3 g tems analysis n doc

المـــــدرســــــة العليـــــا للمـــــواصــــالت بتــــونــــــس

High School for Communications of Tunis

Telecommunication Engineering

Major: Mobile Services and Networks

Graduation Project Report

Topic:

Ericsson 3G Trial Network Optimization

Realized by:

Sofien Jouini

Supervisors:

Mr. Sami Tabbane

Mr. Mohamed Tahar Ferchichi

Project carried out within:

Academic year: 2006-2007

Dedication

Dedication

To my Parents

Acknowledgment

Acknowledgment

First and foremost, I would like to express my deep gratitude and appreciation to

my training supervisor Mr Mohamed Tahar Ferchichi (N&TC Manager in Ericsson)

for his efforts, his consistent and generous support during the project schedule as well

as my recognition for offering me the opportunity to carry this work at Ericsson

Tunisia.

I’m also grateful for Mr Sami Tabbane, my supervisor at Sup’com for all the

help he gave me, his encouragement and advises in both technical and non-technical

matters.

I would like also to express my sincerely thanks for all the working team at the

ELS department of Ericsson Tunisia for their precious help and documentation they

provided me with, Special thanks for Mr Tahar Labidi (N&TC Consultant in

Ericsson).

I also take the opportunity to mention my respect and gratitude for the members

of my PFE evaluation committee for their acceptance to asses my work.

Abstract

Sofien Jouini - iv - PFE 2006/2007

Abstract This project was elaborated in the purpose to optimize Ericsson 3G trial

network in Tunisia and study the impact of its HSDPA upgrade in later phases.

Optimization activity deals with two main issues that were marginalized during

the first and second phases of Ericsson 3G project, neighbours list optimization and

isolation between coexisting antennas (2G/3G and 3G/3G). Likewise, a baseline drive

test was conducted to asses the network performance and propose the required

changes.

The study of HSDPA impact aims to predict the network performance after

HSDPA upgrade, to define the key network performances that will be impacted, and

finally to propose an optimum strategy to deploy HSDPA.

Key Words; UMTS, Neighbours list, Co-existence, Initial tuning, HSDPA impact

Summary

Sofien Jouini - v - PFE 2006/2007

Summary

Acknowledgement Abstract General introduction ……………………………………………………………………….…1

I. Chapter 1: Overview to 3G....................................................................................3

I.1. Introduction .............................................................................................................3

I.2. R99 Networks...........................................................................................................3

I.2.1. Architecture & Interfaces ...............................................................................3

I.2.2. Functionalities of RAN Elements....................................................................5

a) NodeB ............................................................................................................5

b) Radio Network Controller (RNC)..................................................................6

c) RXI .................................................................................................................6

I.3. Migration to HSDPA ...............................................................................................6

I.3.1. R99 to R4 to R5 migration ..............................................................................6

I.3.2. HSDPA Definition ..........................................................................................7

I.3.3. HSDPA features..............................................................................................7

a) Short Transmission Time Interval (TTI) ........................................................8

b) Fast radio-dependent scheduling ..................................................................8

c) High-order modulation ..................................................................................9

d) Fast link adaptation.....................................................................................10

e) Fast hybrid ARQ with soft combining .........................................................10

f) Efficient Cell Power Utilization ...................................................................11

I.3.4. HSDPA channels ..........................................................................................11

I.3.5. SW/HW upgrade for HSDPA introduction ...................................................12

a) RBS...............................................................................................................12

b) RNC .............................................................................................................13

I.4. Ericsson 3G Project...............................................................................................13

I.4.1. Architecture ..................................................................................................13

I.4.2 .Coverage.......................................................................................................15

I.4.3. Services .........................................................................................................16

Summary

Sofien Jouini - vi - PFE 2006/2007

a) Traffic classes ..............................................................................................16

b) Radio Access Bearers (RABs)......................................................................17

c) Mapping of 3G services in RABs .................................................................18

d) Services offered by Ericsson 3G Network in Tunisia ..................................19

I.5.Conclusion ..............................................................................................................19

II.Chapter 2: Network optimization ...........................................................................20

II.1. Introduction ..........................................................................................................20

II.2. Neighbors list optimization ..................................................................................20

II.2.1. Definitions ...................................................................................................21

a) Compressed mode algorithm .......................................................................21

b) Generated list by TCPU ..............................................................................22

II.2.2. Problem study..............................................................................................23

II.2.3. Conclusion...................................................................................................24

II.3. Co-existence problems .........................................................................................25

II.3.1. Definitions ...................................................................................................25

a) Spurious emissions ......................................................................................25

b) Receiver blocking ........................................................................................26

c) Isolation .......................................................................................................26

d) RBS sensitivity degradation.........................................................................28

II.3.2. Problem study..............................................................................................29

a) Spurious emission: GSM TX into WCDMA RX ...........................................29

b) Spurious emission: WCDMA TX into WCDMA RX.....................................30

c) WCDMA Receiver blocking .........................................................................31

II.3.3. Conclusion...................................................................................................31

II.4. Initial tuning .........................................................................................................32

II.4.1. Definition and Purpose ...............................................................................32

II.4.2. Process ........................................................................................................32

a) Preparation phase .......................................................................................33

b) Radio Network (RN) audit ...........................................................................33

c) Data collection.............................................................................................34

d) Post processing............................................................................................34

e) Analysis ........................................................................................................34

II.4.3. Encountered problems.................................................................................35

a) Poor coverage..............................................................................................35

Summary

Sofien Jouini - vii - PFE 2006/2007

b) Missing neighbor .........................................................................................36

c) Pilot pollution and wrong parameters configuration ..................................37

d) Not allowed PLMN ......................................................................................38

II.5. Conclusion............................................................................................................39

III. Chapter 3: HSDPA impact...................................................................................40

III.1. Introduction.........................................................................................................40

III.2. Impact of HSDPA; theoretical study...................................................................41

III.2.1. Impact on Ec/No values .............................................................................41

III.2.2. Impact on coverage....................................................................................42

III.2.3. Impact on capacity .....................................................................................44

III.2.4. Impact on traffic distribution .....................................................................44

III.3. Practical study; Simulation with TCPU .............................................................45

III.3.1. TCPU and Monte Carlo method ................................................................46

a) TCPU ...........................................................................................................46

b) Monte Carlo method....................................................................................46

c) Process of Monte Carlo Simulation in WCDMA Analysis .........................46

III.3.2. Simulation process .....................................................................................49

a) Setup common channel power .....................................................................49

b) Setup HSDPA enabled cells.........................................................................50

c) Define HSDPA related RABs .......................................................................50

d) Define WCDMA Bearer Rate Sets ...............................................................51

e) Define HSDPA capable terminal .................................................................52

f) Run network analysis ...................................................................................53

III.3.3. Simulation result ........................................................................................53

a) Impact on coverage......................................................................................53

b) Impact on capacity.......................................................................................56

c) Traffic distribution .......................................................................................57

d) Quality .........................................................................................................59

III.4. Proposal for HSDPA deployment strategy .........................................................60

III.4.1. Proposal 1 ..................................................................................................60

III.4.2. Proposal 2 ..................................................................................................63

III.5. Conclusion ..........................................................................................................64

Figures list

Sofien Jouini - viii - PFE 2006/2007

Figures list

Figure1. 1 : UMTS networks architecture .....................................................................5

Figure1. 2 : MSC architecture evolution from R99 to R4 .............................................7

Figure1. 3 : Downlink data throughput improvement ...................................................7

Figure1. 4 : proportional fair scheduling algorithm......................................................9

Figure1. 5 : QPSK and 16 QAM....................................................................................9

Figure1. 6 : fast link adaptation ...................................................................................10

Figure1. 7 : Fast hybrid ARQ with soft combining .....................................................10

Figure1. 8 : Efficient Cell Power Utilization in HSDPA.............................................11

Figure1. 9 : HSDPA channels......................................................................................12

Figure1. 10 : HS-TX board ..........................................................................................12

Figure1. 11 : Software and hardware upgrade of RNC ...............................................13

Figure1. 12 : Ericsson 3G network architecture ..........................................................14

Figure1. 13 : Grand Tunis area coverage.....................................................................15

Figure1. 14 : Highway and Hammamet Areas coverage .............................................16

Figure1. 15 : UMTS and Radio Access Bearer Service...............................................17

Figure2. 1 : Compressed Mode algorithm impact .......................................................21

Figure2. 2 : 2G-3G neighbours list generation ............................................................22

Figure2. 3 : neighbours list details...............................................................................23

Figure2. 4 : optimized neighbours’ list ........................................................................24

Figure2. 5 : Inter-modulation product..........................................................................26

Figure2. 6 : Wide Band Noise......................................................................................26

Figure2. 7 : Isolation; co-area case ..............................................................................27

Figure2. 8 : Isolation; co-site case ...............................................................................27

Figure2. 9 : Initial tuning activity process ...................................................................32

Figure2. 10 : poor coverage .........................................................................................36

Figure2. 11 : Missing neighbour..................................................................................37

Figure2. 12 : Pilot pollution and wrong parameters configuration..............................38

Figure2. 13 : PLMN not allowed .................................................................................39

Figures list

Sofien Jouini - ix - PFE 2006/2007

Figure3. 1 : Power consumption in RBS .....................................................................41

Figure3. 2 : Coverage reduction ..................................................................................43

Figure3. 3 : IRAT-H & CM area moving ....................................................................45

Figure3. 4 : simulation flowchart with Monte Carlo algorithm...................................47

Figure3. 5 : setup common channel power ..................................................................49

Figure3. 6 : Setup HSDPA enabled cells .....................................................................50

Figure3. 7 : Define HSDPA related RABs ..................................................................50

Figure3. 8 : Define WCDMA bearer rate sets .............................................................51

Figure3. 9 : Defining HSDPA capable terminals.........................................................52

Figure3. 10 : Run network analysis .............................................................................53

Figure3. 11 : Impact on coverage; simulation result....................................................54

Figure3. 12 : Top 10 cells coverage.............................................................................55

Figure3. 13 : Downlink maximum delivered power from RBS...................................56

Figure3. 14 : Average CE consumption in downlink ..................................................57

Figure3. 15 : Number of blocked users due to lack of code resources ........................57

Figure3. 16 : Average number of users in CM (per cell).............................................58

Figure3. 17 : UEs in IRAT handover (per cell) ...........................................................58

Figure3. 18 : Call setup Success Rate..........................................................................59

Figure3. 19 : Downlink Noise Rise..............................................................................59

Figure3. 20 : CPICH power increasing........................................................................60

Figure3. 21 : Total RBS power increasing...................................................................61

Figure3. 22 : Increase CPICH power with constant power for CCHs and DCHs .......61

Figure3. 23 : Uplink / downlink out of synchronization..............................................62

Figure3. 24 : Soft handover area moving.....................................................................62

Figure3. 25 : HsPowerMargin parameter.....................................................................63

Tables list

Sofien Jouini - x - PFE 2006/2007

Tables list Table1. 1 : RABs provided by Ericsson in P4 .............................................................18

Table1. 2 : Mapping of UMTS Service to RABs.........................................................19

Table2. 1 : WBN effect from GSM; calculating minimum distance ...........................29

Table2. 2 : WBN effect from GSM; calculating maximum filter size.........................30

Table2. 3 : WBN effect from WCDMA; calculating minimum distance ....................30

Table2. 4 : Initial tuning prerequisites and results. ......................................................32

Table2. 5 : Data analysis..............................................................................................35

Table3. 1 : Coverage reduction calculation .................................................................43

Abbreviations

Sofien Jouini - xi - PFE 2006/2007

Abbreviations

A A-DCH : Associated Dedicated Channel

ARQ : Automatically request

ASE: Air Speech Equivalent

ATM : Asynchronous Transfer Mode

B BLER : Block Error Rate

BSC : Base Station Controller

BTS : Base Transceiver Station

C CCH : Common Channel

CM : Compressed Mode

CN : Core Network

CQI : Channel Quality Indicator

CSR : Cell Selection / Reselection

CSSR : Call Setup Success Rate

CTR : Cell Traffic Recording

D DCH : Dedicated Channel

G GSM : Global System for Mobile

GRAN : GSM Radio Access Network

GGSN : Gateway GPRS support Node

H HDSL : High Speed Digital Subscriber

Line

HSDPA : High Speed Data Packet Access

HS-DSCH : High Speed Downlink

Shared Channel

HS-SCCH : High-Speed Shared Control

Channels

HS-TXB : HSDPA Transmitter Board

I IF : Inter-Frequency

IM : Inter-modulation

IRATH : Inter Radio Access Technology

Handover

ITU-R : International Telecommunication

Union – Radio communication sector

K KPI : Key Performance Indicator

M MGW : Media Gateway

MSC : Mobile Switch Center

MSS : Mobile Soft Switch

Abbreviations

Sofien Jouini - xii - PFE 2006/2007

O O&M : Operating and Maintenance

(O&M)

OSS : Operating Service and System

P PLMN : Public Land Mobile Network

R RAB : Radio Access Bearer

RBS : Radio Base Station

R99 : Release 99

RX : Receiver

S SC : Scrambling Code

SGSN : Server GPRS Support Node

SHO : Soft Handover

T TCPU : TEMS Cell Planner Universal

TTI : Transmission Time Interval

TX : Transmitter

TXB : Transmitter Board

U UARFCN : UTRA Absolute Radio

Frequency Channel Number

UE : User Equipment

UETR : User Equipment Traffic

Recording

UMTS : Universal Mobile Terrestrial

System

URAN : UMTS Radio Access Network

UTRAN : UMTS Terrestrial Radio

Access Network

W WCDMA: Wideband Code Division

Multiple Access

General introduction

Sofien Jouini - 1 - PFE 2006/2007


The increasing demand for wireless data services and continuous growth of

multimedia applications made straightforward the evolution to third generation

networks, named UMTS (Universal Mobile Terrestrial System). Soon after its first

commercial launch in 2002, UMTS has been successfully adopted by wireless

operators to be in service now by 169 operators worldwide, where 69 others are in

planning, deploying or trial phase. This rapid growth of UMTS led to a focus on its

significant evolutionary phase, named HSDPA (High Speed Data Packet Access). As

HSDPA is a simple upgrade to the existing system that results in a significant increase

in data capacity and throughput, 70% of UMTS networks have been upgraded by

HSDPA. [1]

To meet the perception of those operators, optimization is fundamental for

UMTS as any other radio mobile system. However, optimization is much more

complicated with UMTS. In fact;

UMTS is an interfered system based on WCDMA (Wideband Code Division

Multiple Access) access technology that brought a set of new sophisticated algorithms

such as admission / congestion control, inner / outer loop power control, soft / softer

handover and compressed mode.

In addition, operators chose often to reuse 2G sites for 3G antennas deploying,

making cost efficient the evolution toward 3G. This leads to co-existence problems

that result in High degradation of 3G receivers sensitivity and therefore 3G services

quality.

Likewise, to exploit the spectrum (frequencies band) and the remaining resources

from R99 (first release of UMTS) traffic such as codes, power and load, HSDPA is

being deployed in the same carrier with R99 traffic, highly impacting the network

performance such as coverage, capacity and traffic distribution.



Within this context, my graduation project aims to highlight this complexity in

3G networks optimization dealing mainly with 2G–3G neighbours list optimization,

2G-3G antennas isolation, network performance analysis through drive test activity

and HSDPA impact on R99 traffic.

The report is divided into three chapters;

The first chapter depicts the UMTS architecture, interfaces and network

elements functionalities as well as HSDPA features and Ericsson 3G project

proprieties such as coverage, architecture and services.

The second chapter discusses, in first step, the importance of 2G-3G neighbours

list optimization and propose a methodology for this task. In second step, we will deal

with various problems due to co-existence between either (2G and 3G sites) or (3G

and 3G sites) of different manufactures (Ericsson, Alcatel, …). Finally, we will

analyse the network performance through a drive test performed on “on air” sites.

In the last chapter, we will start by a theoretically study of HSDPA impact.

After, we will simulate the HSDPA upgrade of Ericsson network by the planning tool

of Ericsson, TEMS Cell Planner Universal (TCPU). Finally, we will discuss the

efficiency of two strategies of HSDPA deployment.

Chapter 1 Overview to 3G


I. Chapter 1

Overview to 3G

I.1. Introduction

This first chapter is an overview to UMTS networks, migration to R5 and

Ericsson 3G trial project in Tunisia. It depicts the architecture as well as the main

proprieties of Ericsson project such as coverage, capacity and offered services.

Not all UMTS features are discussed here because it is beyond the scope of my

report. However, HSDPA features are depicted in more details to make easier the

understanding of the last chapter (HSDPA impact).

I.2. R99 Networks

I.2.1. Architecture & Interfaces

The Base Transceiver Station (BTS) and Base Station Controller (BSC) in GSM

are replaced respectively by NodeB and Radio Network Controller (RNC) in UMTS.

So, the GSM Radio Access Network (GRAN) is replaced by UMTS Radio Access

Network (URAN) (Figure 1.1). Likewise, this new architecture has brought a set of

new interfaces that follow the GSM naming convention, where applicable;



a) Iu Interface

This interface connects the Core Network (CN) and the URAN. The Iu can have

two different physical instances, Iu-CS and Iu-PS. The Iu-CS connects the radio

access network to a circuit-switched core network, that is, to Mobile Switch Center

(MSC). The Iu-PS connects the access network to a packet-switched core network,

which in practice means a connection to an SGSN (Server GPRS Support Node) [2].

b) Iub Interface

This interface is situated between the RNC and the NodeB in the UTRAN. In

GSM terms this corresponds to the A-bis interface between the BTS and the BSC.

The RNC manages NodeB over the Iub interface. The following functions are

performed over the Iub interface;

Logical Operating and Maintenance (O&M) functions of Node B

System information management

Traffic management of common, dedicated and shared channels

Timing and synchronization management [2]

c) Iur Interface

The Iur interface connects two RNCs. This interface can support the exchange

of both signaling information and user data. All RNCs connected via the Iur must

belong to the same Public Land Mobile Network (PLMN). The Iur interface exists to

support macro-diversity so that the URAN can manage the problem of soft handovers

by itself.

There is always only one RNC in control of a UE connection which is the

Serving RNC (SRNC). Any other RNC involved in the connection is a slave RNC or

a Drift RNC (DRNC). The connection to MSC is routed via the SRNC (figure 1.1) [2]



Figure1. 1 : UMTS networks architecture

I.2.2. Functionalities of RAN Elements

a) NodeB

NodeB is the UMTS equivalent of BTS in GSM and is called often Radio Base

Station (RBS). Functions that are performed by a NodeB include the following:

Transmitting of system information messages according to scheduling

parameters given by the RNC

Macro diversity combining and splitting of data streams internal to

NodeB

Reporting of uplink interference measurements and Downlink power

information

Radio measurements and indication to higher layers

Inner loop power control

RF processing [2]

RNS BSS

CN



b) Radio Network Controller (RNC)

The RNC controls one or more Node Bs. It may be connected via the Iu

interface to an MSC (via Iu-CS) or to an SGSN (via Iu-PS). The interface between

RNCs (Iur) is a logical interface, and a direct physical connection doesn’t necessarily

exist. An RNC is comparable to a BSC in GSM networks.

Functions that are performed by the RNC include the following:

Iub transport resources management

Control of NodeB logical O&M resources

System information management and scheduling of system information

Traffic management of common and shared channels

Modifications to active sets (in soft handover)

Allocation of Downlink channelization codes

Downlink and Uplink outer loop power control

Admission control

Reporting Management [2]

c) RXI

The RAN aggregator, RXI, is perfectly similar to the HUB for local area

networks. In fact, its role results in aggregating the backhaul traffic from a large set of

RBSs depending on its capacity. It can either be co-located with the RNC for port

expansion or be remotely located for regional transport concentration. [3]

I.3. Migration to HSDPA

I.3.1. R99 to R4 to R5 migration

3G network evolution results in CN architecture change and downlink data

throughput improvement. The 3G MSC (in R99) was divided (in R4) into MSC

Server and Media Gateway (MGW) (Figure 1.2). Data throughput reaches 14.4 Mbps

in 3GPP specifications but the implemented version of equipments (terminals) support

only 1.6 Mbps (terminal category 12) and 3.6 Mbps (terminal category 5). (Figure

1.3)



Figure1. 2 : MSC architecture evolution from R99 to R4

Figure1. 3 : Downlink data throughput improvement

I.3.2. HSDPA Definition

The High Speed Data Packet Access (HSDPA) is a downlink channel concept

that employs:

Radio channel quality-dependent fast link adaptation

Hybrid ARQ

A higher modulation scheme of 16 QAM

Radio resources sharing between users in the time and code domains

On the new downlink channel, defined in 3GPP as the HS-DSCH - High Speed

Downlink Shared Channel, the theoretical maximum bit rate that can be achieved

reaches up to 14.4 Mbps. [4]

I.3.3. HSDPA features

HSDPA supports a set of new features that enables higher capacity, reduced

delay and significantly higher data rates than for ordinary Radio Bearers (RBs);

Short Transmission Time Interval (TTI)

Fast radio-dependent scheduling

2000(R99) 2001(R4) 2002(R5) 2003 2004(R6) 2005 2006(R7)

3GPP 1st Specification Version of HSDPA

1st Commercial launch for HSDPA

1st Commercial launch for WCDMA

Downlink peak data rate

384 Kbps 3.6 Mbps

MSC

MSC Server

MGW

In charge of the processing of the user data

In charge of call control and Mobile Management (MM)

R99 R4



High-order modulation

Fast link adaptation

Fast hybrid ARQ with soft combining

Efficient Cell Power Utilization

a) Short Transmission Time Interval (TTI)

One of the main features of HSDPA is the introduction of a shorter TTI in the

WCDMA air interface of just 2 ms. The TTI for HSDPA is short when compared to

DCH, where it is between 10 – 40 ms.

A shorter TTI allows adjusting the properties of the transmission on the HSDPA

downlink channel 500 times per second and has the following advantages:

Fast changing radio channel conditions (mainly due to fading and

multi-path propagations) can be tracked by the radio functions more

accurately.

Scheduling of users and data packets can be realized much more

efficiently, since it is now possible to receive a fast feedback on the

instantaneous radio channel conditions for individual users.

By reducing the round-trip time for packets in the air interface, the

application response time is perceived as improved service quality and

as a higher data throughput for the application in the terminal

equipment.

The short delays are also beneficial to TCP when downloading many

relatively small objects (like a web page), since TCP round trip time is

also reduced.

b) Fast radio-dependent scheduling

Scheduling is the method to determine which UE to transmit at a given time

instant. One of the basic ideas is to transmit to UEs only at fading peaks, thus

improving the C/I conditions for the radio channel and thereby improving the cell

throughput. The consequence of such solution is that the data rate for different users

may vary greatly.

Another method is to give all users the same priority, but this reduces the cell

throughput. In other words there is a trade-off between fairness for the individual user

and cell throughput. In P4 two scheduling algorithms are implemented:



Round Robin scheduling;

Is a simple scheduler giving each user the same amount of radio resources

(TTIs) and does not take into account the possibility to transmit on only fading peaks.

The algorithm is fair for all users from a resource point of view. All users are given

the same amount of radio resources, but the bit rate will vary depending on

momentary radio conditions.

Proportional fair scheduling;

It utilizes information about the fading peaks to prioritize users with good radio

conditions. It also takes delay into account promoting users that have not been given

any data for a long time. In this way, both user fairness and cell throughput is taken

into account (Figure 1.4).

Figure1. 4 : proportional fair scheduling algorithm

c) High-order modulation

HS-DSCH is able to use 16 QAM if the UE category permits, which allows

twice as high data rates to be transmitted as compared to QPSK (which is used for the

DCH).

Since 16 QAM is more sensitive to interference, the channel conditions need to be

good (high C/I). Once the conditions are fulfilled very high data rates can be

accomplished.

Figure1. 5 : QPSK and 16 QAM

High data rate

Low data rate Time

#2 #1 #2 #2 #1 #1 #1

User 2

Scheduled user

16QAM

2 bits 4 bits

QPSK



d) Fast link adaptation

Based on the 2 ms TTI and new feedback channel from the UE to the system for

reporting of the instantaneous radio channel quality CQI (Channel Quality Indicator),

the transmission parameters, such as error correction coding scheme and modulation

scheme, can be adjusted so as to track fast varying radio channel conditions.

Figure1. 6 : fast link adaptation

e) Fast hybrid ARQ with soft combining

The fast changing quality of any radio channel introduces bit errors in data

packets sent between the transmitter and the receiver of a packet transmission.

In traditional error correction schemes for interactive and best effort data

transmissions, the main solution is to automatically request a re-transmission (ARQ)

of the erroneously received packets. Expecting the retransmitted packet to arrive

without bit errors, the previously received erroneous packet is discarded. In HSDPA

both the erroneous packet and the retransmitted packet are soft-combined together by

the error correction algorithm to more efficiently use earlier sent packets and air

interface resources. By deploying Hybrid ARQ with soft combining the air interface

capacity can be increased while still keeping a high robustness of the error correction

schemes.

Figure1. 7 : Fast hybrid ARQ with soft combining

High data rate

Low data rate

NodeB



f) Efficient Cell Power Utilization

Fast link adaptation considers the cell power available for HSDPA downlink

transmissions. Rather than deploying power control for compensating adverse radio

channel conditions, the HSDPA downlink shared channel is rate controlled. This

allows use of all remaining power of a cell for HSDPA transmission after that the R99

traffic demand has been satisfied. Consequently, no urgent requirement exists to

deploy a second or separate carrier for HSDPA.

As the traffic demand on HSDPA channels increases, WCDMA deployment

strategy will be revised due to a high load on the HSDPA channels having an impact

on the existing R99 traffic channel coverage and capacity.

Figure1. 8 : Efficient Cell Power Utilization in HSDPA

I.3.4. HSDPA channels

HSDPA channels consist of the following:

One High-Speed Downlink Shared Channel (HS-DSCH), used for

downlink data transmission,

One High-Speed Shared Control Channels (HS-SCCH), used for

downlink control signaling,

One Associated Dedicated Channel (A-DCH) pair (UL & DL) per

HSDPA user in connected state, used for control signaling and uplink

data transmission.



Figure1. 9 : HSDPA channels

I.3.5. SW/HW upgrade for HSDPA introduction

The upgrade from WCDMA to HSPA requires a new software package and,

potentially, some new pieces of hardware in the base station and in RNC to support

the higher data rates and capacity.

a) RBS

RBS needs to be equipped with HSDPA capable TXB (Transmitter Board) and

new software which is remotely loaded. Dedicated Channel (DCH) and HSDPA share

the same hardware resources and the hardware is separate for downlink and uplink.

This new generation of TXB card, HSTXB, can be configured to meet either R99

traffic only, HSDPA traffic only or a mix of both traffic types. HS-TXB supports up

to 5 codes per cell carrier and up to 16 HSDPA simultaneous users per cell carrier in

P4 (Figure 1.10)

Figure1. 10 : HS-TX board



b) RNC

RNC capacity has to be increased according to the estimated added traffic on

Iub interface, so one or more sub-racks may be added (Figure 1.11). Also a new

software package must be installed to enable HSDPA related algorithms such as

scheduling and ARQ soft combining

Figure1. 11 : Software and hardware upgrade of RNC

I.4. Ericsson 3G Project

I.4.1. Architecture

The network is composed of 61 sites, whose 44 are located in Grand Tunis area,

10 sites in Highway (fast route to Hammamet) and 7 sites in Hammamet city. The

network reaches now its second phase (P2) and is still not fully deployed with nearby

26 sites in pending phase (figure 1.2).

The Operating Service and System-Radio Control (OSS-RC 2.2) will be

upgraded in the next phase (to be RC 3.1) and moved to Hached Centre. The RNC,

RXI, (SGSN), Gateway GPRS support Node (GGSN) and the Mobile Soft Switch

(MSS) are installed in Hached Centre.

The Technopole Centre contains 1 RBS, the OSS-RC container and the mini-

link traffic node that is directly related, by FH, to mini-link traffic node in Marsa

Centre where another RBS are located too.

The transmission lines between different RBSs and the RXI are HDSL (High

Speed Digital Subscriber Line) with Asynchronous Transfer Mode (ATM) Protocol.

Sub-rack



Figure1. 12 : Ericsson 3G network architecture

OSS-RC 3.1Technopole

Minilink Traffic Node

MSS R4.1

RXI 820

RNC 3810

SGSN / GGSN

Marsa minilink TN

Marsa MSC

Hammamet Area

(7RBS)

Highway Area

(10 RBS)

Grand Tunis Area

(43 RBS)

RBS 3100

FH

FH

HDSL

Hached C

enter

Technopole Center

Marsa C

enter

RBS



I.4.2. Coverage

The project was designed to cover three areas; Grand Tunis, Highway and

Hammamet as it are depicted on the following snapshots from TEMS Cell Planner

Universal (TCPU) (Ericsson tool for WCDMA planning). The plots are a prediction

made with TCPU using the RF propagation model “Ericsson 9999” (modified

“Okurama Hata” model for WCDMA networks).

Figure1. 13 : Grand Tunis area coverage

HSDPA



Figure1. 14 : Highway and Hammamet Areas coverage

I.4.3. Services

a) Traffic classes

From end-user and application point of view four major traffic classes can be

identified as illustrated in the following.

Real time applications:

o Streaming class; preserve time relation between entities of the

stream, e.g. Video

o Conversational class; preserve time relation of the entities with

low delay, e.g. Voice.

Non real time applications:

o Background class; destination is not expecting data, preserve

payload, e.g. email

o Interactive class; request and response pattern with preserved

payload, e.g. Internet browsing



b) Radio Access Bearers (RABs)

3GPP has defined a Radio Access Bearer (RAB) as “The service that the access

stratum provides to the non-access stratum to transfer user data between User

Equipment and Core Network”. [5]

In UMTS Terrestrial Radio Access Network (UTRAN) a RAB is defined as the

logical connection between the CN and UE and is used to provide a connection for a

UMTS service via UTRAN as it is describing below.

TETE MTMT WCDMARAN

WCDMARAN CN Iu

edgenode

CN Iuedgenode

CNGateway

CNGateway

TETE

UMTS

End-to-End Service

TE/MT LocalBearer Service

ExternalBearer Service

UMTS Bearer Service

Radio Access Bearer Service CN BearerService

Backbone BearerService

Iu BearerService

Radio BearerService

UTRA FDD/TDDService

PhysicalBearerService

Figure1. 15 : UMTS and Radio Access Bearer Service



The Radio Access Bearers provided by Ericsson are listed in the table1.1. [6]

Radio Access Bearers Description

PS Interactive 64/64

Implemented with a 64 kbps uplink DCH and 64 kbps downlink DCH.


Downlink DCH with 128 kbps.


Downlink DCH with 384 kbps.

PS Interactive 64/HS (HSDPA)

Implemented with an uplink DCH and downlink HS-DSCH. The DCH has the capacity of 64 kbps and the TTI is 20 Ms. the HS-DSCH has the capacity of 3.6 Mbps and the TTI is 2 ms.

PS Interactive 384/HS (HSDPA)

The DCH has the capacity of 384 kbps and the TTI is 10 ms.

PS Streaming 16/64 + PS Interactive 8/8

The multi RAB consists of a streaming (16/64) and interactive RAB (8/8). The streaming one has a guaranteed bit rate of 56 kbps in downlink (64 Kbps as maximum)

PS Streaming 16/128 + PS Interactive 8/8

The streaming RAB has a guaranteed bit rate of 112 Kbps in downlink (with 128 Kbps as Max)

CS Conversational Speech 12.2/12.2 + PS Interactive 64/64

The interactive RAB implemented with 64 Kbps in uplink/downlink

CS Conversational Data 64/64 + PS Interactive 8/8

Same proprieties with a difference in traffic type which is data here and the bit rate as it is 64/64 for conversational and 8/8 for interactive RAB.

Table1. 1 : RABs provided by Ericsson in P4

c) Mapping of 3G services in RABs

The Core Network maps the UMTS service on the Radio Access Bearer

according to the table 1.2.



UMTS Service Type of Radio Access Bearer Speech(AMR codec) + Emergency call Conversational/speech RAB Internet access Interactive or Background PS RAB Modem V.90 Streaming 57.6 Kbps Circuit Switched (CS) RAB H.324M multimedia Conversational 64 Kbps CS RAB SMS Signalling Radio Bearer(SRB) Table1. 2 : Mapping of UMTS Service to RABs

d) Services offered by Ericsson 3G Network in Tunisia

Voice call

o 3G to 3G within Ericsson Network

o 3G to 2G and vice versa (Ericsson – Alcatel)

Video Call

o 3G to 3G within Ericsson Network

Data Services

o MMS

o Video streaming

o Mobile Positioning

o Internet browsing

o E-post cards

o Multiplayer games

I.5. Conclusion

We have seen in this chapter the architecture, interfaces and RAN elements

functionalities of R99 networks as well as the upgrade from R99 to HSDPA and

finally the different proprieties of Ericsson 3G project and requirements. The whole

of this bibliography is important to understand the next chapter where we will explain

the performed tasks during network troubleshooting activity.

Chapter 2 Network Optimization


II. Chapter 2

Network Optimization

II.1. Introduction

Optimize a network is to tune its design and configuration parameters to meet its

predefined target performance. Network optimization can be either before commercial

launch or after. When it is before, it is called “Initial Tuning”.

Several tasks have to be performed before initial tuning activity. We will focus

mainly on the two most critical tasks that haven’t been achieved for Ericsson 3G

project, neighbors’ list optimization and antennas isolation.

The importance of neighbors’ list optimization will be discussed in the first

section of this chapter. The second section depicts the methodology of calculating the

required isolation between antennas and stresses its impact on network performance.

The last section is a description of the performed initial tuning activity for the on air

sites of the studied network.

II.2. Neighbors list optimization

Neighbor list definition is a basic activity in the planning phase to ensure a good

mobility within the radio mobile network. 3G-3G neighbor’s list definition is easy to

perform in our case because of the weak density of 3G sites in the total area (Grand

Tunis, Hammamet). However, 2G-3G neighbor’s list is much more critical because of



the compressed mode algorithm that has a great impact on network performance as it

will be explained in the following paragraphs.

Nevertheless, we can generate such lists automatically by TCPU but the result

needs to be well reviewed and verified (during drive test). The problem is perfectly

like the frequency planning task in GSM that may be performed automatically by

some tools but result is usually inaccurate.

II.2.1. Definitions

a) Compressed mode algorithm

In GSM, the mobile disposes of an idle frame to perform measurements on other

frequencies (26th frame in dedicated mode and 51st frame in idle mode). However, in

WCDMA, the UE transmits continuously and has no possibility to conduct such

measurements. Thus, it is necessary to give a gap of time for the UE to achieve this

task.

The RNC reserves 7 slots within each frame during a period called compressed

mode period (Figure 2.1). This period of time depends on the number of frequencies

that have to be measured. The UE achieves the measurements on one frequency

within 3 slots, and then 2 frequencies may be measured during one compressed frame.

The algorithm that runs in RNC and monitors such function is named as

compressed mode algorithm.

Figure2. 1 : Compressed Mode algorithm impact

1…………15

1

2

3

4

12

13

1………..15

14

15

Gap of 7 slots

Normal frame (SF =16) Normal frame (SF=16)Compressed frame (SF = 8)

UE performs measurement on other frequencies (IF or IRAT handover)

RBS Total power

RNC CPU load

38 dbm 41 dbm 38 dbm

60 % 65 % 60 %

Lost codes = 16 codes of SF = 256



b) Generated list by TCPU

We may generate the 3G-3G and 2G-3G neighbors’ list by TCPU, following the

two steps described in figure 2.2 and figure 2.3.

A short description of the ambiguous parameters of TCPU windows is set below

each figure.

Figure2. 2 : 2G-3G neighbours list generation

1. Define the candidate neighbor cells (for GSM and WCDMA)

2/3. The maximum/minimum neighbors allowed for origin GSM cells

4. GSM handover margin in dB.

5/6. The maximum/minimum length of the neighbor list generated

between cells using different frequencies (GSM and other WCDMA

frequencies). (Max = 32)

7. The minimum signal quality of the pilot required for a target cell

using WCDMA

1

2

3

4

5

6

7

1



Figure2. 3 : neighbours list details

II.2.2. Problem study

When we define a neighbors’ list, we may fall in one of the following cases:

Case1; Setting a low number of 2G neighbors for 3G candidate cells

may lead to high drop or handover failure rate because of the missing

neighbors problem.

Case2; Defining a high number of 2G neighbors may contribute to a

long Compressed Mode (CM) period because of the long time needed

to achieve all measurements.

Long CM period leads to a high power consumption in downlink to keep the

same quality of connection (bit rate, Eb/No…) and therefore a high lost power in RBS

as well as interference in downlink. Likewise, CM algorithm is performed in RNC

consuming a great amount of resources like CPU load and channelization codes.

(Figure2.1)

Defining a low or high number of 2G neighbors is a tradeoff. Thus, the best

method to get an optimized neighbors’ list is to scan the GSM frequencies during

Serving cell

Candidate neighbours

Generated neighbours



drive test activity and rank them for each candidate cell according to their signal

strength. Next, we compare the generated list by TCPU and the scanning result. The

intersection between two lists will be the primarily optimized neighboring cell list (no

more than 6 neighbors as start number) (Figure 2.4). This list may be modified next

time according to drive test result or data recording functions in OSS-RC. For

example, if we record a high drop rate where GSM covers the overall area (like

“Tunis Center”) we must add another set of 2G neighbors in the same way as in the

first. A good number of GSM neighboring cells that we may start with, is no more

than 6 (this takes more than 15 ms in CM).

Figure2. 4 : optimized neighbours’ list

II.2.3. Conclusion

We didn’t get to implement this method because we didn’t dispose of a scanner

when we performed the drive test activity. As an instantaneous remedy, we defined

the 6 strongest neighbors generated by TCPU as neighboring cells for each 3G

candidate cell.

During the drive test activity that we performed, the UE achieved successfully

the IRAT handovers in areas covered by GSM. But this cannot reflect the reliability

of the generated list (by TCPU) because, simply, the drive routes didn’t mach all the

covered area where any UE can experience a bad 3G coverage at the moment when

GSM covers well the area.

List from scanning result

TCPU generated list

Primarily optimized list

Cell 1.1

Cell 2.1

Cell 3.2

Cell 4.2

Cell 5.1

Cell 5.3

Cell 1.3

Cell 1.1

Cell 5.3

Cell 7.1

Cell 1.3

Cell 2.1

Cell 7.2

Cell 8.1

Cell 1.1

Cell 5.3

Cell 1.3

Cell 2.1 < >



II.3. Co-existence problems

Coexistence problems result from either the transmitter or receiver

imperfections. 3GPP and GSM specifications guarantee a minimum performance that

should be respected by both WCDMA and GSM product vendors. However, these

specifications cannot often resolve the problem. Thus, we must isolate the coexisting

antennas by ensuring a sufficient separation distance between them or adding filters in

the critical cases.

In this project, we are interested in the effect of GSM / WCDMA transmitter

(TX) on WCDMA receiver (RX) and Ericsson WCDMA RBS receiver blocking.

Otherwise, we will not discuss the effect of WCDMA TX on GSM RX or GSM RX

blocking because it is beyond the scope of this project.

II.3.1. Definitions

a) Spurious emissions

ITU-R Recommendation M.329-7 defines spurious emissions as “Emission on

frequencies which are outside the necessary bandwidth and the level of which may be

reduced without affecting the corresponding transmission of information.

Spurious emissions include harmonic emissions, parasitic emissions, inter-modulation

products and frequency conversion products but exclude out-of-band emissions.” [7]

Inter-modulation (IM) products:

They are created when two or more frequencies mix in non linear devices in the

transmit path or the receive path. IM products of order n are the sums and differences

in n terms of the original frequencies. The strengths of the IM products decline with

higher orders (we consider only the third order). (Figure 2.5)



Figure2. 5 : Inter-modulation product

Wide Band Noise (WBN) (harmonic or parasitic emission).

Figure2. 6 : Wide Band Noise

b) Receiver blocking

Receiver blocking is the effect of a strong out of band signal, present at the input

of the receiver, on the receiver’s ability to detect an in-band wanted signal. The

blocking signal reduces the specified receiver sensitivity by a certain value of dB. [7]

c) Isolation

Isolation between systems is defined as attenuation between transmitter port in

the interfering system and receiver port (victim). It is the total path loss due to feeder

losses, propagation and attenuation in any extra filter or other devices.

Antennas can be either co-sited or co-area case as it is explained below ( Figure

2.7, Figure2.8).

P1P2

M3

M5M7

M3

M5

Power

Frequency

f1f2 2f1-f2

Frequency

Power

Out of the wanted band emission (WBN)

Wanted band (Normal emission)



Figure2. 7 : Isolation; co-area case

Figure2. 8 : Isolation; co-site case Isolation = ∑ losses between RBS and BTS = Lfi + Lfj + Lt + Lpath - Gai - GASC - Gaj (2.1)

Lpath = 32.4 + 20log (d) +20log (f) (2.2)

Where;

f: the frequency of transmitter

d: the distance between antennas

3G RBS

GSM BTS

ASC gain GASC

Path-Loss (Lpath) Distance > 10m

Gai Gaj

Isolation

RBS port

Feeder Loss Lfi Feeder Loss

Lfj

3G RBS

GSM BTS

Isolation = 30 db Distance < 10m

TMAASC

TMA loss Lt



Ericsson recommends a minimum isolation of 30 db for co-sited antennas to

guarantee its RBS performance. This isolation is achieved by a minimum separation

distance between antennas depending on their proprieties (beamwidth, tilt,

azimuth…).

d) RBS sensitivity degradation

The RBS sensitivity can be expressed as: [5]

RBSSens = Nt + Pint + Nf + 10 log (RUser) + Eb/No [db] (2.3)

Where;

Nt : the thermal noise power density = -174dBm/Hz

Pint : the received level of the external interferer [dBm]

Nf : the noise figure (3 dB with TMA, 4 dB without)

RUser : the user bit rate

Eb/No : the required bit energy above the noise spectral density for

minimum call quality [dB]

RBS sensitivity degradation is expressed as following;

RBS sensitivity degradation = ∆ RBSSens = 10log (1+ Pint / N) [dB] (2.4)

Where;

N = kTBNf [W]

k : Boltzman constant (1.38·10-23 J/K)

T: the thermal noise temperature (290 ºK)

B : the receive bandwidth [Hz]

The maximum allowed sensitivity degradation which corresponds to the noise

rise caused by an external source has to be specified. Ericsson recommends (0.11 db)

as maximum degradation of RBS sensitivity to keep its network performance [8]. This

means that the maximum allowed external power (Pint) due to either spurious

emission or blocking is (-120 dbm / 3.84 MHz).

This value will be taken into consideration during the following study.



II.3.2. Problem study

a) Spurious emission: GSM TX into WCDMA RX

Inter-modulation (IM) effect from DCS 1800;

The Downlink maximum frequency for GSM 1800 is Fmax =1880 MHz, and the

minimum frequency is Fmin = 1805 MHz. Thus, (2Fmax – Fmin) = 1955 MHz.

Ericsson 3G system operates on 2100 band (2130 MHz in uplink) and therefore

there is no problem of inter-modulation (1955 << 2130).

Wide Band Noise effect from GSM 900 and 1800;

GSM specifications limited the spurious emission from GSM BTS as following

[9];

o Co-area case; -62 [dBm / 100 kHz] = -46 [dBm / 3.84 MHz]

o Co-site case; -96 [dBm / 100 kHz] = -80 [dBm / 3.84 MHz]

To keep a certain performance of RBS receiver, Ericsson proposes

[-120dbm/3.84 MHz] as maximum allowed power coming from spurious emission.

o Required Isolation (co-area) = -46 – (-120) = 74 dB

o Required Isolation (co-site) = -80 – (-120) = 40 dB

Co-area isolation:

Considering our case where we have feeders (of 20 m), TMA losses in GSM

sites and no losses in 3G sites due to the using of ASC in uplink.

Isolation = Lfi + Lfj + Lt + Lpath - Gai - GASC - Gaj = Lfi + Lfj + Lt + [32.4 + 20log

(d) + 20log (f)] - Gai - GASC - Gaj (2.5)

Lfi

(db)

Lfj(db) Lt(db) f(MHz) Gai(dbi) GASC(db) Gaj(dbi) d(Km)

GSM900 0.8 1.5 0.2 2130 18 1.5 18 3.174 GSM1800 1.3 1.5 0.2 2130 18 1.5 18 2.996

Table2. 1 : WBN effect from GSM; calculating minimum distance

For the reason that we are studying dense urban areas (Grand Tunis,

Hammamet), we cannot fulfill this condition (d= 2 or 3 Km). Thus we need to

determine the size of filters that must be added to meet the isolation requirements. We



consider d = 1Km (average distance between GSM and 3G sites in “Grand Tunis”

area).

Lfi

(db)

Lfj(db) Lt(db) f(MHz) Gai(dbi) GASC(db) Gaj(dbi) Filters(db)

GSM900 0.8 1.5 0.2 2130 18 1.5 18 10 GSM1800 1.3 1.5 0.2 2130 18 1.5 18 9.5

Table2. 2 : WBN effect from GSM; calculating maximum filter size Co-site isolation:

We guarantee, as I explained previously, 30 db of isolation between co-sited

antennas. As the required isolation is 40 db, we must add filters of 10 db (40-30).

b) Spurious emission: WCDMA TX into WCDMA RX

Some Ericsson 3G sites are co-existed with Alcatel ones and therefore we shall

study this case.

3GPP specifies that the power of any spurious emission (of WCDMA

transmitter) shall not exceed - 96 dBm/100 kHz (= -80 dBm/3.84 MHz). [9]

To ensure -120 dbm as maximum spurious emission at WCDMA RX we need an

isolation of 40 db (-80 – (-120)).

Co-area isolation:

We consider the following values which are the same for GSM case (LASC is

ASC insertion loss in downlink).

Lfi

(db)

Lfj(db) LASC(db) f(MHz) Gai(dbi) GASC(db) Gaj(dbi) d(Km)

WCDMA 1.5 1.5 0.2 2130 18 1.5 18 0.06

Table2. 3 : WBN effect from WCDMA; calculating minimum distance

60 m is the minimum required distance to overcome the spurious emission

problem coming from 3G transmitters. This value (60 m) is easily to reach in reality

(d > 400m in “Grand Tunis” and “Hammamet” areas) and therefore there is no

harmful effect of others 3G sites operating in the same area.



Co-site isolation:

With a guaranteed isolation of 30 db, we must add filters (in WCDMA TX) of

10 db (40-30).

It is important to note that we are studying the worst case. In fact, the WCDMA

RBS performance is better than 3GPP specifications due to high competence between

3G products vendors. Ericsson, for example, designs its RBS to a spurious emission

35 db lower than that required by 3GPP. Thus, if ALCATEL, ZTE and HUWAWI

RBSs were designed with an equivalent performance of Ericsson RBS or at least

guarantee a maximum spurious emission 10 db lower than 3GPP specifications (- 90

dbm/3.84 MHz), we will not need to add filters. However, we have no information

about RBS performance of these vendors (ALCATEL…). Thus, we achieve our

calculations according to3GPP specifications.

c) WCDMA Receiver blocking

According to the WCDMA specifications, the WCDMA RBS has to be designed

to cope with GSM 900 1800 TX signals of up to –15 dBm. [9]

The installed Ericsson RBSs (3000 family) has been designed to cope with

GSM900 and GSM1800 TX signals of up to +20 dBm. If GSM RBS is transmitting at

its maximum power: 25W (44dBm), the isolation needed is 24 dB (44-20). This value

of isolation is guaranteed even in the near field zone of the antenna just with the

coupling loss of the antennas, and the feeder and jumper losses. In case of far field

zone, the propagation loss is enough to fulfill this requirement. Thus, we don’t need

filters.

II.3.3. Conclusion

According to the previous study, it is clear that the coexistence problems result

in spurious emission from either ALCATEL GSM 900/1800 TX or WCDMA TX of

other 3G sites (ALCATEL, HUWAWI and ZTE). Otherwise, there is no harmful

effect of GSM 1800 inter-modulation products or the high transmitted power from 2G

antennas that leads to WCDMA receiver blocking.

To overcome these problems, we recommend the installing of filters of 10 db in

2G / 3G transmitters (not receivers!) co-existing with Ericsson 3G sites.



II.4. Initial tuning

II.4.1. Definition and Purpose

The purpose of the Initial Tuning is to make sure the radio network works well

after the network has been built. The service verifies that the radio network design

and the corresponding network data have been implemented correctly, that the

implemented design is consistent with the proposed design.

Prerequisites and results of Initial tuning activity are given by table 2.2.

INITIAL TUNING

PREREQUISITES ACTIVITIES RESULTS

- Cluster plan completed

- All planned sites

integrated, tested and in

working condition per

cluster

- Radio Network Design

and Network Data

implemented

- Network not in

commercial service

- Preparation

- Radio Network Auditing

- Drive test route plan

- Data Collection

- Post-processing

- Analyzing/Change proposal

Report

- Verification that the

critical items have been

cleared.

- Initial Tuning Analysis

report/Verification report

- Presentation of results

RND Acceptance

Table2. 4 : Initial tuning prerequisites and results.

II.4.2. Process

Figure2. 9 : Initial tuning activity process

Preparation RN audit Post processing Analysis

Fulfil Requirements

Final Report

Change proposal

NO

YES

Data collection



a) Preparation phase

During the preparation phase a couple of activities are carried out such as:

Definition of Clusters; As a start, the total area to be covered according

to service requirements will be divided into small areas called clusters.

Each cluster is aiming to contain 10-15 sites that are located close to

each other. Each cluster can then be separately tuned in different time

frame

Definition of Drive Test Routes; It is essential that the drive test routes

are well planned, excessive duplication of drive routes or missing major

roads as well as driving too much outside the cluster will potentially

confuse the performance statistics. The routes shall be planned so that

soft/softer handover can be observed in important areas and overlapping

between them be as minimum as possible.

Collect radio network design information; The 3G project is wholly

designed and configured by Ericsson. Thus, we dispose of all required

design and configuration information for network audit.

Preparing equipment; The following set of equipments is required for

drive test activity:

o TEMS Scanner (including GPS)

o 2 TEMS UEs (one for long call, one for short call)

o 2 SIM card for the UEs

o TEMS hardware key.

o Data collection PC (PC with TEMS Data Collection)

b) Radio Network (RN) audit

Since all results of the initial tuning are highly depending on a well implemented

network, a radio network audit should be done before starting the initial tuning to

ensure a good result of the service.

The purpose of consistency check is to find inconsistencies in the network and

fix them prior to drive testing. By fixing inconsistencies we save time and speed up

the tuning process. In order to perform the design/consistency check, network

configuration data should be collected through OSS-RC (installed in OSS-HACHED

center).



c) Data collection

Data collection can be retrieved from two different sources: TEMS Investigation

(drive testing) and Traffic Recording (UETR/GPEH/CTR).The General Performance

Event Handling (GPEH), Cell Traffic Recording (CTR) and User Equipment Traffic

Recording (UETR) functions are useful for an advanced problem analysis. In this

activity, we will be closed to TEMS investigation only. Two types of measurements

can be performed:

For scan mode, the PSCH, the SSCH and the CPICH of the sites in the

cluster shall be scanned.

For dedicated mode, two types of calls should be performed for both

speech and video:

o Long calls to evaluate the coverage, quality and retain-ability

performance of the cluster. Long calls will be measured as

continuous calls. As soon as a dropped call occurs a new call

will be placed. Also calls of 10 minutes duration can serve this

purpose.

o Short periodic calls to evaluate the accessibility performance.

The purpose of this test is to ensure that calls can be originated

from all cells on the network and to measure the Call Setup

Success Rate (CSSR) as well as the Call Complete Success Rate

(CCSR). A speech call can be set up every 90-130 seconds, and

there will be a pause of 10 seconds.

d) Post processing

The collected data will be processed in order to simplify analysis and to extract

field measurement performance statistics for reporting. In Ericsson, we dispose of

TEMS Investigation Root Analysis tool 6.0 that provides us with the required

information.

e) Analysis

Among others the following criteria will be analyzed:



RF analysis UE analysis

- SC plan verification

- Coverage

- Pilot pollution

- Neighbors’ list

- Accessibility

- Retain-ability

- Throughput (packet)

- PDP context activation failure

(packet)

- Session error (packet)

Table2. 5 : Data analysis

II.4.3. Encountered problems

During drive test activity, we didn’t face an ambiguous problem. All

encountered problems are classic. Below, we describe briefly each problem.

a) Poor coverage

Problem description: This Drop occurs in region where CPICH RSCP and/or

CPICH Ec/No are measured in critical values not suitable for a proper connection.

TEMS investigation shows a poor coverage with CPICH RSCP = -127 dbm and

Ec/No = -28 db. As it is shown in figure 2.10, the UE experiences a bad radio

condition before the drop call and enters in idle mode after (it receives system

information type 3 which is broadcast in idle mode).



Figure2. 10 : poor coverage

Change proposal: in this case we recommend tilting up the antenna to cover the

hall of coverage (Tunis Center). There is no need to add a new site because of the

short distance between cells.

b) Missing neighbor

Problem description: The drop occurs when the signal quality is bad on the Best

Serving cell with the contemporary possibility for the UE to perform a SHO on a

better cell that is not declared as a Neighbor. The Active Set best server is cell of SC

= 248. During the call, cell of SC = 464 becomes the strongest cell but is not added to

the active set, as it is not defined as neighboring cell (Figure 2.11). The cell of SC =

464 acts as an increasing interferer until eventually the call is released.



Figure2. 11 : Missing neighbour

Change proposal: simply, we have to declare the cell of SC = 464 as

neighboring cell for cell of SC = 248.

c) Pilot pollution and wrong parameters configuration

Problem description: The following snapshot shows 3 problems (figure2.12).

Pilot pollution; we have more than 3 strong cells (active set size = 3)

where the difference between the measured values of Ec/No is less than

5db.

Wrong parameter configuration; the cell of SC = 184 is stronger than

that of SC = 0 (Ec/No [184] = -7 > Ec/No [0] = -14). However, the

serving cell hasn’t been replaced. This due to the parameter “individual

offset” that was set higher than 7 db!

Missing neighbor; it is clear that cells of SC = 280, 424 and 304 have to

be declared as neighboring cells for the serving cell (SC = 0)



Figure2. 12 : Pilot pollution and wrong parameters configuration

Change proposal: we recommend 3 changes.

Setting “Individual offset” value to 0 db (recommended by Ericsson)

Declare cells of SC 280, 424 and 304 as neighboring cells for cell of SC

0

Tilt down the antennas of all these sites to overcome the problem of

pilot pollution

d) Not allowed PLMN

Problem description: The UE tries to camp on other PLMN (UARFCN =

10768) as it shown in Figure 2.13. There is no roaming between Ericsson 3G network

(UARFCN = 10663) and Alcatel 3G network (UARFCN = 10768). This is why the

UE failed in location area update.



Figure2. 13 : PLMN not allowed

II.5. Conclusion

In this chapter, we have seen the different tasks conducted during the

optimization activity of the studied network.

We highlighted the importance of 2G–3G neighbor’s list and its impact on

network performance as well as we proposed a simple method to optimize it.

We have also depicted the coexistence problems such as inter-modulation

products, Wide Band Noise effect and WCDMA receiver blocking. We reached

defining the required isolation between antennas (2G-3G and 3G-3G) and then the

required filters sizes.

Likewise, we have shown the “initial tuning” process and analyzed the

encountered problems during such activity (drive testing). All problems we have

detected through TEMS investigation are classic. Otherwise, there weren’t an

ambiguous reason for any problem.

Chapter 3 HSDPA impact


III. Chapter 3

HSDPA Impact

III.1. Introduction

Evolution to HSDPA is a mandatory step to make the difference between second

and third generation networks. In fact, the maximum theoretically bit rate in 3G is 384

Kbps which is the same given by EDGE.

For operators, HSDPA upgrade is smooth and cost-efficient regarding WCDMA

deployment cost. These reasons make straightforward the upgrade from WCDMA to

HSDPA for the most 3G operators.

Ericsson has prepared its trial network for this step by installing a scalable RBSs

and upgrading its RNC to P4 (corresponds to R5 in 3GPP). However, the network

performance may be highly impacted by HSDPA and we will come back again to

dimensioning, planning and optimization phases.

In this chapter, we will study this perceived impact of HSDPA on Ericsson 3G

network. The study is divided in two parts. In the first part, we study theoretically the

issue. In the second part, we will simulate the HSDPA upgrade to prove the expected

results from first part. At the end of chapter, we will propose two strategies of

HSDPA deployment and discuss their efficiency.



III.2. Impact of HSDPA; theoretical study

III.2.1. Impact on Ec/No values

HSDPA traffic consumes all the remaining power in RBS after serving R99

traffic. It is the best effort traffic regarding the R99 one (Figure 3.1).

Figure3. 1 : Power consumption in RBS

Giving the following expression; Ec/No = PCPICH / (PIntra + PExtra + Noise) (3.1)

Where;

PIntra : the internal power delivered by RBS

PExtra : the external power received by UE from other cells

Noise : the interference caused by environment and other systems

Thus, the increasing of PIntra from 75 % of RBS total power (as maximum

power at antenna for R99 traffic only), to 100 % (full power) in HSDPA case will

reduce the Ec/No value.

If we ignore the term (PExtra+ Noise) against PIntra, we obtain Ec/No = PCPICH /

PIntra. (3.2)

Considering;

X1 = (PCPICH / 0.75*PTot) : the value of Ec/No in R99 only.

X2 = (PCPICH / PTot) : the value of Ec/No in (R99 + HSDPA) case.

PTot : the total power delivered by the RBS.

X2/X1 = 0.75, and thus the difference in Ec/No = 10* log (0.75)

CS traffic

Best effort traffic

Admission Control limit (For R99 traffic)

Best effort traffic (PS)

CS traffic

HSDPA traffic 100%

75%

0%

∆ Ec/No = -1.25 db

RBS power



III.2.2. Impact on coverage

We separate the impact on coverage from that on Ec/No value due to fact that

the cell coverage reduction can be only calculated from the delta of downlink

interference margin and not the difference in Ec/No value we have calculated

previously (-1.25 db).

Following the below procedure to calculate the reduced coverage for a

maximum loaded cell: (Equation (3.3) is the downlink budget of CPICH channel). [5]

Lpmax = PCPICH– SUE – BPC – BIDL – BLNF – LBL – LCPL – LBPL +Ga – LJ (3.3)

Where;

Lpmax : the maximum path loss due to propagation in the air [dB]

PCPICH : CPICH power at antenna [dBm]

SUE : the UE sensitivity [dBm]

BPC : the power control margin [dB]

BLNF : the log-normal fading margin [dB]

BIDL : the noise rise or the downlink interference margin [dB]

LBL : the body loss [dB]

LCPL : the car penetration loss [dB]

LBPL : the building penetration loss [dB]

Ga : the sum of RBS antenna gain and UE antenna gain [dBi]

LJ : the jumper loss [dB]

So, the only term related to RBS transmitted power is BIDL. = 1+ K * Ptot,ref /Lsa

Where;

K = (µ + Fc) / (Nt*Nf*Rchip)

Lsa = Lpmax + BPC + BLNF + LBL + LBPL – Ga + LJ : signal attenuation

µ : the non-orthogonality factor at the cell border

Fc : the ratio between the received inter-cell and intra-cell interference

Ptot,ref : the total transmitted power of RBS at antenna

If Ptot,ref passes from 75% to 100% of to the total RBS power, BIDL increases by

X = K * 0.25* Pnom,ref / Lsa . We calculate the X value in the below table (Table 3.1)

using Ericsson project inputs.



Term µ Fc Nt (dbm/Hz) Nf (db) Rchip K

value 0.64 2.1 -174 7 3.84*10^6 1.75*10^13

Lpmax(db) BPC(db) BLNF(db) LBL(db) LBPL(db) Ga(db) LJ(db) Lsa(db) Lsa(linear)

130 0 4.9 0 18 18 0.2 135.1 3.23*10^13

Pnom(W) Pnom(dbm) LASC(db) Lj (db) Lf(db) Pnom,ref(dbm) Pnom,ref(W) X

17.5 42.43 0.2 0.2 2 40 10 1. 36

Table3. 1 : Coverage reduction calculation

Considering (L1, R1) and (L2, R2) the (path loss, cell range) respectively of

R99 loaded cell (75 % of total power) and HSDPA loaded cell (100%);

L1 = 134 + 35.22log (R1)

L2= 134 + 35.22 log (R2)

The reduction in coverage is calculated through the difference in path loss

values for the 2 cases: L1-L2=35.22 log (R1/R2) = X = 1.36

Figure 3.2 illustrates the cell breathing effect due to a high HSDPA traffic.

`

Figure3. 2 : Coverage reduction

R1/R2 = 12 %

HSDPA enabled cell (Maximum load)

HSDPA disabled cell (Maximum load) R1

R2 12%



III.2.3. Impact on capacity

HSDPA has brought a significant improvement for 3G networks capacity by:

Efficient usage of all the remaining power from R99 traffic.

Good monitoring of system resources by means of its new features such

as fast scheduling algorithm and short TTI.

Increasing the cell throughput as 3 or 4 times as in the R99 case.

However, R99 traffic will be limited as much higher as the HSDPA user’s

number in the cell. In fact;

The power admission threshold for R99 traffic should be decreased to a

value that allows an acceptable throughput for HSDPA users. This has a

direct impact on the blocking and down switching rate of R’99 traffic in

the cell.

Introduction of the HS-DSCH requires all remaining cell power after

serving all R99 users; this is why all HSDPA enabled cells transmit

close to their maximum power limit. This raises the downlink

interference in the cell and leads to a higher blocking rate.

HS-DSCH shares orthogonal code resources with R99 traffic using

codes of spreading factor (SF) 16. The deployed release of HSDPA

may use up to 5 codes with SF 16 which means that 82 codes with SF

256 may be reserved to HSDPA traffic.

HSDPA impacts Ec/No values that trigger the Compressed Mode (CM)

algorithm. This one has an intensive impact on system capacity by

consuming power, channelization codes (in RBS) and CPU load (in

RNC) as twice as the simple mode do.

III.2.4. Impact on traffic distribution

Three Mobility Management (MM) algorithms depend on Ec/No thresholds:

Inter Radio Access Technology Handover (IRAT-H)

Inter-Frequency (IF) handover

Cell Selection / Reselection (CSR)



Such algorithms handle the traffic distribution within radio mobile system;

either between 3G network layers, by means of IF and CSR thresholds, or between

Radio Access Technologies (RATs) (GSM – UMTS), through IRAT and CSR

parameters. The following figure shows an example of traffic distribution change due

to Ec/No varying.

`

Figure3. 3 : IRAT-H & CM area moving

III.3. Practical study; Simulation with TCPU

After explaining how HSDPA impacts R99 traffic, we are going to see such

impact on the studied project of Ericsson.

We keep the same inputs that had been used in the planning phase of Ericsson

3G network (antenna type, tilt, azimuth, propagation model, powers configuration,

RBS type…). However, we will configure 2 HSDPA terminals and enable HSDPA

for all cells by a process that I will depict in details in next paragraphs.

Before starting the simulation on TCPU, we have to get the background

information about TCPU and Monte Carlo algorithm by which TCPU is running.

CM starts

IRAT- H starts-16 db

-12 db

No Coverage

Ec/No values

Area enters in CM

UE moves from UMTS to GSM

Area enters in IRAT-H

Ec/N0=-12db

Ec/No= -18 db

RBS



III.3.1. TCPU and Monte Carlo method

a) TCPU

TEMS Cell Planner Universal is a GSM / WCDMA planning tool that provides

an advanced and accurate analysis with flexible parameter settings to support different

planning methods, with or without Monte Carlo simulations. Depending on the design

stage and level of detail of a network plan, TCPU permits us selecting different levels

of calculation speed and accuracy for the predictions and simulations.

b) Monte Carlo method

“Monte Carlo methods” are a widely used class of computational algorithms for

simulating the behaviour of various physical and mathematical systems, and for other

computations. They are distinguished from other simulation methods by being

stochastic (nondeterministic), usually by using random numbers (in practice, pseudo-

random numbers),as opposed to deterministic algorithms. Because of the repetition of

algorithms and the large number of calculations involved, Monte Carlo is a method

suited to calculation using a computer, utilizing many techniques of computer

simulation. [10]

c) Process of Monte Carlo Simulation in WCDMA Analysis

The Monte Carlo simulator of TEMS Cell Planner Universal is designed to

reflect as closely as possible the UTRAN behaviour in terms of setting up, managing,

and cancelling user connections. It simulates all UTRAN algorithms such as

admission and congestion control. Below is the flowchart of Monte Carlo algorithm

used by TCPU simulations: [11]

http://en.wikipedia.org/wiki/Computation

http://en.wikipedia.org/wiki/Algorithm

http://en.wikipedia.org/wiki/Simulation

http://en.wikipedia.org/wiki/Physics

http://en.wikipedia.org/wiki/Mathematics

http://en.wikipedia.org/wiki/Stochastic

http://en.wikipedia.org/wiki/Nondeterministic

http://en.wikipedia.org/wiki/Random_number

http://en.wikipedia.org/wiki/Pseudo-random_number

http://en.wikipedia.org/wiki/Pseudo-random_number

http://en.wikipedia.org/wiki/Deterministic_algorithm

http://en.wikipedia.org/wiki/Computer

http://en.wikipedia.org/wiki/Computer_simulation

http://en.wikipedia.org/wiki/Computer_simulation



Figure3. 4 : simulation flowchart with Monte Carlo algorithm.

Step1: Generating users

Depending on the traffic demand defined for each WCDMA bearer, users are

generated at random locations. For the whole trial the user distribution is kept

constant. The probability of occupying a certain location in the network (bin) with a

specific traffic (WCDMA bearer) depends on the traffic demand in the network. Over

time, the user distribution is Poisson distributed with a mean number of users equal to

the specified traffic demand.

Step 2: Sorting Cells According to Selected Ranking Algorithm

For each bin occupied with traffic, all cells covering the bin are ranked in

priority for connection attempts. A best server list is generated for each bin. The

selected ranking algorithm defines the order of the cells for which the connection

attempt is made from a specific bin and WCDMA bearer.

WCDMA analysis input parameters

Generate users for all WCDMA bearers

Connect users to cells

Calculate achieved C/I

Modified Tx power in uplink & downlink

Disconnect users (admission & congestion

control)

Converged ?

Output DL power & UL load

Collect statistics for all trials

Generate plots

Generate statistic reports

Resort cells (if applicable)

Num

ber of random trials

Sort cells according to the ranking algorithm

YesNo



Step 3: Connection attempts

Starting with the highest ranked cell, each cell is checked to see if a user can

connect. During the connection attempt the following constraints are tested:

o Downlink CPICH quality

o Required uplink UE TX power

o Required downlink RBS TX power

To meet the criteria, each cell must exceed the minimum threshold by at least

the fading margin for the radio link to be available.

Step 4: Calculate Achieved C/I for All Connected Users

For each mobile, the achieved C/I is calculated based on the uplink and

downlink power settings and the interference known from the previous iteration. For

mobiles in soft or softer handover, maximum ratio combining is performed on the

downlink. For users in softer handover, maximum ratio combining is performed on

the uplink, and for all users in soft handover, selection combining is performed on the

uplink.

Step 5: Modify Tx powers on UL and DL

Over several iterations the transmit power for all cells and all user terminals is

modified to match as closely as possible the achieved C/I to the target C/I. The target

C/I is calculated from the user-defined uplink and downlink Eb/Io values and the

spreading factors used for the respective WCDMA bearer.

Step 6: Disconnect Users - Admission and Congestion Control

The next algorithm checks capacity resources for all cells and disconnects users

that would exceed the user-defined thresholds for the following parameters:

o Maximum number of UL / DL Air Speech Equivalent (ASE)

o Maximum UL interference (noise rise)

o Maximum downlink power limit

o Maximum number of users on spreading factor 8/16/32

When any of these thresholds is exceeded, users are disconnected from cells in

overload based on their QoS criteria and the priority class defined for the WCDMA

bearer. In general, packet switched WCDMA bearers (Interactive and Background

class) are disconnected before the circuit switched WCDMA bearers (Conversational

and streaming class).



Step 7: Convergence check

Once a possible overload situation is resolved for all cells, the system must

verify if the network has achieved a stable state, in which the power changes are

minimal between the iterations. One of the following three criteria can be chosen for

decision;

o UL noise rise (cell based) convergence

o UL noise rise and DL power (cell based) convergence

o UL and DL C/I (user based); System converges both on the

uplink and downlink for each user connection.

III.3.2. Simulation process

This paragraph describes the process of HSDPA configuration in the studied

network. We start by configuring HS-SCCH power, HSDPA RAB (A-DCH),

enabling HSDPA for all cells and so on. The ambiguous parameters are referred to by

a number and brief explanation according to that number is set below each window.

a) Setup common channel power

Figure3. 5 : setup common channel power

1. HS-SCCH (db): difference between HS-SCCH and PCPICH power.

1



b) Setup HSDPA enabled cells

Figure3. 6 : Setup HSDPA enabled cells

1. Maximum orthogonal codes for HSDPA. Refers to SF 256 codes.

2. Maximum HSDPA users allowed in the cell.

3. Maximum users with SF=4 on uplink allowed for the cell.

c) Define HSDPA related RABs

Figure3. 7 : Define HSDPA related RABs

1. In cases of congestion, congestion control disconnects users first on

bearers with low priority. Priority 1 = highest

2. Options include: Background, Conversational, Interactive or

Streaming

3. Options include: Average or Peak

4. Indicates if uplink and downlink rate switching is handled

simultaneously.

1

2

3

1

23

4



d) Define WCDMA Bearer Rate Sets

Figure3. 8 : Define WCDMA bearer rate sets

1. The maximum DL / UP bit rate of the rate set.

2. Used to calculate the interference in DL (UP) generated by the bearer

3. Maximum power that can be allocated for A-DCH ( in DL)

1

2

3



e) Define HSDPA capable terminal

Figure3. 9 : Defining HSDPA capable terminals

1. The sum of losses in reception / transmission such as body loss,

building penetration loss, and feeder loss

2. Maximum power available for terminal type

3. Maximum / Minimum output power of terminal type.

4. Maximum codes with SF = 16 that can be allocated for terminal

5. Shortest time interval for scheduling between users

6. Physical layer throughput (effective throughput)

Category 12

parameters

Category 5

parameters

3

4

5 6



f) Run network analysis

Figure3. 10 : Run network analysis

1. The minimum required pilot channel quality in dB of the HS-SCCH

required by the terminal to detect the HS-SCCH correctly and to be able

to set up a connection on the HSDPA channels.

2. Select one of the 3 defined scheduling methods

3. Defined in comparison with general Round Robin method

III.3.3. Simulation result

a) Impact on coverage

1

2

3



Figure3. 11 : Impact on coverage; simulation result

R99 T

raffic only R

99 + HSD

PA T

raffic



Figure 3.11 shows the studied network both before and after deploying HSDPA.

We depict the two snapshots from TCPU in the same page and with no

separation space or paragraph to make clearer the impact of HSDPA traffic on

network coverage.

For legend, we have chosen to separate Ec/No values into 4 ranges because of

the following;

Ec/No < -8 db: good quality of signal and no of the already discussed

algorithms (IRATH, CM, and CSR) may be triggered in this interval.

-12 db < Ec/No < -8 db: acceptable quality of signal, UE enters in

compressed mode when Ec/No = -12 db

-18 db < Ec/No < -12 db: signal can be decoded with a modest quality.

UE is in compressed mode. UEs existing in this area consume a great

amount of resources (power, codes …)

Ec/No < -18 db: signal cannot be decoded. UE moves to GSM (if it is

possible)

The coverage regression of each area, which corresponds to one of the above

four ranges of Ec/No values, meets our expectations from the theoretical study in the

first part of this chapter. Figure 3.12 depicts the top 10 cell coverage percentile (10

cells among 126). This chart stresses the previous study and the above snapshots

(figure3.11)

0%10%20%30%40%50%60%70%80%90%

100%

Coverage (%)

1 2 3 4 5 6 7 8 9 10

Top 10 Cells

Coverage per Cell

HSDPA R99

Figure3. 12 : Top 10 cells coverage



b) Impact on capacity

Resources that I got to evaluate their consumption during simulation, are RBS

total delivered power (Figure 3.13) and downlink Channel elements (CE)

(Figure3.14). As HS-DSCH uses SF 16 (up to 5 codes per user), the number of

blocked users due to the lack of SF 16 codes reflects well the HSDPA impact

(Figure3.15).

I see that it is not significant to compare the amount of used codes in the two

cases (R99 traffic only, R99 + HSDPA). In fact, it is obvious to say that the number

of users using codes of SF 16 increases with HSDPA deployment because HSDPA

uses codes of SF 16. And thus, it is better to show the blocked users number in each

case.

The difference between the average consumed power in the two cases is about

3db, which means that HSDPA traffic consumes as twice as R99 one (figure 3.13).

According to the Channel Element concept, an increasing in this resource (CE)

consumption reflects an increasing in hardware and software consumption in RBS

(figure 3.14).

Downlink Max Power (dbm)

32

34

36

38

40

42

44

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120All cells

Pow

er (d

b)

HSDPA R99

Figure3. 13 : Downlink maximum delivered power from RBS

3db



Figure3. 14 : Average CE consumption in downlink

Blocked user ( SF = 16)

0

2

4

6

8

10

12

14

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

Top 50 Cells

bloc

ked

user

s

HSDPA R99

Figure3. 15 : Number of blocked users due to lack of code resources

c) Traffic distribution

The number of users in CM increases with HSDPA traffic (figure3.16). The

probability to camp on other layer (frequency) or Radio Access Technology (RAT)

increases as larger as the CM area.

05

10152025303540

CE

1 2 3 4 5 6 7 8 9 10Top 10 Cells

Average Downlink CE

HSDPA R99



0

1

2

3

4

5

6

7

Average users number

1 2 3 4 5 6 7 8 9 10Top 10 Cells

Avearage number of users in CM

HSDPA R99

Figure3. 16 : Average number of users in CM (per cell)

The growing up of UEs in IRAT Handover (due to coverage regression) that is

depicted in figure 3.17 confirms well our expectation (in part one of this chapter). So,

users in cell border experience a bad quality of connection (Ec/No < -16) and attempt

to camp on other layer or GSM. In this simulation, we have not configured a

multilayer system (more than one frequency. Thus, the UE tries directly to camp on

GSM cells (defined in neighbors’ list).

0

2

4

6

8

10

12

Number of UEs

1 2 3 4 5 6 7 8 9 10

Top 10 Cells

UEs in IRAT Handover

HSDPA R99

Figure3. 17 : UEs in IRAT handover (per cell)



d) Quality

The Call Setup Success Rate (CSSR) reflects the subscriber perception about

network availability. Decreasing in this KPI (Key Performance Indicator) means

degradation in network performance and therefore a bad user perception (figure 3.16).

Downlink Noise Rise (NR) degrades the UE receiver sensitivity leading to high

Block Error Rate (BLER) that reduces the downlink data throughput and therefore the

service quality (especially multimedia services) (figure 3.17).

0.00

20.00

40.00

60.00

80.00

100.00

CSSR (%)

Bardo Soukra jamil Phénix Gammarth R-V-TTop 10 Cells

Call Setup Success Rate (%)

HSDPA R99

Figure3. 18 : Call setup Success Rate

00.5

11.5

22.5

33.5

4

NR (db)

1 2 3 4 5 6 7 8 9 10

Top 10 Cells

Downlink Noise Rise (db)

HSDPA R99

Figure3. 19 : Downlink Noise Rise



III.4. Proposal for HSDPA deployment strategy

III.4.1. Proposal 1

Ec/No = CPICH power / PIntra , as we explained previously (paragraph III.2.1).

Thus, we can increase CPICH power by 2db to compensate Ec/ No value degradation

(-1.25 db).

This is a bad solution (at least for Ericsson) as all common and even

dedicated channels powers are configured in proportion of CPICH power (values are

put as margins). Thus, we will fall in one of the two following problems;

Case 1: Increase CPICH power and keep the same values (margins)

for other common and dedicated channels power (Figure 3.19)

Figure3. 20 : CPICH power increasing

As all power values are set relatively to CPICH power, an increase by 2db in

CPICH power will increase all common and dedicated channels power by the same

value (2db). Thus, the total power delivered by an RBS may increase above the

threshold of admission control (set for R99 traffic only and not for HSDPA)

(Figure3.20).

Total power (R99) = [CCH power] + [DCH power]

If each power threshold increases with 2 db the total power threshold increases

by 20 db! Likewise, the reserved power for HSDPA will be highly decreased that a

minimum guaranteed bit rate cannot be achieved.

+ 2db

No changes in m

argins



Figure3. 21 : Total RBS power increasing

Case 2: increase CPICH power and decrease the margins (difference

with DCH and CCH powers) so that only CPICH power will be increased

(Figure 3.21).

Figure3. 22 : Increase CPICH power with constant power for CCHs and DCHs

In this case many problems will occur in network such as out of synchronization

between uplink and downlink and soft handover area change as it explained in figures

3.23 and 3.24.

+ 20 db

Admission Control Threshold (R99 traffic)

DCH Power

DCH Power

CCH Powers

Maximum RBS

power HSDPA Power

HSDPA Power

100%

75 %

RBS power

- 2 db

+ 2 db



Figure3. 23 : Uplink / downlink out of synchronization

Figure3. 24 : Soft handover area moving

Ec/No = -18 db

Eb/No = 4. 8 db (Target value for speech bearer)

Area where UE can access to Speech RAB

Area where UE can camp on cell

Out of synchronization area

Cell A Cell B

Cell ACell BSoft

handover Area

CPICH power = + 2db



III.4.2. Proposal 2

Ericsson defines a power margin (“HsPowerMargin”) that permits us to control

the reserved power for HSDPA traffic and therefore its impact (figure 3.24).

Figure3. 25 : HsPowerMargin parameter

The maximum degradation of Ec/No value is Max (∆ Ec/No) = HSDPA power

(db).

Max (∆ Ec/No) = [ Max RBS Power (100 %) – R99 power threshold (75 %)

– HsPowerMargin ] = [ 0.25 * Max RBS Power – HsPowerMargin ]

If we increase “HsPowerMargin”, we will reduce HSDPA impact but

we will limit the data throughput and capacity in downlink.

If we decrease “HsPowerMargin”, we improve both capacity (number

of HSDPA users in cell) and data throughput but this will impact R99

traffic.

Thus, it is a tradeoff between a low and high value of “HsPowerMargin”.

Unfortunately, this parameter (“HsPowerMargin) is not configured in TCPU and

therefore we cannot simulate its impact to find its optimum value. Therefore, tests

must be performed on field when we deploy HSDPA.

HsPowerMargin

CCH power

DCH power (CS + PS traffic)

HSDPA power

RBS power

R99 power Threshold

HSDPA power Threshold

100%

75%

90%



Ericsson May develop a new feature that permits a dynamic change of

“HsPowerMargin” according to the R99 and HSDPA traffic on cell (like “on demand

bursts” for GPRS). This will be better for good management of network.

III.5. Conclusion

We have studied in this chapter the HSDPA impact on R99 network

performance.

We demonstrated this impact on Ericsson 3G Trial Network in Tunisia through

simulation performed with TCPU. Simulation result has confirmed the theoretical

study.

However, we didn’t get to define an exact strategy of HSDPA deployment with

no impact on R99 traffic. This may be one of the current researches of 3GPP or 3G

products vendors (like Ericsson). In my opinion, the experience on field (live traffic)

will help us to make the decision on what strategy has to be followed to minimize

such impact.

General Conclusion


General Conclusion

We have studied several issues in this project wholly related to Ericsson 3G

network optimization either before deploying HSDPA or after.

In the current phase, HSDPA is not yet deployed in all sites but just three. The

performed tasks within this project are mainly; neighbor’s list optimization, isolation

between co-sited and co-area antennas and drive test performing and analyzing. These

tasks are very important for all radio mobile networks, especially for WCDMA as an

interfered and complicated system. We didn’t encounter sophisticated problems

during the drive test activity; problems are classic such as lack of coverage and pilot

pollution.

For the coming phase, where HSDPA is expected to be deployed in all sites, it is

necessary to study its impact on current network performance to realize how to

proceed to minimize the expected drawbacks. The simulation result that we got

confirms well our theoretically study. Thus, it’s necessary to keep in mind that

HSDPA impacts coverage, capacity and traffic distribution (especially between 2G

and 3G systems). In fact, if no tuning will be performed after HSDPA deployment,

R99 traffic users will experience a high degradation of services quality which is too

bad for Tunisian operator since its first experience in 3G market.

We demonstrated that the increasing of Pilot power is not a remedy for the

studied problem. However, we saw that the power margin variable defined by

Ericsson to control the amount of power reserved for HSDPA traffic can be the

appropriate solution of our issue. We didn’t get to simulate the impact of this

parameter to determinate its optimum value. This for the fact that the TCPU version

we dispose of doesn’t support this feature.

References


References

[1]. “http://3gamericas.org/English/”, Global 3G Status: UMTS and HSDPA

Deployments. 2007 May 23

[2]. Juha Korhonen, “Introduction to 3G Mobile Communications”, Artech House

mobile communications series), Second Edition, ISBN 1-58053-507-0, 2003

[3]. “Ericsson WCDMA RXI Product Description” (Commercial description),

Ericsson AB, 221 01 FGC 101 454 Uen Rev C, 2005-02-11

[4]. Harri Holma, Antti Toskala, “HSDPA / HSUPA for UMTS”, JOHN WILEY &

SONS Ltd, First edition, ISBN 0-470-01884-4, 2006

[5]. “WCDMA Radio Network Design” (Student Book), Ericsson AB, LZU 108 5173

R5A, 03-06-2005

[6]. “Mobile Packet Switched Access for WCDMA and GSM”, Ericsson AB, 14/155

16-HSD 101 13/4 Uen, 12-05-2006

[7]. grouper.ieee.org/groups/802/20/Contribs/C802.20-03-98.doc

[8]. Stevan Filipovic ZG/ND, “Required isolation GSM-UMTS”, Ericsson AB, 02-03-

2001

[9]. “Universal Mobile Telecommunications System (UMTS); UTRA (BS) FDD;

Radio transmission and reception”, ETSI, TS 125 104, V3.11.0, 2002-12

[10]. http://en.wikipedia.org/wiki/Monte_carlo_algorithm

[11]. “TEMS Cell Planner Universal Technical Guideline” (Monte Carlo Simulator),

Ericsson TEMS products, 2004-02-05

http://3gamericas.org/English/

http://www.3gamericas.org/pdfs/Global_3G_Status_Update.pdf

http://www.3gamericas.org/pdfs/Global_3G_Status_Update.pdf

http://en.wikipedia.org/wiki/Monte_carlo_algorithm

3 g tems analysis n doc

Technology

Transcript of 3 g tems analysis n doc