaging_Lev_DL/UL_Window_Size
= 6 SACCH frames (2.9 s)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

52

52

BSS Parameters
Level-triggered Power Control: Exercise!
Current BS Power = ?
Path loss

(including BS + MS
antennas and cables)

= 124 dB

PC_Upper_Threshold_Lev_DL = - 77 dBm
PC_Lower_Threshold_Lev_DL = - 87 dBm
BS_TxPwr_Max = 43 dBm
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

53

53

BSS Parameters
Quality-triggered Power Control
Reduce power if quality is good
Further reduction in output power under good
conditions to reduce interference

Increase power if quality is bad


Interference might affect quality irrespective of
signal level

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

54

54

BSS Parameters
Quality-triggered Power Control
Common default settings
PC_Lower_Threshhold_Qual_DL/UL = 4 (RXQUAL)
Triggers a power increase at poor quality

PC_Upper_Threshhold_Qual_DL/UL = 0 to 2
Triggers a power reduction at good quality
Arguably Level criterion is more suitable for power
reduction
High load networks trade a bit of speech quality for
more capacity

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

55

55

BSS Parameters
Quality-triggered Power Control
Common defaults for averaging parameters
The averaging interval for Quality (PC) is shorter
than the one for Level (PC)
Quality can change more suddenly

PC_Averaging_Qual_UL/DL_Windowsize
= 4 (=1.9s)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

56

56

BSS Parameters
Handover
Optimising handover performance:
maximise time spent on the best cell
maintain good
speech quality
minimise number
of handovers
minimise dropped
calls
generate as little
interference as possible
distribute traffic
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

57

57

BSS Parameters
Handover
Parameters can be adjusted to move cell
boundaries and overlap to
Improve success rate for critical handovers
Eliminate local interference at the cell edge
Share traffic load between cells

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

58

58

BSS Parameters
Handover
HO Parameter tuning is a second order solution.
First check:
frequency plan
antenna tilts
neighbour lists
traffic capacity

There is interaction with the systems


Measurement Averaging parameters
Power Control parameters

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

59

59

BSS Parameters
Handover
Handover and Power Control
types and priorities:
Level Power Control
Quality Power Control
Power Budget Handover
Level Handover
Quality Handover
Intracell Handover
Distance Handover
Umbrella Handover
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

60

60

BSS Parameters
Power Budget Handover
Other cell offering a smaller path loss
takes account of fixed and dynamic power control

BS_TxPwr = 37 dBm
(Traffic channel)

BS_TxPwr = 43 dBm
(BCCH channel)

124 dB

Serving cell

126 dB

RxLev_DL
= -87dBm

RxLev_N1
= -83dBm

Neighbour cell

The best neighbour offers 4dB better signal, but the


path loss is 2 dB worse, therefore no handover.
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

61

61

BSS Parameters
Power Budget Handover (PowerBudGeT)
Path loss difference PBGT(n) calculated in the
BSC for each of the 6 neighbours
= [BS_TXPWR - RXLEV_DL] - [bsTxPwrMax(n) - RXLEV(n)]

BS_TXPWR: Actual serving cell transmit power

Neighbour with largest value of PBGT(n) and


PBGT(n) > Handover Margin is first in line for
handover

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

62

Power Budget Hanover is used to execute a handover to a cell that has a


lower path loss than current serving cell. The path loss of cell n must be
lower by margin PBGT(n) for handover to occur. Value of PBGT(n) must not
be too low to avoid ping-pong (multiple handovers)

(C) Network Consultants / TNC Ltd 2005

62

BSS Parameters
Power Budget Handover
Common default setting:
HO_Margin_PBGT _N = 4 to 6 dB
Large enough to protect against ping-pong
Small enough to ensure HO before signal disappears

In a well designed network, the majority of


handovers are of the PBGT type (90%)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

63

63

BSS Parameters
Power Budget Handover : Exercise!
PL = K + 35.7 log d

4dB or 6dB

HO_Margin= 4dB or 6dB

? km
1km
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

1km
Network Consultants / TNC Ltd 2005

64

64

BSS Parameters
Power Budget Handover
Example: Moving cell boundary to avoid
local interference
BS1: f1
Se

r
rve

BS2: f3

BS3: f3
Interference

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

65

In this example BS2:f3 is interfering with BS2:f3. The interference is


occurring in the handover overlap region between BS2 and BS1.

(C) Network Consultants / TNC Ltd 2005

65

BSS Parameters
Power Budget Handover
Example: Moving cell boundary to avoid local
interference

Interference
4dB

-2dB 4dB

8dB

Only possible with sufficient overlap between cells

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

66

66

BSS Parameters
Power Budget Handover
Example: Moving cell boundary to avoid
local interference
Default setting:
HO_Margin_PBGT_1 = ho_Margin_PBGT_2 = 4 dB
Handover zone around Equal Path Loss Boundary

New setting:
hoMarginPBGT_1 = -2 dB (HO from BS2 to BS1)
hoMarginPBGT_2 = 8 dB (HO from BS1 to BS2)

Note that the handover zone is now only 6 dB


wide instead of 8 dB before
Slightly increased risk of ping-pong handovers

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

67

The interference can be avoided by moving the handover boundary of BS2


inwards, effectively reducing the service area of BS2 and expanding the
service area of BS1.

(C) Network Consultants / TNC Ltd 2005

67

BSS Parameters
Power Budget Handover
Common defaults for averaging parameters
The averaging interval for Power Control is shorter than
the one for Handover
Increases the chance that PC is executed before HO
Discourages ping-pong

Ho_AveragingLev_DL/UL_windowSize = 8
averagingWindowSizeAdjCell = 8
Ho_ThreshholdLev_DL/UL_nx = 4
Ho_ThreshholdLev_DL/UL_px = 3

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

68

68

BSS Parameters
Level Handover
Urgent handover on low signal
Switch from PBGT to Level Mode when signal less
than HO_Threshold_Lev.
In Level Mode, new handover margin parameter:
HO_Margin_LEV_N applies

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

69

If serving cell signal is less than HO_Threshold_LEV and a neighbour cell


has a signal that is stronger by value of HO_Margin_LEV_N, then execute a
level handover to that neighbour cell.

(C) Network Consultants / TNC Ltd 2005

69

BSS Parameters
Level Handover
Since the signal is less than
HO_Threshold_LEV, the
smaller HO_Margin_LEV
can be used instead

The PBGT margin is too large,


and the serving signal falls
below the sensitivity before a
handover can take place

HO_Threshold_Lev
HO_Margin_PBGT_N
Receiver sensitivity
HO_Margin_LEV_N

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

70

70

BSS Parameters
Level handover
Common default settings
Level Threshold: 5 dB above effective Rx sensitivity
HO_Threshold_LEV_DL = -97 dBm (Downlink)
HO_Threshold_LEV_UL = -102 dBm (Uplink)

Hysterysis: 2 dB less than for PBGT HO


HO_Margin_Level = 2 to 4 dB

Effective BS receiver sensitivity (Uplink):


Lower than MS sensitivity
+ Diversity gain + Pre-amplifier

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

71

71

BSS Parameters
Example: Critical handover on hilltop

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

72

72

BSS Parameters
Example: Critical handover on hilltop
Signal of candidate
neighbour not sufficient
until summit is reached

Signal rises slightly


as mobile reaches
top of hill
Signal drops rapidly
as mobile passes
over summit
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

73

73

BSS Parameters
Example: Critical handover on hilltop
Delays due to
Measurement averaging
HO_Interval
HO_Margin

mean that serving signal


can be too low before HO
can be completed

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Delay

74

74

BSS Parameters
Example: Critical handover on hilltop
Counter measures:
Shorter averaging window size
Increase HO_Threshold_Lev
Smaller HO_Margin
Faster power control

Additionally make HOs


more reliable:
Shorter neighbour lists
Sufficient capacity in target cell

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Delay

75

75

BSS Parameters
Quality triggered handover
Handover to cell with worse signal because
of interference
Quality handover threshold
HO_Threshhold_Qual_DL/UL = 5
Triggers the handover request
One higher than Power Control threshold

Quality handover margin, hoMarginQual


HO_Margin_Qual_N = -2 dB
Enables HO to a cell with a lower RxLev

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

76

76

BSS Parameters
Quality triggered handover
PC_Threshold_Qual = 4
HO_Threshold_Qual = 5
Increase power before attempting HO!
When Frequency hopping, RxQual = 5 is OK
Adjust threshold to 6 for hopping network
(On hopping frequencies)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

77

77

BSS Parameters
Intracell (Interference) handover
Change of frequency and/or timeslot within the cell
If no other target cell is available, an intracell handover
can improve an interference situation
Interferer, f1 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 ...
Server, f1 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 ...
Current interference situation may be different on a different
frequency or different timeslot,

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

unless ... ?

Network Consultants / TNC Ltd 2005

78

78

BSS Parameters
Intracell handover
Effect depends on the network traffic load
No effect if
Frequency Hopping is used
In case of downlink interference from a BCCH
carrier, if Server has 1 TRX only

Possibly a feature to handover from hopping


to non-hopping frequencies and vice versa?

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

79

Only effective on TCH carriers where DTX, PC, partial load my provide
interference free timeslots. For heavy traffic load no free time slots may be
available.

(C) Network Consultants / TNC Ltd 2005

79

BSS Parameters
Intracell handover
To avoid an endless loop of intracell
handover attempts:
Max_Number_Of_Repetitions = 5

Then go back to try other HO types

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

80

80

BSS Parameters
Intracell handover, default parameters
2 criteria indicate an interference situation:
Poor quality (high RXQUAL, same as Quality HO)
HO_Threshhold_Qual_DL/UL = 5

AND
Sufficient signal level
HO_Threshold_Interference_DL/UL
= 10dB above sensitivity?

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

81

81

BSS Parameters
Distance Handover
Maximum allowable BS-MS distance nearly
exceeded
MS_Range_Max = 61 (Bits Timing Advance, TA)
Max. Timing Advance (63) corresponds to about 35 km

In most cases, other handover criteria are more


effective than the Distance criterion alone
Combinations with other criteria can be useful,
e.g. elevated bridges, cross-water propagation etc.

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

82

Timing advance values can go up to 233s which is equivalent to 70 km


round trip distance from the BS. This is equal to a radial distance of 35 km.
1 TA value = 1 bit interval = 3.7 s
63 TA values = 63 x 3.7 s = 233 s
Cell range can be doubled up to 120 km roundtrip by using additional
timeslot to cope with the higher propagation delay. This is O.k. in rural areas
where capacity is not a big issue.

(C) Network Consultants / TNC Ltd 2005

82

BSS Parameters
Distance Handover
No TA information in idle mode
Sea
msRangeMax

Urban

Optimisation v 4.0

Distance handover AFTER


call setup: ensure neighbour
relationship is ONE-WAY

Network Consultants / TNC Ltd 2005

83

This example shows a bridge across a river between two urban areas. The
bridge is set low so that nearby cells have no line of sight and hence do not
provide sufficient coverage, In this case, the best server on the bridge is a
site that is much further away and is providing coverage across the water.

(C) Network Consultants / TNC Ltd 2005

83

BSS Parameters
Defaults for distance averaging
Ms_Distance_Averaging_windowSize = 20
Ms_Distance_Ho_Threshhold_nx = 8
Ms_Distance_Ho_Threshhold_px = 5

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

84

84

BSS Parameters
Umbrella handover
Umbrella cell
Quick area coverage
at network start
Covers gaps between
small cells later on
ic
af f
Problem: Umbrella cells are
Tr
prone to congestion
Handover mechanism required that relieves umbrella
cells from traffic
e.g. time based:Check in regular time intervals (parameter
hoPeriodUmbrella) if a handover into a small cell is
possible (RxLev_N > Umbrella_Threshold)

Also used for macro-microcell handovers

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

85

85

BSS Parameters
Handover timer parameters:
Minimum time between 2 handover requests,
minHOReqInt
Min_HO_Req_Int = 3 sec.
Long enough to reduce amount of handover signaling
Short enough to allow 2nd attempt before call drops

Time between 2 handover assessments, HO_Period


HO_Period = 2 sec.
Long enough to avoid ping-pong handovers
Short enough to allow correction after false
handover decision
Longer than PC_Interval

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

86

86

BSS Parameters
Exercise

R xQ u al

ho ThreshQ ual

pcU pperT heshQ u al

3
2
1

pcLow erThesh Q ual

0
pcUpperThreshLev

pcLowerThreshLev

hoThreshInterference

hoThreshLev

Sensitivity

R xLev

87

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

RxQ ual
c

h o T h re s h Q u al
p cU p p e rT h es h Q u al

6
5
4
3
2

p cL o w erT h es h Q u a l

1
0

(C) Network Consultants / TNC Ltd 2005

pcUpperThreshLev

pcLowerThreshLev

hoThreshInterference

hoThreshLev

Sensitivity

R xLev

87

BSS Parameters
Defaults for level averaging parameters
Handover
hoAveragingLevDL/UL_windowSize = 8
averagingWindowSizeAdjCell = 8
hoThreshholdLevDL/UL_nx = 4
hoThreshholdLevDL/UL_px = 3

Power Control

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

pcAveragingLevDL/UL_windowSize = 6
pcLowerThreshholdLevDL/UL_nx =2
pcLowerThreshholdLevDL/UL_px =2
pcUpperThreshholdLevDL/UL_nx = 2
pcUpperThreshholdLevDL/UL_px = 2

Network Consultants / TNC Ltd 2005

Given in SACCH-multiframes

88

88

BSS Parameters
Defaults for quality averaging parameters
Handover
hoAveragingQualDL/UL_windowSize = 4
hoThreshholdQualDL/UL_nx = 4
hoThreshholdQualDL/UL_px = 3

Power Control

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

pcAveragingQualUL_windowSize = 4
pcLowerThreshholdQualDL/UL_nx = 4
pcLowerThreshholdQualDL/UL_px = 3
pcUpperThreshholdQualDL/UL_nx = 32
pcUpperThreshholdQaulDL/UL_px = 32

Network Consultants / TNC Ltd 2005

89

89

BSS Parameters
Other handover parameters
rxLev_Min_Cell_#
-100 dBm (for GSM class 4 MS)
Minimum signal level of handover target cell
Approximately the same as MS receiver sensitivity

Can be used to prefer one neighbour over


another
e.g. for national roaming

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

90

90

BSS Parameters
Other handover parameters
Queuing
Waiting times in case of (target) cell congestion:
timeLimitHandover = 3 sec.
timeLimitCall = 5 sec.
How long does it take on average before a channel
becomes idle in a congested cell?

Max. number of queuing calls


maxQueueLength = 15%
Given as percentage of the number of TCHs
in the cell (15% corresponds to 1 per TRX)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

91

91

BSS Parameters
Other parameters Handover priorities
Priority for choice of handover target cell
Default: hoPriority_# = 3
Cells which are often congested can be given a
lower priority and thus relieved from traffic

!!! This can change the networks handover


behaviour drastically

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

92

92

BSS Parameters
Other parameters Handover priorities
Traffic load distribution
hoLoadFactor_# = 1
Priority of a cell is automatically reduced
by 1 in case of congestion (starts at 3)

hoLoadThreshhold = 85%
Percentage of TCHs busy for status Congested

Also: Traffic Reason Handover


Also: Directed retry for call setup

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

93

Directed Retry forces a mobile to attempt handover to second best server if


the best server is not available.

(C) Network Consultants / TNC Ltd 2005

93

BSS Parameters
Traffic load distribution
All measures that reduce the effective cell size can
also relieve the cell from traffic
Antenna tilt
Power reduction
(bsTxPwrMax and
msTxPwrMax equally)
Moving handover boundaries
(if sufficient overlap)
... only if there is significant
traffic in the affected areas

Usually a temporary solution only

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Default
handover boundaries

94

94

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

95

QUALITY ASSESSMENT

(C) Network Consultants / TNC Ltd 2005

95

Quality Assessment
Methods of assessing quality
Intensive drive tests with Test
Mobile System (TMS)
Network Statistics
MSC

OMC

Customer complaints

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

96

96

Quality Assessment
Quality criteria
The customers quality expectations are very simple:

Availability of the service anywhere, anytime


Call setup time within limits
Good speech quality during the call
Normal termination of the call

but SUBJECTIVE

To fix problems we need to quantify them:


Poor signal levels
High blocking rates
High bit error rates
Dropped calls/handover failures
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

97

97

Quality Assessment
Quality criteria
Call Success Rate, CSR
e.g. Successful call:
Registration possible
Call setup within 7 seconds
90% of the time RXQUAL < 4
Normal termination after 2 minutes

Optimisation target: e. g. 90% CSR

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

98

98

Quality Assessment
Quality criteria
Call Success Rate, CSR
Optimisation target: e. g. 90% CSR
Mature network target: e.g. 95% CSR
Including calls made indoors

Targets for each city / district


Worst cell: 90%

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

99

99

Quality Assessment
Test Mobile Systems
Collect MS measurement report data
(Downlink only!):
Incar
or

Outcar?

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Serving signal level


BER (RXQUAL)
Channel number
CI and LAC
Timing Advance TA
BSICs, signal and
power levels of neighbouring cells
Layer-3-Messages

Network Consultants / TNC Ltd 2005

100

100

Quality Assessment
Test Mobile Systems
Antenna on roof or mobile in car?
No indoor measurements few coverage problems
Attenuator to simulate indoor service?

Measurements at all times of day


Filter to produce busy hour statistics

Results for all cells


Filter to produce stats for each cell

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

101

101

Quality Assessment
Test Mobile System
Statistical mode
Used for wide area quality assessment and trend
identification
Automatic call generator generates test calls in
regular intervals
e.g. 2 minute calls with 30 seconds idle

Attempts to simulate subscriber behaviour

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

102

102

Quality Assessment
Test Mobile System
Criterion

Otherwise

Registration possible

No service

Call set up in 7 seconds

Blocked or Delayed

90% of RxQual 4 or MOS > 3

Noisy

Normal completion after 2 mins

Dropped

MOS = Mean Opinion Score


Objective voice quality measure defined by ITU
Compares recorded waveform with received
waveform
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

103

103

Quality Assessment
Test Mobile Systems
Drive test statistics
Call
attempts

calls

No
service

Successful

Blocked Dropped Noisy


calls
calls
calls

Delayed
setup

Outcar

350

92.6%

0.6%

1.4%

2.0%

3.4%

0.0%

Incar

357

75.6%

3.9%

2.8%

4.2%

13.5%

0.0%

Competitor 1

Competitor 2

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

104

Note the number of samples used. Is this sufficient?

(C) Network Consultants / TNC Ltd 2005

104

Quality Assessment
Test mobile system
Binomial distribution

Confide nce of 1%Confidence


a ccura
for
give
n num
of 1% cy
accuracy
for given
number
of calls be r of ca lls
100.0%

100.0%

90.0%
80.0%

Confidence Probability

90.0%

Confidence Probability

80.0%
70.0%
60.0%

70.0%
60.0%

Dropped call rate

50.0%

2%

40.0%

5%
10%

30.0%

20%

20.0%

Dropped call rate

10.0%

50.0%

0.0%
0

40.0% Optimisation v 4.0

100

200

300

400

500 600 700 800


Num ber of Calls

900 1000 1100 1200

2%
105

Network Consultants / TNC Ltd 2005

5%
10%

30.0%

20%

Notes:

20.0%
10.0%
0.0%

100

200

(C) Network Consultants / TNC Ltd 2005

300

400

500 600 700 800


Nu m b e r o f Calls

900

1000 1100 1200

105

Quality Assessment
Test Mobile System
Trace mode
Continuous call (re-established when dropped)
Used for

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Locating individual problem areas


Detailed analysis in problem areas
Quality assessment on rural highways etc.
BS Testing and Functional Testing

Network Consultants / TNC Ltd 2005

106

106

Quality Assessment
Test Mobile System
Statistical analysis of RXLEV, RXQUAL etc.
Coverage Statistics for City
100

Coverage Statistics for City

90

100
Probability of signal < x-axis %

80

90

Probability of signal < x-axis %

80

70

70

60
Network 1
Network 2

50

Network 3
Network 4

40

30

20

10

60
0
-110

-100

-90

-80

-70

-60

-50

-40

Network 1

Signal level, dBm

50

Network 2

107

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

Network 3
Network 4

40

Notes:

30

20

10

0
-110

-100

-90

-80

-70

-60

-50

-40

Signal level, dBm

(C) Network Consultants / TNC Ltd 2005

107

Quality Assessment

TMS in Trace Mode: Evaluation software


Optimisation v 4.0

Network Consultants / TNC Ltd 2005

108

Notes:

(C) Network Consultants / TNC Ltd 2005

108

Quality Assessment
Test Mobile System
Import into planning tool
Terrain or clutter
background
Comparison of
measured network
performance
vs. prediction
(eg handover
boundaries)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

109

109

Quality Assessment
OMC network performance statistics
BSC

BSC

MSC
OMC

BSC

BSC

MSC
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

110

110

Quality Assessment
OMC network performance statistics
Advantages over test drives:
Less labour intensive and time consuming
More comprehensive, large number of samples
Not limited to time of test drive
Uplink and downlink analysis possible
Real subscriber behaviour mix of outdoor, indoor, incar use, call patterns etc.,
but also including faulty MS, bad installations etc.

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

111

111

Quality Assessment
OMC network performance statistics
Disadvantages, limitations:
Limited geographical resolution: Entire cell
Call attempts in uncovered areas are not counted
Customer perception could be much worse

Network must have minimum load for reliable


statistics

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

112

112

Quality Assessment
OMC network performance statistics
Example: Identification of defective RXU
D A IL Y C E L L T C H R E P O R T

D A IL Y C
EL
PO RT
fr o m
2 4L
/0 1T
/9 7CtoH2 7R
/0 1E
/9 7
f r o m 2B4S /ID0/C1E/L9L 7
ID : t4o
8 -12
4 -7
0 6/201 1 / 9 7
B S ID /C E L L ID : 4 8 -1 4 -0 6 2 1

B S /S ec tor

C ell ID

c all attem p ts

b loc ked c alls

d rop p ed c alls

b loc k ed c all rate

0 6 2 1 /3

0621

541

11

2 .0 3

B S /S e c to r

C e ll ID

c a ll a tte m p ts

0 6 2 1 /3

0621

541

D A IL Y C E L L Q U A L IT Y R E P O R T
b lo c kfreodm c2 4a/0lls
1 /9 7 to 2 7 /0 1d/9r 7o p p e d
B S ID /C E L L ID : 4 8 -1 4 - 0 6 2 1

11

BS/
S ec tor

C ell ID

U p lin k
R XQ U A L
0 ..4

0 6 2 1 /3

0621

61951

d rop p ed c all rate


1 .6 6

c a lls

b lo c k e d c a ll r a te

d r o p p e d c a ll r a te

2 .0 3

1 .6 6

D AUIpLlinY
C DEowLn lin
Lk Q DUowAn linLk I T YU p R
k
lin kE P O
U pR
lin kT
R XQ U A L
QUAL
UAL
U AL
f r o m R2X0Q4..4U/A0L 1 / R9X7
t o 2R7X0Q/..40
1 / 9 R7XQ
5 ..7
5 ..7
5 ..7
%

B S ID /C
E L L I D6 :9 3 4 8 - 1943 .3-40 6 2 16 .6 6
65681

4423

D ow n lin k
R XQ U A L
0 ..4
%

D ow n lin k
R XQ U AL
5 ..7
%

9 8 .9 6

1 .0 4

N u m b e r o f 'c a ll a tte m p ts': 5 4 1

BS/
S e c to r

C e ll ID

0 6 2 1 /3

0621

e dk c a lls ': 1D
1 o 'b
k ek
d c a ll ra
U p linN ku m b e r o f 'bUlopc klin
wlonclin
Dteo':w n2 .0
lin3 k%
R X Q UNA
L b e r o fR'dXroQp pUe dA cLalls': 9R X'dQrop
U pAeLd c a ll rate
R X':Q1U.6A
um
6 L
%
0 .. 4N u m b e r o f 'n o5isy
.. 7c a lls ':
0
..
4
5
.
.7
35 'n o is y c a ll ra te ':
6 .4 7 %
N u m b e r o f 'g o o d c a lls':

61951

4423

Optimisation v 4.0

N u m b e r o f 'c a l l a t t e m p t s ': 5 4 1

4 8 6 'c a ll su c c e s s ra te ': 8 9 .8 3 %

65681

693

'b l o c k e d c a l l r a t e ':

N u m b e r o f 'd r o p p e d c a l l s ': 9

'd r o p p e d c a l l r a t e ': 1 . 6 6 %

N u m b e r o f 'n o i s y c a l l s ':

'n o i s y c a l l r a t e ':

N u m b e r o f 'g o o d c a l l s ':

(C) Network Consultants / TNC Ltd 2005

35

U p lin k
R XQ U AL
5 . .7
%

D o w n lin k
R XQ U AL
0 . .4
%

9 3 .3 4

6 .6 6

9 8 .9 6

Network Consultants / TNC Ltd 2005

N u m b e r o f 'b l o c k e d c a l l s ': 1 1

Notes:

U p lin k
R XQ U AL
0 . .4
%

D o w n lin k
R XQ U AL
5 . .7
%
1 .0 4

113

2 .0 3 %
6 .4 7 %

4 8 6 'c a l l s u c c e s s r a t e ': 8 9 . 8 3 %

113

Quality Assessment
OMC network performance statistics
Example: After replacement of defective RXU
D A IL Y C E L L T C H R E P O R T
f r o m 2 8 /0 1 / 9 7 t o 0 2 / 0 2 /9 7

D A I LB SYID /C
CEEL LLIDL: 4 8T- 1C
H REPORT
4 -0 6 2 1
f ro m 2 8 /0 1 /9 7 to 0 2 /0 2 /9 7
B S /S e c to r

C e ll ID

c a ll a tte m p ts

0 6 2 1 /3

0621

853

B S /S e c to r

C e ll I D

c a ll a t t e m p t s

0 6 2 1 /3

0621

853
BS/
S e c to r

0 6 2 1 /3

0 6 2 1 /3

2 .5 8

22

C e ll ID

U p lin k
RXQ UA L
0 ..4

0621

112399

U p lin k
R XQ UA L
5 ..7

D o w n lin k
R XQ U AL
0 ..4

D o w n lin k
R XQ U AL
5 ..7

d r o p p e d c a ll r a t e

2 .5 8

U p lin k
R XQ U A L
0 ..4
%

1 .5 2

U p lin k
RXQUAL
5 ..7
%

D o w n lin k
R XQ U AL
0 ..4
%

D A IL Y C E L L Q U A L IT Y R E P O R T
/ 9 7 t9o6 .7 80 2 / 03 .222 / 9 7 9 8 .6 8
3743 f r o m
1 1 4 6 02
7 8 / 011
535

D o w n lin k
R XQ U AL
5 ..7
%
1 .3 2

B S ID /C E L L ID : 4 8 -1 4 -0 6 2 1

22

'b lo c k e d c a ll r a te ': 2 .5 8 %
'd ro p p e d c a ll r a te ': 1 .5 2 %

3743

b lo c k e d c a ll r a t e

13

N u m b e r o f 'd r o p p e d c a lls ': 1 3

112399

1 .5 2

B S ID /C E L L ID : 4 8 - 1 4 -0 6 2 1

UNpu lin
lin
o yw cna lin
D3 . o
n lin k
m bk
e r o f 'n o is y Uc aplls
': k
2 6 'nDo is
ll r akte ':
0 5w %
R X Q U A L7 9 2 'cRaX
QUAL
RXQ UAL
R XNQu mUbA
L
e r o f 'g o o d c a lls ':
ll s u c c e s s ra te ': 9 2 .8 5 %
0 ..4
5 ..7
0 ..4
5 ..7

Optimisation v 4.0
0621

13

D A IL Y C E L L Q U A L IT Y R E P O R T

N u m b e r o f 'b lo c k e d c a lls ':

C e ll I D

22

2 8 /0 1 / 9 d
7 rt o
b lo c k e d f rco m
a lls
o p0 2p/ 0e2d/9 c7 a lls

N u m b e r o f 'c a ll a t te m p t s ': 8 5 3

BS/
S e c to r

d ro p p e d c a lls
B Sb lo Ic kDe d /c aClls E L
L I D : 4 8b-lo1c k e4d c-a0ll ra6te2 1d ro p p e d c a ll ra te

U p lin k
R XQ U AL
0 ..4
%

U p lin k
RXQ UAL
5 ..7
%

D o w n lin k
R XQ U AL
0 ..4
%

D o w n lin k
RXQ UAL
5 ..7
%

9 6 .7 8

3 .2 2

9 8 .6 8

1 .3 2

Network Consultants / TNC Ltd 2005


114607

1535

114

N u m b e r o f 'c a l l a t t e m p t s ': 8 5 3

Notes:

N u m b e r o f 'b l o c k e d c a l l s ':

22

'b l o c k e d c a l l r a t e ': 2 . 5 8 %

N u m b e r o f 'd r o p p e d c a l l s ': 1 3

'd r o p p e d c a l l r a t e ': 1 . 5 2 %

N u m b e r o f 'n o i s y c a l l s ':

26

'n o i s y c a l l r a t e ':

N u m b e r o f 'g o o d c a l l s ':

792

(C) Network Consultants / TNC Ltd 2005

3 .0 5 %

'c a l l s u c c e s s r a t e ': 9 2 . 8 5 %

114

Quality Assessment
Customer complaints
An additional source of information, but
difficult to handle

Complaint

Customer service must have clear simple procedure toForm


collect all relevant information

Evaluation:
Pins on map to identify problem areas
Possibly better resolution than OMC statistics
Large number of potential test persons

Categorise for clues on possible optimisation measures

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

115

115

Quality Assessment
Limitations
Regardless of the assessment method, some
problems only show under traffic load
Congestion
Interference
(except for BCCH downlink interference)

Ongoing quality control/optimisation after launch

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

116

116

Quality Assessment
Systematic quality assessment
Quality targets may change:
Emerging mobile markets:
Customers tend to accept occasional
coverage holes and poor quality

Mature markets: Focus on quality


Customers take coverage for granted
and start focussing on service quality
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

117

117

Quality Assessment
Systematic quality assessment
Quality targets may change:
Launch target:
95% CSR outdoor
90% for worst cell

Mature network target:


98% CSR outdoor / 95% CSR indoor
95% for worst cell / 90% for worst cell

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

118

118

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

119

Optimisation Measures

(C) Network Consultants / TNC Ltd 2005

119

Optimisation Measures

Overlaid
microcells

Fine tuning of antenna


orientations and tilt
Adding
TRX

Repeaters
Preamps.

Frequency
changes

Sectorisation

Half
rate
Cell parameter
settings

Underlay/
Overlay

Frequency Hopping,
Power Control, DTX

Dual band Dual mode

Cell Splitting

Effect

Possible network optimisation measures:

Cost, Effort
Optimisation v 4.0

Network Consultants / TNC Ltd 2005

120

The diagram above give a graphic depiction of cost / benefits realised by


applying different optimisation measures to solve problems in the network.
It can be clearly seen that the most efficient (least cost, maximum effect)
optimisation solutions can be achieved by applying advanced network
features such as power control, frequency hopping, DTX.
Also changes to site configuration such as antenna tilts and azimuth are
more effective than change cell parameter setting.
The most effective and COSTLY, solutions involve adding new hardware
such as sectorisation, micro-cells and sites.

(C) Network Consultants / TNC Ltd 2005

120

Optimisation Measures
Optimisation measures
Fine tuning
Adding features
Frequency hopping
Power Control
Discontinuous transmission
Hierarchical cell structures

Antenna orientations, height and tilt


Handover and power parameters
Neighbour lists
Frequency changes

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

121

121

Optimisation Measures
Optimisation measures
Network extension
Dual band, dual mode
Adding TRX
Half (dual) rate coding
Preamplifiers
Sectorisation
Cell splitting
Repeaters
Overlaid micro- and picocells

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

122

122

Optimisation Measures
Antenna fine tuning
Typical antenna beam patterns

Typical 3dB
beamwidths:

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

123

123

Optimisation Measures
Antenna fine tuning
Horizontal plane:
Possible coverage weaknesses between sectors
Interference reduction
Traffic load distribution

Vertical plane:
Interference reduction
Possible coverage weaknesses in the short to medium
distance range
Traffic load distribution

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

124

124

Optimisation Measures
Example: Weak coverage between sectors
Best server
Other BS
Problem
area

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

125

125

Optimisation Measures
Example: Weak coverage between sectors
Sector B: 210
65 horizontal beamwidth

Problem
area

Measured RXLEV_DL before antenna change


Optimisation v 4.0

Network Consultants / TNC Ltd 2005

126

Problem area is an area of weak coverage situated between sectors.

(C) Network Consultants / TNC Ltd 2005

126

Optimisation Measures
Example: Weak coverage between sectors
Sector B: 180
65 hor. beamwidth

Problem
area

Measured RXLEV_DL after antenna change

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

127

In this case, the simplest and most cost effective solution to this coverage
problem was to re-orientating sector 3 to point at 180, thereby increasing
the coverage in that area.
However, it is also important to verify that by changing the azimuth of sector
3 that coverage is not been adversely affected in an important area, and
that any new interference is not causing problems.

(C) Network Consultants / TNC Ltd 2005

127

Optimisation Measures
Raising or lowering antenna height
to improve coverage
to reduce interference

Rarely possible due to constructional


constraints

Need large change to be effective


(4 to 5 dB path loss change when doubling
antenna height)
unless the obstruction situation changes
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

128

128

Optimisation Measures
Adjusting antenna height: Exercise!

25m

PL = K 13.8 log h + (44.9-6.55 log h) log d

1km

h2

4km

What is the difference in C/I performance with h2 = 25m and 50m?


What is the ratio of the number of frequencies per cell in each situation?
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

129

129

Optimisation Measures
Example: Weak coverage below main lobe

7vert. beamwidth,
2 electrical downtilt,
no mechanical downtilt

Problem area:
City centre
(Indoor coverage)

Measured RXLEV_DL before antenna change


Optimisation v 4.0

Network Consultants / TNC Ltd 2005

130

Slide shows a City Centre area which situated in the main lobe of the
antenna and close to the site. Marketing requires in-building coverage in the
city centre, however, Test mobile data shows that the measured downlink
RxLev is not sufficient.

(C) Network Consultants / TNC Ltd 2005

130

Optimisation Measures
Example: Weak coverage below main lobe

Measured RXLEV_DL before antenna change, terrain height


Optimisation v 4.0

Network Consultants / TNC Ltd 2005

131

Further analysis of this test drive show that the city centre is situated in a low
lying area below the antenna which is located on a hill above. It is likely that
the coverage to the city centre is being provide by lower part of the antenna
vertical lobe, which is heavily attenuated by the antenna vertical pattern.

(C) Network Consultants / TNC Ltd 2005

131

Optimisation Measures
Example: Weak coverage below main lobe
BS Ant.
160m

2 electrical down

0
tilt

3dB-poin
t: 5.5
= ar
ctan (
160/1
200) =
7.6

City

1200m

Antenna pattern and angles

Solution: Add 4 mechanical downtilt

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

132

Adding a 4mechanical down tilt will place the main beam of the antenna
directly on to the city centre and hence increase the coverage.

(C) Network Consultants / TNC Ltd 2005

132

Optimisation Measures
Example: Weak coverage below main lobe

City

Innenstadt

Tool simulation with 0 and 4 mechanical downtilt


Optimisation v 4.0

Network Consultants / TNC Ltd 2005

133

Tool simulations can be carried out to see the effects of before and after
applying the down tilt. As the picture on the right shows, the down tilt has the
effect of increasing the signal level to the required in-building level (red) in
the city centre.

(C) Network Consultants / TNC Ltd 2005

133

Optimisation Measures
Example: Weak coverage below main lobe

Verification test drive


with 4additional
mechanical downtilt

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

134

A test mobile drive test is carried out after the application of the down tilt to
verify that the measured RxLev levels are meeting the requirements.

(C) Network Consultants / TNC Ltd 2005

134

Optimisation Measures
14: -18dB
4: -3dB
0

Antenna downtilt
Interference:
-18dB?

Typical vertical
beam pattern

14: -18dB
Possible loss
of coverage

Coverag
e

In practice, reflections fill up nulls and sidelobes


especially in urban areas

(Nebenkeule)

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

135

In very high density networks where capacity must be maximised, then


systematically tilt so that the first null in the antennas vertical pattern points
towards the horizon. This will minimise interference.

(C) Network Consultants / TNC Ltd 2005

135

Optimisation Measures
Systematic antenna downtilt
Minimum: Vertical main lobe on cell edge

= arctan

hBS
r

hBS
r

Maximum: First null angle


Otherwise side lobe will increase interference

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

136

Many operators buy antennas with 2of vertical electrical downtilt as


standard. This is based on the assumption that the main lobe should point at
the cell edge. In urban areas of course, the main lobe should point at the
roof tops at the cell edge, not at the ground. Also the amount of interference
reduction from 2of downtilt is minimal.

(C) Network Consultants / TNC Ltd 2005

136

Optimisation Measures
Electrical vs. mechanical downtilt
0

Electrical, with null-filling

Mechanical

Advantages:

Back lobe tilts down instead of up


Null-filling becomes possible (better lower side lobe)
Antenna remains upright (aesthetics)
Mechanical downtilt can be added

Disadvantages:
Antennas are more expensive
Slightly compromised interference reduction potential
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

137

137

Optimisation Measures
Reduced flexibility with omni cells
Only electrical downtilt,
same for entire cell
Only 1 set of handover /
power parameters for
entire 360 cell edge
No load sharing
between sectors

Omni site

Sectorised cell site with


different downtilt angles
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

138

138

Optimisation Measures
Other ways of improving coverage
Increasing BS output power?
Cell parameters bsTxPwrMax, msTxPwrMax

Better BS RX sensitivity or pre-amplifier


Must be matched by higher
BS TX power for balanced
link budget

nk:
Downli

MS
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

-109 dBm

Uplink

-104 dBm

42 dBm
37 dBm

BS
139

139

Optimisation Measures
Other ways of improving coverage
Before

Cell splitting, sectorisation

After

Difficult, expensive
Primarily used for capacity enhancement

Hierarchical cell structures


Micro- and picocells
Repeaters
New sites needed
Rep:f1
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

BSC

Rep:f1

BS:f1

Micro:f4

Network Consultants / TNC Ltd 2005

140

140

Optimisation Measures
Cell parameter optimisation
Only a few groups of parameters contain
potential for optimisation effects:
Power parameters
Handover parameters
Averaging parameters
BSIC and LAC (Base Station Identity and Location
Area Codes)
Some other parameters
Some of these have to be treated with extreme care as
they can have very severe effects on the network!

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

141

141

Optimisation Measures
Cell parameter optimisation
Default parameter sets:
Standard settings suitable for most cases
Starting point for possible optimisation
There may be different sets of default parameters,
e.g. for:
Different HW/SW configuration (e.g. different BTS
types, with/without pre-amplifier etc.)
Different areas (e.g. urban, highway etc.)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

142

142

Optimisation Measures
Cell parameter optimisation
Other parameters have no direct
optimisation potential
e.g. CI (Cell Identifier)
but maintaining data consistency is important
and a major difficulty in network optimisation

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

143

143

Optimisation Measures
Neighbour lists
Missing and too many neighbours cause problems

Missing neighbours

Too many neighbours


Inaccurate sig lev measurements

Cell dragging
Ideally: 6 to 8
neighbours
In practice:
8 to 12
Optimisation v 4.0

False handover decisions


Interference
Poor quality

Dropped calls

Too many neighbours


also makes frequency
planning much more
restricted

Network Consultants / TNC Ltd 2005

144

MS collects (100 / # of neighbours) samples of signal strength of neighbour


cells.
Therefore the more neighbours the fewer samples used to make handover
decisions the higher the probability of a false decision.

(C) Network Consultants / TNC Ltd 2005

144

Optimisation Measures
Neighbour cell lists
Cell dragging
Server
Defined neighbours
Missing neighbour

f1

Interferer

results in
poor serving signal levels
increased interference

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

f1

145

145

Optimisation Measures
Example: Missing neighbour definition

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

146

Picture above shows an area of very poor coverage next to BS 0044. It is


surprising to see such a poor signal level (< -95 dBm) next to the site.

(C) Network Consultants / TNC Ltd 2005

146

Azimuth

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Cell ID.

Cell ID.

Azimuth

Ch. No.

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Ch. No.

783

147

783

147

Optimisation Measures
Example: Missing neighbour definition
Poor signal levels and dropped calls in the town
that is to be covered by BS 0044
0044 is rarely used as the serving cell in the town
0225-1 (about 12 km away, TA=22!) acts as the
best server in the area between 0014 and 0044
No neighbour relation between 0225-1 and 0044

Calls on 0225-1 are dragged into the town,


no handover to 0044 is possible

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

148

148

Optimisation Measures
Example: 0044-1 and 0225-1 now defined as neighbours

Azimuth

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Cell ID.

Cell ID.

Azimuth

Ch. No.

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Ch. No.

783

149

783

149

Handover 0225-1
> 0044-1
Handover
0044-1
> 0255-1
Azimuth

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Network Consultants / TNC Ltd 2005

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Cell ID.

Cell ID.

Azimuth

Ch. No.

0013 1

20

793

0013 2

120

795

0013 3

300

not relev.

0014 1

Omni

805

0044 1

Omni

808

0225 1

210

798

0225 2

350

not relev.

0499 1

40

Ch. No.

783

150

783

150

Optimisation Measures
Optimising neighbour cell lists
Too many neighbour definitions:
inaccurate measurements
Only about 100 measurement samples are
possible during one measurement period for all
defined neighbour cells together
[ e. g.:
6 neighbours: 16 - 18 samples per cell
32 neighbours: 3 - 4 samples per cell
Fewer samples lead to inaccurate averages,
false handover decisions, interference

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

151

151

Optimisation Measures
Optimising neighbour cell lists
Too many neighbour definitions:
frequency planning difficulties
Too many neighbours can make adjacent channel
frequency planning impossible...

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

152

152

Optimisation Measures
Optimising neighbour cell lists
Too many neighbour definitions:
inappropriate handovers

Patches of coverage
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

153

153

Optimisation Measures
Frequency changes

Example: Identification of downlink


interference with network statistics

B S /C e ll c a ll
S e c to r ID a tte m p ts

b lo c k e d
c a lls

d ro p p e d
c a lls

b lo c k e d
c a ll
ra te

d ro p p e d
c a ll
ra te

0 0 8 8 :2
0088:2

120

62

4 .4 6

2 .3 0

2691

D A ILY C E LL T C H R E P O R T
f ro m 2 8 / 0 1 /9 7 to 0 3 /0 2 / 9 7
B S ID /C E L L I D : 4 2 -1 4 -0 0 8 8 : 2

D A I L Y C E L L Q U A L IT Y R E P O R T
f ro m 2 8 /0 1 /9 7 to 0 3 / 0 2 /9 7
B S ID /C E L L ID : 4 2 -1 4 -0 0 8 8 :2
B S /C e ll
S e c to r ID

U p lin k
U p lin k
D o w n lin k D o w n lin k U p lin k
U p lin k
R XQ U AL R XQ U AL R XQ U AL R XQ U AL R XQ U AL R XQ U AL
0 ..4
5 ..7
0 ..4
5 ..7
0 ..4
5 ..7
p r oz .
p ro z .

0 0 8 8 :2

354137

8465

345801

16801

9 7 .6 7

2 .3 3
2.33

D o w n lin k
R XQ U AL
0 ..4
p ro z .

D o w n lin k
R XQ U AL
5 ..7
p ro z .

9 5 .3 7

4 .6 3
4.63

N u m b e r o f 'c a ll a tt e m p t s'

2691

N u m b e r o f 'b lo c k e d c a lls '

120

'b lo c k e d c a ll ra t e '

: 4 .4 6 %

N u m b e r o f 'd ro p p e d c a lls'

62

'd ro p p e d c a ll ra te '

: 2 .3 0 %

N u m b e r o f 'n o is y c a lls '

116

'n o isy c a ll ra te '

: 4 .3 1 %

N u m b e r o f 'g o o d c a lls'

2393

'c a ll s u c c e ss ra te '

: 88.93%
8 8 .9 3 %

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

154

The report shows that for BS0088:2, the call success rate is poor, and also
that the proportion of poor downlink RxQual measurement values are rather
high.

(C) Network Consultants / TNC Ltd 2005

154

Optimisation Measures
Example: Downlink interference,
frequency change
Downlink RXQUAL rate is twice as high as Uplink
Faulty TXU test was negative
Downlink interference
Verification through handover statistics:
BS:Sec. Total_HOUL_Qual UL_Lev UL_Interf DL_Qual DL_Lev PBGT
0088:1

6380

12%

19%

1%

11%

17%

40%

0088:2

2677

5%

14%

0%

32%

18%

31%

0088:3

5464

6%

9%

1%

20%

10%

54%

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

155

Handover statistics show an unusually high number of handover triggered by


poor downlink quality, in same proportion as those triggered by power
budget. This is an indicator that we have a poor downlink quality, probably
being caused by downlink interference from another site.

(C) Network Consultants / TNC Ltd 2005

155

Optimisation Measures
Example: Downlink interference,
frequency change
Interferer: BS 0087:3 (about 5 km away)

797

6
797

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

156

Analysis of the surrounding sites, shows a possible interference source from


BS0087:3 (channel 797)

(C) Network Consultants / TNC Ltd 2005

156

Optimisation Measures
Example: Downlink interference
RXQUAL measurement with TMS in Camp
Mode (to avoid QUAL-handovers)

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

157

157

Optimisation Measures
Example: Downlink interference,
After frequency change at BS 0088:2 from
797 to 817

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

158

158

Optimisation Measures
Example: Downlink interference
Network statistics after
frequency change
B S /C e ll c a ll
S e k to r ID a tte m p ts

b lo c k e d
c a lls

d ro p p ed
c a lls

b lo c k e d
c a ll
r a te

d r op p ed
c a ll
r a te

0 0 8 8 :2

68

56

2 .9 3

2 .4 1

2321

D A IL Y C E L L T C H R E P O R T
f r o m 2 4 /0 2 / 9 7 t o 0 2 /0 3 / 9 7
B S ID /C E L L ID : 4 2 -1 4 -0 0 8 8 :2

D A IL Y C E L L Q U A L I T Y R E P O R T
f r o m 2 4 /0 2 / 9 7 t o 0 2 /0 3 / 9 7
B S ID /C E L L ID : 4 2 -1 4 -0 0 8 8 :2
B S /C e ll
S e k to r ID

U p lin k
U p lin k
D o w n lin k D o w n lin k
R XQ U AL R XQ U A L R XQ UA L R XQ UA L
0 ..4
5 ..7
0 ..4
5 ..7
p roz.

U p lin k
R XQ UA L
0 ..4
p roz.

U p lin k
RXQ U AL
5 ..7
p roz.

D o w n lin k D o w n lin k
RXQ U AL RXQ U AL
0 ..4
5 ..7
p roz.

0 0 8 8 :2

257077

9 7 .2 5

2 .7 5

9 7 .8 4

7264

258636

5705

N u m b e r o f 'c a ll a t te m p t s '

2321

N u m b e r o f 'b lo c k e d c a lls '

68

'b lo c k e d c a ll r a te '

: 2 .9 3 %

N u m b e r o f 'd r o p p e d c a lls '

56

'd r o p p e d c a ll r a te '

: 2 .4 1 %

N u m b e r o f 'n o is y c a lls '

60

'n o is y c a ll r a t e '

: 2 .5 9 %

N u m b e r o f 'g o o d c a lls '

2137

Optimisation v 4.0

'c a ll s u c c e s s r a te '

2 .1 6
2.16

: 92.07%
9 2 .0 7 %

Network Consultants / TNC Ltd 2005

159

After frequency change, the call success rate is higher, and the proportion of
poor RxQual for the downlink has decreased to the same proportion as that
of the uplink.

(C) Network Consultants / TNC Ltd 2005

159

Optimisation Measures
Example: Downlink interference,
Handover statistics after frequency change
BS:S

Total_HO UL_Qual UL_Lev

UL_Interf DL_Qual DL_Lev

PBGT

0088:1

5867

12%

22%

1%

11%

17%

37%

0088:2

2582

5%

19%

0%

4%

25%

47%

0088:3

4619

6%

9%

0%

20%

8%

57%

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

160

160

Optimisation Measures
Frequency changes
Highly effective against severe cases of
interference
Far-reaching side effects
Re-planning of other frequencies may become
necessary
Often difficult, especially in high-density areas

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

161

161

Optimisation Measures
Frequency changes
Can be done at either cell
(interferer or victim)
Often randomly significantly easier on one side
than on the other

Incompatibility of interferer and victim must


be kept in the planning tools interference
matrix!
For subsequent retuning efforts
Until major redesign in the area

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

162

162

Optimisation Measures
Frequency changes
Short term solution:
BCCH/TCH swapping

Victim
Before

Interferer

BCCH: 794
TCH: 797

BCCH: 794

Can produce some interference relief if the affected


frequency is a BCCH carrier
BCCH: 797
After

TCH: 794

BCCH: 794

Doesnt work for all types of interference


Effect depends on TCH traffic load
Results in non-optimal mix of BCCH and
TCH sub-bands
Usually a temporary measure only

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

163

163

Optimisation Measures
Increasing network capacity
The relationship between quality and capacity
In a congested network, quality can deteriorate very
quickly:
Congestion
Extended call
setup times

Poor speech
quality

Interference/
Noise
Unavailability
of service
Violation of all 4 basic quality criteria
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Dropped
calls
164

164

Table of Contents
95 Quality assessment

2 Introduction

Quality criteria

Definitions
Why optimise?

Call success rate

Project phases

Test Mobile Systems

Group interfaces

Network performance statistics

Network expansion

Customer complaints

119 Optimisation measures

6 System basics

Antenna fine tuning

GSM air interface


Measurements

Improving Coverage

Idle mode procedures

Cell parameter optimisation

Call setup procedures

Optimising neighbour cell lists


Frequency changes

32 BSS parameters

Common problems
Advanced techniques

Idle mode
Measurement averaging
Power control
Handover

165 Organisation
Timetables
Teams
Preparation
Databases
Implementation procedures

Optimisation v 4.0

Network Consultants / TNC Ltd 2005

165

Organisation

(C) Network Consultants / TNC Ltd 2005

165

Organisation
The relationship between network
design and network optimisation
Thorough network design/planning from the start
can reduce the optimisation effort significantly!
In a poorly planned network, achievable optimisation
effects without major redesigns are rather marginal

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

166

166

Organisation
The relationship between network
design and network optimisation
A close link between the two activities is
necessary:
Network planners must be involved in the optimisation
Optimisation results must be fed back into the planning process

Or better still: a small team is responsible for


design, expansion and optimisation of a
local area.
Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

167

167

Organisation
Optimisation stages
Initial optimisation
Minimum service quality for commercial launch
Or after a major network change (turn on cycle)

Ongoing optimisation
Continuous quality control and improvement
Design,Construction,Installation

Optimisation v 4.0

Ongoing Optimisation
Launch

BSn

Area
complete

BS1
BS2

Initial...

Network Consultants / TNC Ltd 2005

Network
extensions?

168

The calendar is often organised into turn-on cycles when groups of new
sites are brought on air together, around once every 10-15 weeks. Each
turn-on is followed by intensive optimisation of the frequency plan and
neighbour lists.
A short cycle runs continuously to fix problems as they are found, possibly
resulting in a frequency re-tune once a week.

(C) Network Consultants / TNC Ltd 2005

168

Organisation

Drop-dead
date

tia
in i
e(
ns
iv
Int
e

3m
on
th
an Tra
d T ffic
RX
f
co orec
u n ast
t
Fre
qu
en
cie
sa
nd
pa
ram
ete

g
nin
lan
SP
wB
Ne

12

mo
n th
an Tra
d B ffic
S c fo
ou rec
as
nt
t

rs

l) o
pti
mi
sa

tio

The Optimisation and Growth cycle

Turn-on
date

Ongoing optimisation
Optimisation v 4.0

Network Consultants / TNC Ltd 2005

169

Drop-dead date is the date that the site construction is completed.


Turn-on date is the site on air date.

(C) Network Consultants / TNC Ltd 2005

169

Organisation
Project phases
Cluster definition

An iterative
process!

BS testing / Functional testing

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Systematic quality assessment


Optimisation

Network Consultants / TNC Ltd 2005

170

170

Organisation
Cluster definition
Sub-networks small enough
for a two person team to manage
Large enough to be self contained
Not too many relationships
with neighbouring clusters

About 15 to 30 sites
About

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

20km2

Network Consultants / TNC Ltd 2005

171

171

Organisation
Cluster definition
Wanted: Stable sub-networks
Ideally:
All BS within the sub-network
fully operational,
plus:
All handover neighbours
outside the sub-network
Nearest co- and adjacent
channel interferers

Otherwise:
Repeated optimisation effort!

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

172

172

Organisation
BS testing
of single BS in the cluster
Objectives:
Verify network configuration against current
planning status
Identify and eliminate equipment faults (hard/software) and installation errors before
optimisation begins

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

173

173

Organisation
BS testing
When?
Whenever a single BS reaches the
Ready-to-launch status

Who?
Installation personnel after completion of job
Vendor and operator staff together
Can be a contractual issue

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

174

174

Organisation
BS testing
Antenna installation check
visual/manual on site
height, orientation and tilt

Basic cell parameters and functions


BCCH and other carrier frequencies,
BSIC, CI, LAC
Neighbour list, consistency
Handover and power parameters
Call setup on all timeslots and speech quality
Handover to other sectors

Park BS until next ToC

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

175

175

Organisation
Functional testing
Objectives
Verify site integration into current plan

When?
When cluster is complete, just before systematic
quality assessment and optimisation starts

Why again?
Plan may have changed since BS testing

Who?
Teams who will also carry out the drive tests

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

176

176

Organisation
Procedure
Identify areas with quality problems
Drive tests, network statistics, customer complaints

Trace source of the problem


Trace measurements, tool predictions
Coordination with other teams

Agree and implement solution


Update databases!
Verify effect

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

Optimisation

Suggest solution, predict side effects!

177

177

Organisation
Strict database management is essential!
Planning tool database
Current status of the network must be available at all
times

OMC database
Maintain parameter consistency
Save into next implementation cycle

Measurement database
Date and time, equipment used

Optimisation database
History of optimisation measures taken and removed,
comments

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Network Consultants / TNC Ltd 2005

178

178

Organisation
Approximate time plan for a sub-network

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

Milestone

Effort

Progress

Sub-network definition
Network data established
BS testing
Functional testing
1st quality assessment
Trace measurements
Optimisation
2nd quality assessment
Quality target reached

5%
10%
15%
20%
10%
10%
20%
10%

5%
15%
30%
50%
60%
70%
90%
100%

Network Consultants / TNC Ltd 2005

179

179

!"
#
& #'( )

Optimisation v 4.0

(C) Network Consultants / TNC Ltd 2005

#
%

Network Consultants / TNC Ltd 2005

180

180

Você também pode gostar