GPRS Plan Guide

GPRS - Radio Network Planning Relevant Aspects

Site

Ludwigsburg

Cellular Operations Department

Originator(s)

A. Grtner / U. Birkel

GPRS - Radio Network Planning Relevant Aspects

Domain:RCD

Division:Operations

Rubric: Radio Network Planning

Type:Guidelines

Distribution codes:

Distribution:

Stuttgart:Vlizy:

C. BrechtmannP. Godet

K. EckertE. Baudienville

R. CollmannC. Fortuit

D. AdolphsL. Carre

K.-D. FraschJ.M. Gabriagues

F. EngmannR. Rouvrais

K. DanielP. Keryer

K. HeinleinJ.P. Jardel

H.-G. TuechleF. Collin

Cc:

M. Albani

J.D. Calvet

Abstract:

This document describes the GSM feature GPRS from the radio network planning point of view. Basics on GPRS are described in the first part of the document, the second part focuses on GPRS engineering guidelines. Note: The document contains industry confidential information.

Keywords:GPRS; packet data; engineering rules;radio network planning; RNP

Approval

Name

C. Brechtmann

K. Eckert

R. Collmann

Date

Signature

Table of Contents

51Change History

2Referenced Documents5

3Scope6

4Abbreviations6

5Basics on GPRS8

5.1Introduction8

5.2Logical Channels, PDCH Multiplexing and dynamic channel allocation10

5.2.1Mapping of logical channels on PDCH10

5.2.2PDCH Multiplexing11

5.2.3TBF Establishment13

5.2.4PDCH Dynamic Allocation15

5.2.5GPRS Network Operation19

5.3RNP relevant GPRS aspects21

5.3.1Channel coding21

5.3.2Measurements21

5.3.3Power Control22

5.3.4Cell re-selection and re-direction instead of handover22

5.3.5Routing Areas23

5.3.6Discontinuous Reception (DRX)24

5.3.7Quality of Service24

5.3.8GPRS Traffic Model25

5.3.9BSS/CAE Parameter25

6GPRS Radio Network Planning26

6.1RNP Design Thresholds based on Reference Performance Point26

6.1.1Reference Performance Point26

6.1.2RNP Design Thresholds29

6.2Initial GPRS Design32

6.2.1GPRS Link Budget and Cell Ranges32

6.2.2Frequency Planning33

6.2.3Traffic dimensioning issues34

6.3Detailed RNP design36

6.3.1Coverage Planning36

6.3.2Interference Analysis36

6.3.3Coding scheme and Throughput density map37

6.3.4GPRS Capacity37

6.4GPRS RNP strategies37

6.4.1Migration to GPRS in case of existing GSM network37

6.4.2Evaluation of capacity gains based on network expansion strategies38

7Summary39

ANNEX A Channel Coding and dynamic coding scheme adaption41

The reference performance point41

Dynamic Coding Scheme Adaptation41

Annex B Dependency of data throughputs of C/I and Level43

Interference43

Sensitivity47

Annex C Quality of Service (QoS)48

Service Precedence (Priority)48

Delay48

Reliability48

Throughput49

ANNEX D Cell Redirection (Cause 25)50

Index51

List Of Tables

21Table 1: Achievable data rates for the Coding Schemes CS 1 to 4

Table 2: GPRS Averaging parameters for uplink and downlink measurements21

Table 3: Reliability Classes24

Table 4: C/I requirement according to ETSI for GPRS carriers29

Table 5: Level requirement according to ETSI for GPRS carriers30

Table 6: GPRS link budgets can be set up similar to CS link budgets32

Table 7: Hata parameters for cell range calculation33

Table 8 : Achievable cell ranges in a coverage driven environment (Hata formula) for TU5033

Table 9: Remaining Erlangs for circuit switched cell traffic (2%Blocking)35

Table 10: C/I for a BLER = 10-1 (including the implementation margin of 2 dB)45

Table 11: Saturation throughput for CS-1 and CS-246

Table 12: Throughput at reference performance point46

Table 13: Signal strength needed for a BLER = 10-1 (TU50, no FH)47

List Of Figures

8Figure 1: New logical entities in the GPRS architecture: SGSN and GGSN

Figure 2: Alcatel GPRS solution9

Figure 3: PDCH Multiplexing11

Figure 4: PDCH Multiplexing of different users on PDCHs12

Figure 5: UL TBF establishment( one phase access on CCCH)13

Figure 6: DL TBF establishment( one phase access on CCCH)14

Figure 7: Coordination of dynamic channel allocation via BSC and MFS A93515

Figure 8: Dynamic allocation of PDCHs per Group: PDCH resource control16

Figure 9: TBF handling per link17

Figure 10: TBF resource allocation17

Figure 11: Parameter and Bitmap example18

Figure 12: Subdivision of a Location Area in GPRS Routing areas (RA)23

Figure 13: Definition of Reference Performance Point27

Figure 14: Dependency of data throughputs as a function of level (no interference)28

Figure 15: GPRS coverage for different coding schemes (schematic)36

Figure 16: Block Error Rate over C/I of the GPRS coding schemes (TU50, no FH)43

Figure 17: Block Error Rate over C/I of the GPRS coding schemes (TU3, no FH)44

Figure 18: Block Error Rate over C/I of the GPRS coding schemes CS-1 and CS244

Figure 19: Maximum throughput in kbyte/s over C/I (TU50, no FH)45

Figure 20: Maximum throughput in kbyte/s over C/I for CS-1 and CS-246

Figure 21: BLER over Eb/N0 of the GPRS coding schemes (TU50, no FH)47

1 Change HistoryXE "change history:in document"

Date

Edition

Status

Author

Comments

March1999

01

Draft

A. Grtner

Document Creation

Aug. 1999

01

Proposal 01

U. Birkel

For internal review

Sept. 1999

01

Release

U. Birkel

2 Referenced DocumentsXE "referenced documents:in document"

[1]

ETSI specification

GSM 03.64 version 6.1.0 Release 1997 "General Packet Radio Service (GPRS); Overall description - radio interface"

[2]

3BK 112030055 DSZZA

"GPRS Traffic XE "Traffic" Model XE "Traffic Model" and Performances, Release B6.2"

[3]

ETSI document

Tdoc SMG2 WPB 99/97 "GPRS simulation results in TU 3 and TU 50, no FH"

[4]

ETSI document

Tdoc SMG2 GPRS 218/97 "Evaluation of Channel Coding Schemes XE "Coding Schemes" CS XE "CS" -2 and CS-4"

[5]

ETSI document

Tdoc SMG2 WPB 100/97 "C/Ic and Eb/N0 Radio Performance for the GPRS Coding Schemes XE "Coding Schemes" "

[6]

ETSI specification

ETSI specification GSM 02.60 version 6.1.1 Release 1997 "General Packet Radio Service (GPRS); Service Description

[7]

ETSI specification

GSM 03.60 version 6.2.0 Release 1997 "General Packet Radio Service (GPRS); Service Description, Stage2"

[8]

ETSI specification

GSM 05.02 version 6.3.0 Release 1997 "Multiplexing and Multiple Access on the Radio Path"

[9]

ETSI specification

GSM 05.05 version 6.3.0 Release 1997 "Radio Transmission and Reception"

[10]

3BK 10204 0458 DTZZA

GPRS BSS Technical Feature List

[11]

GPRS general pres. (B6.2)

http://aww.mcd.alcatel.com/mcd/technic/index.htm

[12]

3BK 11202 0256 DSZZA

GPRS Telecom Presentation

[13]

NEM Meeting 7/99 Velizy

Presentation of F. Collin: GPRS Parameter Tuning

[14]

3BK 11202 0065 DSZZA

Power Control XE "Power Control" and Handover Algorithms

[15]

3BK 09417 FCAA PBZZA

RNO/RNP CMA Logical Parameter File Interface B6.2

[16]

3BK ?not available

GPRS Dimensioning

[17]

3BK 25021 0001 UZZZA

Product Sheet Release B 6.2

3 ScopeXE "Scope"

The document describes GPRS Radio Network Planning relevant aspects. It is the scope of the document to provide an RNP engineering guideline for GPRS networks.

Chapter 5 gives an overview on GPRS, focussing on the PDCH XE "PDCH" handling and user multiplexing as well as on basic GPRS features, which are interesting from the RNP point of view.

GPRS RNP engineering rules are specified in Chapter 6 by defining the design figures and by describing the methodology for initial and detailed GPRS radio network designs.

4 AbbreviationsXE "abbreviations:in document"

ARQ

Automatic Retransmission Request

BCCH

Broadcast Control Channel

BCS

Block Check Sequence

BER

Bit Error Rate

BLER XE "BLER"

Block Error Rate XE "Block Error Rate"

BSC

Base Station Controller

BSS

Base Station System

BSSGP

GP BSS GPRS Protocol

BTS

Base Transceiver Station

BVC

BSSGP Virtual Connection

CCCH

Common Control Channel

CS XE "CS"

Circuit-Switched (By extension includes all no GPRS links)

CS XE "CS" -x

Coding Scheme x (x = 1, 2, 3, 4)

DL

Down link

DRX XE "DRX"

Discontinuous Reception

DSP

Digital Signal Processor

FEC

Forward Error Correction

FH

Frequency Hopping

Gb

Telecommunication interface between BSS and SGSN XE "SGSN"

GGSN XE "GGSN"

Gateway GPRS Support Node

Gi

Telecommunication interface between a GGSN XE "GGSN" and a PDN.

Gn

Telecommunication interface between SGSNs and between SGSN XE "SGSN" + GGSN XE "GGSN" .

GPRS

General Packet Radio Service

GPU

GPRS Processing Unit

Gs

Telecommunication interface between SGSN XE "SGSN" and MSC/VLR

GSM

Global System for Mobile communication

HPLMN

Home PLMN

IMSI

International Mobile Subscriber Identiy

IP

Internet Protocol

ISDN

Integrated Service Digital Network

LA

Location Area

LLC

Logical Link Control

MAC

Medium Access Control

MFS XE "MFS"

Multi-BSS Fast packet Server

MO

Mobile Originated

MS

Mobile Station

MSC

Mobile Services Switching Centre

MT

Mobile Terminating

NCx

Network Controlled cell re-selection mode x (x = 0, 1, 2)

NMC

Network Management Centre

NS

Network Service

NSS

Network Switching System

NS-VC

Netwok Service Virtual Connection

O&M

Operation and Maintenance

OAM

Operation, Administration and Maintenance

OMC-R

Operation and Maintenance Centre - BSS

PACCH

Packet Associated Control Channel

PCH

Paging Channel

PCU

Packet Control Unit

PDCH XE "PDCH"

Packet Data Channel

PDP

Packet Data Protocol

PDU

Protocol Data Unit

PLMN

Public Land Mobile Network

PPCH

Packet Paging Channel

PRACH

Packet Random Access Channel

PS

Packet-Switched

PSPDN

Public Switched Packet Data Network

PTM

Point-To-Multipoint

PTM-G

Point-To-Multipoint Group

PTM-M

Point-To-Multipoint Multicast

PTP

Point-To-Point

PVC

Permanent Virtual Connection

QoS

Quality of Service XE "Quality of Service"

RA

Routing Area

RACH

Random Access Channel

RLC

Radio Link Control

SDCCH

Stand-alone Dedicated Control Channel

SDU

Service Data Unit

SGSN XE "SGSN"

Serving GPRS Support Node

SMS

Short Message Service

SPDCH

Slave Packet Data Channel

TBF XE "TBF"

Temporary Block Flow

TCP

Transport Control Protocol

TDMA

Time Division Multiple Access

TLLI

Temporary Logical Link Identity

TRX

Transceiver

TS

Timeslot

UL

Uplink

USF

Uplink State Flag

VLR

Visitor Location Register

Basics on GPRS

4.1 Introduction

GSM is based on a circuit switched (CS XE "CS" ) concept, which means that a connection is established between two terminals, that is maintained during the whole session, reserving the channel resource exclusively. GPRS (General Packet Radio Service) has been introduced as a new GSM feature to provide end-to-end packet-switched data transmission between MS users and fixed packet data networks. With GPRS, the radio interface resources are allocated dynamically; the transmission rate can be varied. The available channel resource is shared by multiple connections, which means that a user only occupies network resources when data packets are transmitted (bursty traffic nature). The mobile users are multiplexed together on one or on several radio TS reserved for GPRS. This is very efficient if only data bursts have to be transmitted and not continuous data streams.

Packet switched services can be subdivided into connection oriented services and connectionless services which are also called datagram services. In the latter case, each packet contains the complete destination and originating address and it passes through the network independently from other packets, so that packets between two communication partners can take different paths through the network and even overtake each other. So, the routing of the data packets is possible without a connection establishment. For connection oriented services, the path through the network is given explicitly for the duration of the virtual connection. The set-up and release of the virtual connection needs a certain signalling and switching effort. Both the GSM radio part and the network architecture are affected by the implementation of GPRS. As shown in Figure 1, the new backbone network architecture comprises two main elements:

the SGSN XE "SGSN" (Serving GPRS Support Node) which manages mobility (keep track of MS location), security functions (encryption), data compression, access control and charging.

the GGSN XE "GGSN" (Gateway GPRS Support Node) provides interaction with the external packet data networks.

BSS

SGSN

GGSN

TE

GPRS

MS

MSC/VLR

HLR

Data

Signalling

ISDN, PSTN

GPRS Backbone

INTERNET,

X25

Figure 1: New logical entities in the GPRS architecture: SGSN XE "SGSN" and GGSN XE "GGSN"

The Alcatel solution consists of a software upgrade in the BSS and a new hardware element called Alcatel 935 MFS XE "MFS" (Multi-BSS Fast packet Server), which is typically located beside the transcoder at the MSC site and can be shared by several BSSs as shown in Figure 2. In addition, it comprises a backbone network architecture containing the mentioned GPRS Support Nodes.

MSC

BTS

BSC

Abis

Abis

OMC-R

MFS

(PCU)

Ater

TC

SGSN

A

Gb

Gb

Figure 2: Alcatel GPRS solution

Alcatels GPRS development program schedule is:

Step 1 end of 1999 shall be in line with SMG 28 (Release B 6.2): This first GPRS implementation step includes GPRS ETSI phase 1 features, which are absolutely needed for an operator to provide GPRS service. It is a short term solution for a first commercial introduction of GPRS. This solution is foreseen to work with prototype GPRS MS.

Step 1 shall be in line with SMG 29 around mid 2000 (Release B 6.2 M): Additional interesting features, which are needed by an operator in a service deployment phase will be incorporated (e.g. Master PDCH XE "PDCH" ). This solution is then foreseen to work with commercial GPRS Mobiles and is not compatible with SMG 28.

For more detailed information on the technical features of the Alcatel GPRS implementation please refer to [10], [17].

4.2 Logical Channels XE "Logical Channels" , PDCH XE "PDCH" Multiplexing XE "PDCH Multiplexing" and dynamic channel allocation XE "dynamic channel allocation"

In implementation step 1 only one TRX per cell can be dedicated to GPRS service. On this carrier, any time slot can be configured to allow the transmission of data packets between the mobile station and the network. This physical channel or timeslot is then called a PDCH XE "PDCH" (Packet Data Channel).

Per TRX the available time slots are dynamically shared by circuit switched (CS XE "CS" ) and GPRS traffic, i.e. n timeslots can be used for CS and the remaining 8-n timeslots for GPRS, while n is variable between min. 0 and max. 8 (dynamic channel allocation XE "dynamic channel allocation" ).

The packets of different users can be transmitted on one PDCH XE "PDCH" (user multiplexing), furthermore one user can transmit packages on several PDCHs (multislot capability XE "multislot capability" ).

Data packets are transmitted between the MS and the NW on a PDCH XE "PDCH" via a Temporary Block Flow XE "Temporary Block Flow" (TBF XE "TBF" ).

Before the PDCH XE "PDCH" multiplexing and the TBF XE "TBF" establishment XE "TBF establishment" is described in more detail, the mapping of the physical channel on logical channels is explained in the next section.

4.2.1 Mapping of logical channels on PDCH XE "PDCH"

The packet data logical channels are mapped on a PDCH as follows:

PCCCH (Packet Common Control CHannel) used to initiate packet transfer

PRACH (Packet Random Access CHannel)

PPCH (Packet Paging CHannel)

PAGCH (Packet Access Grant CHannel)

PBCCH (Packet Broadcast Control CHannel) used for broadcasting system information

PTCH (Packet Traffic XE "Traffic" CHannel) used for data transmission and associated signalling

PDTCH (Packet Data Traffic XE "Traffic" CHannel)

PACCH (Packet Associated Control CHannel) mapped on one PDCH XE "PDCH" and is dynamically assigned to one of the PDCHs which are assigned to the corresponding MS.

PTCCH (Packet Timing Advance Control CHannel) used for continuous timing advance mechanism. It is mapped on one of the PDCH XE "PDCH" carrying the PACCH of that MS.

A PDCH XE "PDCH" is called GPRS Master Channel (MPDCH) when it carries a PBCCH and a PCCCH channel. If a MPDCH is available or not and on which TS it is configured, is broadcasted on the BCCH. If no MPDCH is available the BCCH and the CCCH of the circuit switched design are used instead. The usage of a MPDCH has thus the following advantages:

Reduced BSC and CCCH load and thus improved access delay

On the PBCCH the GPRS specific system information is broadcasted, including a list of GPRS neighbourcells. Otherwise (no MPDCH available) the neighbours specified for CS XE "CS" traffic will also be used for GPRS cell reselection, which means that target cells which do not support GPRS might be selected. Thus the usage of a MPDCH is recommended.

If no MPDCH is supported (step 1 implementation) the following log. channels are used:

CCCH (Common Control CHannel) used to initiate packet transfer with

RACH, PCH and AGCH

BCCH (Broadcast Control CHannel) used for broadcasting system information

PTCH and PTCCH as mentioned above

Note: In some documents the SPDCH (Slave PDCH XE "PDCH" ) is used to identify PDCHs which are no MPDCH (Master PDCHs).

4.2.2 PDCH XE "PDCH" Multiplexing XE "PDCH Multiplexing"

As for CS XE "CS" traffic, the access scheme is Time Division Multiple Access (TDMA), with 8 physical channels (= timeslots) per carrier. A timeslot allocated for GPRS is called a Packet Data Channel (PDCH XE "PDCH" ). The sharing of PDCHs is based on blocks of 4 consecutive TDMA frames. Several users can be multiplexed block wise on one PDCH (E.g. on PDCH 1 = TS 0 as shown in Figure 3). On the other hand one user can occupy several PDCHs, depending on its multislot capability.

Timeslots usable for PDCHs are grouped into PDCH XE "PDCH" groups. The maximum amount of PDCHs per group is accordingly 8. In step1 there is only one PDCH group available (since only 1 TRX per cell).

As shown in Figure 3 twelve blocks (B0 to B11) form a 52-multiframe. The frames 25 and 51 are idle frames and the frames 12 and 38 are used for the PTCCH ( Total 4*12 + 4 = 52 frames in 240 ms. 1 block includes 4 TDMA frames (=18.462ms). One PDCH XE "PDCH" in one block occupies 4 timeslots with 114 bits each (= 456 bits per RLC block). Thus the brutto throughput XE "throughput" rate is 456 bit/18.462ms = 24.7 kbit/s per timeslot. However the effective data throughput rate is lower since bits for USF, BSC and header are used (see chapter 6.5.5.1 in [1]).

Figure 3: PDCH XE "PDCH" Multiplexing XE "PDCH Multiplexing"

Depending on the coding scheme different effective data throughput XE "throughput" rates per timeslot can be achieved (( 20kbit/s/timeslot see chapter 5.3.1).

For packet data transfer, the data is transmitted on a Temporary Block Flow (TBF XE "TBF" ). For a TBF one or more PDCHs are allocated and it comprises a number of blocks. A TBF is temporary and is maintained only for the duration of the data transfer. Each TBF is unidirectional, i.e. uplink and downlink TBFs for the same link are uncorrelated. The TBF of each user is identified by a Temporary Flow Identity (TFI).

The maximum amount of PDCHs per TBF XE "TBF" is limited by O&M or by the mobiles multislot capability, whichever is lower. The maximum amount of TBFs per PDCH XE "PDCH" is also limited by O&M (user multiplexing).

As shown in Figure 4 the TBFs with TFI = 24 and 17 are using several PDCHs, whereas the user with TFI = 5 does not support the multislot capability and therefore is only allocated on one PDCH XE "PDCH" . Further the three TBFs are multiplexed block wise on the allocated PDCHs.

This packet oriented approach allows optimum usage of the available radio resource.

Figure 4: PDCH XE "PDCH" Multiplexing XE "PDCH Multiplexing" of different users on PDCHs

The control of the multiplexing of different MSs on an uplink PDCH XE "PDCH" uses the USF (Uplink State Flag) mechanism. The USF is a token which is distributed by the network at UL TBF XE "TBF" establishment XE "TBF establishment" (one USF per allocated PDCH). The uplink multiplexing is scheduled by USF values included in the header of each RLC downlink block. The USF value in downlink block Bn schedules the uplink block Bn+1, i.e. MS which has been allocated this USF, can use Bn+1 either as a PDTCH or a PACCH. On the master PDCH, a specific USF value is reserved (USF = FREE) to schedule a PRACH. Another USF value is reserved to schedule a block for PACCH related to a downlink TBF.

TBF XE "TBF" Establishment

The establishment of a TBF XE "TBF" can be initiated either by the MS or by the network [11]:

UL TBF XE "TBF" establishment XE "TBF establishment"

Packet access can be done in either one phase or 2 phases. 2 phase access is necessary to request a RLC unacknowledged mode and to send the MS multi-slot class, when access is on CCCH. In this document only the 1 phase access will be explained.

The packet access uses either the PCCCH (if MPDCH available) or the CCCH

The establishment can also be done on PACCH if a DL TBF XE "TBF" is on-going

The UL TBF XE "TBF" establishment XE "TBF establishment" scenario in Table 5 explains the example of the one phase access on the CCCH (for other scenarios see [11]).

Figure 5: UL TBF XE "TBF" establishment XE "TBF establishment" ( one phase access on CCCH)

(1) The Channel Request is received on the RACH and indicates one phase access. In case the request can be satisfied, an Immediate Assignment message is sent to the MS with a TFI, one allocated PDCH XE "PDCH" with its USF, the initial timing advance value (calculated on reception of the Packet Channel Request) and the Timing advance Index (to be used for continuous timing advance index). A timer is activated to give time to the MS to receive this command.

(2) At timer expiry a Packet UL Assignment message is sent to MS, assigning the same resources as those assigned previously, but without the initial timing advance value. Then the network forces the MS to send a Packet Control Acknowledgement (polling indication) to be sure that the UL TBF XE "TBF" has been successfully established.

(3) Then the MS listens to the allocated PDCHs to detect its USF. On reception of the Packet Control Ack, the network begins to schedule UL blocks, with the USF mechanism

(4) The MS transmits UL blocks when allowed by the network. The MS shall provide its TLLI in RLC data blocks, until the end of the contention resolution (i.e. reception of the Packet UL ack/Nack with its TLLI)

(5) The network acknowledges as soon as one of these blocks is correctly received (i.e. the MS using the TBF XE "TBF" is non-ambiguously identified)

DL TBF XE "TBF" establishment XE "TBF establishment"

The procedure may be entered either when the MS is in packet idle mode (access on PCCCH or CCCH) or when the MS is in packet transfer mode (i.e. an UL TBF XE "TBF" is already established, then the access is performed on the PACCH).

The DL TBF XE "TBF" establishment XE "TBF establishment" scenario is explained at the example of the one phase access on the CCCH (for other scenarios see [11]).

Figure 6: DL TBF XE "TBF" establishment XE "TBF establishment" ( one phase access on CCCH)

(1) An Immediate assignment is sent with a TFI, one PDCH XE "PDCH" and a Timing Advance Index (no initial timing advance value is provided). A timer is activated to give time to the MS to take into account this message. At timer expiry, a packet DL Assignment message is sent with the TFI, PDCHs (additional PDCHs may be allocated, since only one PDCH can be allocated when using CCCH) and the timing advance index.

(2) The network forces the MS to acknowledge to be sure that the DL TBF XE "TBF" has been successfully established and to be able to compute an initial timing advance value.

(3) The initial timing advance value is sent to the MS

(4) Then, data transfer begins

4.2.3 PDCH XE "PDCH" Dynamic Allocation

The PDCH XE "PDCH" Dynamic allocation is used for the coordination between circuit switched and GPRS traffic within one TRX (PDCH group) for radio resource handling. This coordination is performed through the Alcatel BSCGP interface between BSC and A935 XE "A935" MFS XE "MFS" as shown in Figure 7. PDCH (de)-allocation requests are sent from the MFS to the BSC and BSC CS XE "CS" traffic load is indicated from the BSC to the MFS.

Figure 7: Coordination of dynamic channel allocation XE "dynamic channel allocation" via BSC and MFS XE "MFS" A935 XE "A935"

Timeslots usable for PDCH XE "PDCH" are grouped into PDCH groups (only one in step 1). One PDCH group contains time-slots belonging to the same TRX, having the same frequency configuration without holes (=consecutive timeslots).

4.2.3.1 PDCH XE "PDCH" resource control

As the circuit switched services should be served with priority, a parameter defining the maximal number of PDCHs within one carrier/group can be set (MAX_PDCH XE "PDCH" _GROUP, see below). When the maximal number of time slots is reached, no more time slots can be allocated to packet data traffic. In case of high BSC high traffic load, this maximum value is automatically reduced to MAX_PDCH_HIGH_LOAD, so that there are enough timeslots reserved for circuit switched traffic. The following parameters are tuneable in order to achieve an optimum resource utilisation and can be set via O&M:

MIN_PDCH XE "PDCH" _GROUPMinimum number of PDCHs per cell which can be activated for GPRS (Range 0 to 8, Default = 0)

MAX_PDCH XE "PDCH" _GROUPMaximum number of PDCHs allocated in a PDCH group in case of normal BSC load (Range 1 to 8, Default = 8)

MAX_PDCH XE "PDCH" _HIGH_LOADMaximum number of PDCHs allocated in a PDCH group when the BSC has indicated a high BSC traffic load (Range 0 to 8, Default = 1)

Figure 8 shows the dynamic allocation algorithms of PDCHs. One can observe that timeslots are allocated by the BSC on MFS XE "MFS" request (and deallocated, if no longer needed). If the maximum number of PDCHs is reached, new requests are rejected for that group. If the CS XE "CS" load situation changes from normal load to high load, the MFS limits the number of allocated PDCHs to MAX_PDCH XE "PDCH" _HIGH_LOAD. The MFS deallocates PDCHs as soon as exceeding PDCHs become empty and marks them as not usable for new TBFs (BSC soft pre-emption).

time

MIN_PDCH_GROUP

MAX_PDCH_HIGH_LOAD

MAX_PDCH_GROUP

Allocated PDCHs

Cell is created

A GPRS MS requests

4 timeslots

Normal BSC load

Normal BSC load

High BSC load

Maximum number of

PDCHs is reached

High BSC load: the MFS

deallocates PDCHs as soon

as exceeding PDCHs

become empty.

Maximum number of

PDCHs in BSC high load

situation

Figure 8: Dynamic allocation of PDCHs per Group: PDCH XE "PDCH" resource control

4.2.3.2 TBF XE "TBF" resource allocation:

The TBF XE "TBF" resource allocation is used to coordinate the allocated amount of TBFs per PDCH XE "PDCH" , requesting additional resources/TBFs (if available in the PDCH group) in case of busy GPRS traffic. O&M parameters define thresholds used to invoke allocation of new PDCHs according to the algorithm as described in this section.

According to Figure 9 the TBF XE "TBF" resource allocation is controlled by the packet switched traffic occurrence per PDCH XE "PDCH" . The following states are distinguished: Empty,Active, Busy and Full. The according thresholds can be adjusted via O&M parameters.

N_TBF XE "TBF" _PER_PDCH XE "PDCH" :Threshold defining the number of TBF supported by a PDCH (UL or DL), above which, the MFS XE "MFS" serves new TBF preferably by requesting additional resources to the BSC. Range: [1;16], Default value : 2

MAX_UL_TBF XE "TBF" _PDCH XE "PDCH" Maximum number of UL TBFs per PDCH. Range: [1;7], Default value: 7

MAX_DL_TBF XE "TBF" _PDCH XE "PDCH" Maximum number of DL TBFs per PDCH. Range: [1;9],.Default value: 9.

MAX_PDCH XE "PDCH" _PER_TBF XE "TBF" :Maximum number of PDCHs which can be allocated to a single TBF. Range: [1;5], Default value: 5 ( Theoretical max. bitrate per MS in Alcatel step 1 implementation (CS XE "CS" -2): 5*12kbit/s = 60kbit/s/user. However the bit rate per user is also limited by the multislot capacity of the mobiles (see chapter 5.2.4.4).

Figure 9: TBF XE "TBF" handling per link

The TBF XE "TBF" resource allocation algorithm is performed in 4 steps as shown in Figure 10, within the MFS XE "MFS" and is triggered by a TBF establishment XE "TBF establishment" request as shown in figure 7.

Figure 10: TBF XE "TBF" resource allocation

Step1: The multi-slot class of the MS (when known) is interpreted and the O&M parameter MAX_PDCH XE "PDCH" _PER_TBF XE "TBF" . Based on this the maximum applicable amount of PDCHs for this TBF request can be determined.

Step2: The TBF XE "TBF" establishment XE "TBF establishment" on one direction shall comply with constraints imposed by potential concurrent TBF on the other direction

Step 3: To select a PDCH XE "PDCH" the following command is applied

If n_consecutive_pdch ( n_MS_requested then the already allocated PDCHs can serve the request

If n_allocated_pdch= already allocated PDCHs can be used

n_allocated_pdch

the total number of already allocated PDCHs

n_MS_requested

number of PDCHs determined in step2

Figure 11 gives an example of the mentioned parameters (left) and a bitmap (right).

Figure 11: Parameter and Bitmap example

Step 4: (after BSC PDCH XE "PDCH" allocation)

If n_consecutive_pdch = n_MS_requested the request is served on the n_consecutive_pdch

If n_consecutive_pdch > 0 AND n_consecutive_pdch < n_MS_requested then the request is served on n_consecutive_pdch and the MFS XE "MFS" tries to extend the selected combination by using adjacent PDCHs in the busy state.

If n_consecutive_pdch = 0 and at least one PDCH XE "PDCH" with less than MAX_UL/DL_TBF XE "TBF" _PDCH TBFs the request is served on the usable PDCHs

If n_consecutive_pdch = 0 and no other usable PDCHs the UL request is rejected and the DL request is queued during t < PDU_Life_time

4.2.3.3 BSC PDCH XE "PDCH" allocation Principles

The aim of the BSC is to keep one set of free TSs on the PDCH XE "PDCH" group chosen for GPRS support. If no time-slot is already allocated for GPRS on the PDCH group, the time-slots on the centre of this PDCH group are seen as the set of channels to keep free (i.e. with the lowest priority, for CS XE "CS" traffic). If some TS have been allocated, the adjacent time-slots are seen as the channels to keep free.

4.2.3.4 MS CAPABILITIES

In order to evaluate bitrates per user, the resource allocation has also to be considered from the MS point of view. Different MS classes and types are defined for GPRS:

First of all three classes of GPRS MSs are supported:

Class-A MS can operate GPRS and other GSM services simultaneously.

Class-B MS can monitor control channel for GPRS and other GSM services simultaneously, but can only operate one set of services at one time.

Class-C GPRS MS can exclusively operate GPRS services.

The maximum number of PDCHs granted to a MS depends in its multi-slot class. The MS constraints determining the maximum capacity per user are the following:

-MS type

type 1: simplex MS (either transmit or receive)

type 2: duplex MS (receive and transmit simultaneously)

-Max. number of receive/transmit timeslots per TDMA frame. Concerning the multislot capacity, in the year 2000 Alcatel plans to provide mobiles supporting only 1 TS for the UL and 1 TS for the DL, Sagem will probably offer 3+1 (3 DL and 1 UL).

-Min. time (in timeslots) between receive, transmit and measurements

These values are defined in [12] and in [8] in more detail.

4.2.4 GPRS Network Operation4.2.4.1 GPRS ATTACH function

The GPRS attach function is performed according to the following steps and is similar to CS XE "CS" IMSI Attach: Authenticates the MS, Generates the Ciphering Key, Allocates TLLI, Copy subscriber profile from HLR to SGSN XE "SGSN"

After Attach: The location of the MS is tracked. Communication between MS and SGSN XE "SGSN" is secured. Charging information is collected. SGSN knows what the subscriber is allowed to do. HLR knows the location of the MS in the accuracy of the SGSN

4.2.4.2 PDP Context Activation

No data transmission is possible before the PDP address is activated. A user may have several PDP addresses and can activate each of the subscribed PDP addresses separately.

PDP Activation procedure: MS sends request to SGSN XE "SGSN" , SGSN informs GGSN XE "GGSN" , the GGSN creates a context and sends acknowledge to SGSN, SGSN sends acknowledge to MS ( Data transmission is possible

When PDP context is activated, the SGSN XE "SGSN" has a logical tunnel between MS and GGSN XE "GGSN" and MS can send data packets as well as computers in external data networks can send packets to MS using MSs PDP address as destination.

4.2.4.3 Mobility Management

Instead of Location Area, GPRS uses Routing Areas XE "Routing Areas" to group cells. RA is a subset of LA (see chapter 5.3.5). Mobiles can be in the following states from the NW point of view:

IDLE: MS is not known by the network

READY: MSs location is known in accuracy of cell, MS must inform its location after every cell change, MS can initiate Mobile Originating transfer at any time, SGSN XE "SGSN" does not need to page MS before Mobile Terminating data transfer

STANDBY: MSs location is known in accuracy of Routing Area, MS must inform its location after every Routing Area change (no need to inform cell changes within the same RA), Before the network can perform Mobile Terminating data transfer, the MS must be paged within the RA. MS may initiate Mobile Originating data transfer at any time.

Note that the upper states are defined from the mobility management point of view, from the radio resource point of view the following states are distinguished:

PACKET IDLE MODE: In this mode, the MS is not allocated any radio resource on a PDCH XE "PDCH" , it listens to the PBCCH and PCCCH or, if those are not provided by the network, to the BCCH and the CCCH.

PACKET TRANSFER MODE: In this mode, the MS is allocated radio resource on one or more PDCHs for the transfer of LLC PDUs.

RNP relevant GPRS aspects

4.2.5 Channel coding XE "Channel coding"

On the radio interface, data can be potected according to four different coding schemes, CS1 to CS4 (only CS XE "CS" -1 and CS-2 are implemented in step 1). These coding schemes offer different redundancy levels and thus different peak data throughputs:

CS XE "CS"

RLC data block (incl. header) [bytes]

RLC data unit [bytes]

RLC data unit throuput [kbit/s]

CS XE "CS" -1

22

20

8

CS XE "CS" -2

32

30

12

CS XE "CS" -3

38

36

14,4

CS XE "CS" -4

52

50

20

Table 1: Achievable data rates for the Coding Schemes XE "Coding Schemes" CS XE "CS" 1 to 4

The most protected mode is CS1, which is therefore always used for signalling. Depending on the radio condition CS1 or CS2 are used for traffic. This choice is done by the MFS XE "MFS" , according to the reception quality and level measurements performed by the BTS. The performance of coding schemes and the coding scheme adaptation algorithms will be described in more detail in Annex A.

4.2.6 Measurements XE "Measurements"

Downlink Measurements XE "Measurements"

MS measurement reports are transmitted from the MS to the MFS XE "MFS" in each DL(N) Acknowledge command containing the averaged level C of the serving (C_VALUE) and RXQUAL averaged over the received blocks. The reporting period depends on the parameters DL_ACK_PERIOD or DL_NACK_PERIOD (Range: 1 to 64, Default=16). This period defines the number of received data blocks between each DL Ack. So only when the MS receives data, the measurement report is sent. Thus the reporting period is not tuneable and it varies significantly.

Uplink Measurements XE "Measurements"

At each received block, level and quality are computed (over 4 TDMA frames) and sent to the MFS XE "MFS" . The measurements are averaged, the averaging algorithms are quite complex.

The relevant parameters for UL and DL are [13]:

LEVEL

QUALITY

DOWNLINK

T_AVG_T, T_AVG_W

DL_(N)ACK_PERIOD

UPLINK

K_AVG_L

KAVG_Q

Table 2: GPRS Averaging parameters for uplink and downlink measurements

If the Parameter PC_MEAS_CHAN = 0 the measurements are performed on the BCCH-TS. If PC_MEAS_CHAN =1, the measurements are performed on ALL blocks on one of the used PDCHs.

The Timing advance procedure is used to derive the correct value for timing advance that the MS has to use for the UL transmission of radio blocks. It comprises two parts:

Initial advance estimation

Continuous timing advance update

The evaluation is similar to the CS XE "CS" approach and is updated every 0.96 sec.

A further important figure which is related to the Radio Link is the Retransmission rate which is not measured or reported via the air interface but reported by the MFS XE "MFS" . However it is not used for Radio Link Relevant Algorithms (initially it was planned to use this parameter for dynamic coding scheme adaptation, but has been withdrawn since it is not always an accurate parameter).

4.2.7 Power Control XE "Power Control"

In step 1 power control is only implemented on the uplink in an open loop configuration, i.e. the MS adapts its output power per block (i.e. 4 timeslots) based on the received average signal strength assuming the same path loss in UL and DL. When accessing the NW on the (P)RACH the MS uses the output power defined by (GPRS_)MS_TXPWR_MAX_CCH, which is broadcasted by the (P)BCCH. In case of closed loop power control (not available in step 1) the output power is commanded by the NW site as for CS XE "CS" traffic.

4.2.8 Cell re-selection XE "Cell re-selection" and re-direction XE "re-direction" instead of handover XE "handover" 4.2.8.1 Cell re-selection XE "Cell re-selection"

There are no handovers for GPRS mobiles. Roaming to other cells is performed by cell (re)-selection in packet idle mode. When a mobile in packet transfer mode leaves the coverage range of a BTS or suffers from interference, the link quality will degrade and retransmission will be activated. In this case the GPRS mobiles remain connected to their serving cell until the call is dropped or the TBF XE "TBF" transmission is finished (which is more probable, due to the short duration of the data transfer per block flow). In case of a call drop cell re-selection in packet idle mode is performed and the data package is resent on the new cell. If the data transmission is finished without call drop, reselection in packet idle mode is performed and the next data package will be send on the new cell.

Furthermore the quality criteria of the radio link is for GPRS MS additionally dependant on the retransmission rate. Therefore are the RXQUAL/RXLEV based Power Control XE "Power Control" and Handover algorithms as implemented for CS XE "CS" traffic no longer applicable.

I.e. for cell roaming the GPRS mobile performs autonomous cell re-selection in packet idle mode. The MS measures the received signal strength on the (P)BCCH frequencies of the serving cell and the neighbour cells, as indicated in the GPRS (if MPDCH available) or the CS XE "CS" neighbourcell list and calculates the received level average of each frequency. As in GSM the 6 strongest neighbours are continuously monitored. The MS applies existing GSM reselection algorithms (C1 and C2) in case of cells without MPDCH or GPRS reselection parameters in case of cells with MPDCH (C31 and C32).

The parameters for cell (re)-selection are broadcasted on the (P)BCCH. The C31, C32 algorithms are quite similar to the C1, C2 criterion and are described detailed in (GSM 05.08 version 6.4.0): Only cells providing a receive level larger then RXLEV_ACCESS_MIN are potential candidates, the strongest candidate will be selected according to C1. In case of a required location update Cell_Reselect_Hysteresis is applied [14].

4.2.8.2 Cell redirection (cause 25)

This feature is not implemented in B6.2. This feature will be used when a mobile which wants to use a GPRS service establishes a call on a cell which does not support GPRS. The BSC indicates to the MS a neighbour cell which supports GPRS and in which the mobile can communicate. In such a case, the BSC assigns a SDCCH to the MS at receipt of the Channel Request message indicating a packet establishment cause. The BSS starts making measurements and processing averages. The Handover preparation entity is asked to start the GPRS redirection. Once a cell supporting GPRS and having sufficient radio conditions is detected, a handover XE "handover" alarm cause GPRS redirection (=cause 25) is sent to the Handover management entity. This entity will manage the list of candidate cells and send it to Internal Channel Change entity. Then a procedure of cell change order is triggered. In case of procedure failure, the channel is released. However, since it is not implemented in B6.2 is shall only be specified in more detail in Annex E.

4.2.9 Routing Areas XE "Routing Areas"

If the SGSN XE "SGSN" wants to transfer data in DL and if he MS is not in the READY state, then a paging command needs to be send to the MS. Thus the signalling effort for paging is assumed to be larger for GPRS compared to circuit switched traffic. However if the GPRS transfer is a mobile originated call, then an UL TBF XE "TBF" will be first established and put the MS into the READY state. In that case no paging is required.

Thus for GPRS mobiles, Routing Areas XE "Routing Areas" (RA) have been defined, which should be smaller than Location Areas (LA). An RA identifies one or several cells. Each cell is characterised by a Routing Area Code (0256) and an RA_COLOUR (07).

The location of an MS in STANDBY state is known in the SGSN XE "SGSN" on an RA level. This means that the MS is paged in the RA where the MS is located when mobile-terminated traffic arrives in the SGSN and no UL TBF XE "TBF" is ongoing.

An RA is a sub-set of one, and only one, Location Area (LA). An RA is served by only one SGSN XE "SGSN" as shown in Figure 12.

Figure 12: Subdivision of a Location Area in GPRS Routing areas (RA)

4.2.10 Discontinuous Reception (DRX XE "DRX" )

This function allows an MS, not to monitor all PCCCH blocks, but only blocks defined by its paging group. The MFS XE "MFS" determines the paging group from the IMSI value. To reduce paging group periodicity (15.36 s), this period is divided by the SPLIT_PG_CYCLE value. The MS applies existing GSM DRX XE "DRX" procedures if there is no MPDCH.

4.2.11 Quality of Service XE "Quality of Service"

The main part of QoS handling is performed in the SGSN XE "SGSN" . Only some basic functions are performed on BSS side. The subscribers QoS profile consists of the following parameters, which are negotiated between network and subscriber or are set to default values, before transmission. Their implementation in step 1 of GPRS is mentioned in this section. A more detailed description is given in Annex C.

Service precedence (priority): in step 1 of the GPRS implementation, the BSS assumes that all data units have the same precedence class attribute, i.e. they have all the same priority.

Delay XE "Delay" : Step 1 of the GPRS implementation only realises delay class 4, the "Best Effort", i.e. no mean or peak delay times can be guaranteed to the user.

User data throughput XE "throughput" : The user throughput is defined by maximum bit rate (peak throughput) and mean bit rate (mean throughput) given in kbit/s. The effective bitrate per user depends on the mobile multislot capability and the PDCH XE "PDCH" multiplexing.

Reliability XE "Reliability" : Five reliability classes are defined according to the traffic type as indicated in Table 3.

Reliability XE "Reliability" Class

RLC Block Mode

Traffic XE "Traffic" Type

1

Acknowledged

Non real-time traffic, error-sensitive application that cannot cope with data loss.

2

Acknowledged

Non real-time traffic, error-sensitive application that can cope with infrequent data loss.

3

Acknowledged

Non real-time traffic, error-sensitive application that can cope with data loss, GMM/SM, and SMS.

4

Unacknowledged

Real-time traffic, error-sensitive application that can cope with data loss.

5

Unacknowledged

Real-time traffic, error non-sensitive application that can cope with data loss.

NOTE:For real-time traffic, the QoS profile also requires appropriate settings for delay and throughput XE "throughput" .

Table 3: Reliability XE "Reliability" Classes XE "Reliability Classes"

4.2.12 GPRS Traffic XE "Traffic" Model XE "Traffic Model"

Traffic XE "Traffic" calculations for packet switched systems require a different approach. The cell capacity can not be defined by a blocking probability (Erlang B) as for CS XE "CS" traffic. The system capacity is rather defined by data rates per cell or per user. Based on the GPRS configuration the total offered bit rates per cell can be given. In order to determine the amount of users it is necessary to multiplex all users on this total cell capacity. Thus the required bitrate per user needs to be specified. Since the traffic occurrences differ depending on the data application a traffic model is required. The Alcatel GPRS traffic model is described in [2]. Three different user profiles have been defined (Pager and Traffic Management, Wireless NW Computers and Telemetry). Each of them with unsymmetrical UL/DL traffic occurrences, whereas higher traffic rates occur on the downlink.

4.2.13 BSS/CAE XE "CAE" Parameter

There are only a few GPRS related BSS Parameter which have to be set by the Radio Network Planner (with A955 V5). These parameters are listed in [15] as RNP relevant CAE XE "CAE" Parameter. RNO related parameters are set to default values. The RNP parameters are:

En_GRPS in the ADJ File in order to enable/disable GPRS traffic within the cell.

PDC file (PCH group definitions), containing the following parameters:

RNP_CELL_ID: Identifier of the cell

TRX_INDEX: Sequency number of the TRX per cell

PDCH XE "PDCH" _GROUP_ID: Identifier of the PDCH group consisting of TRX-TS, sharing same radio configuration (meaning, being allocated to the same TRX)

TSNO: Sequence Number of the TRX_TS per TRX used for GPRS traffic (07)

Even though these are the only GPRS related BSS parameter, which should be set by the RNP engineer, it is recommended to cross check if the other GPRS parameters are set to reasonable defaults. Further important parameters, which might help to understand the GPRS concept from the BSS parameter point of view have been already mentioned in the previous chapters. For more detailed information on these and many other parameters refer to [15].

GPRS Radio Network Planning

4.3 RNP Design Thresholds XE "Design Thresholds" based on Reference Performance Point XE "on Reference Performance Point"

As soon as a subscriber leaves the coverage range of a cell or enters an area of high interference the radio link quality degrades and the Block Error Rate XE "Block Error Rate" (BLER XE "BLER" ) increases. In GPRS the notion BLER is introduced, to consider an error rate per block independent of retransmission. The BLER is the probability that a block (456 bits in case of CS-4) is erroneous. Meaning that the block sequence check sum is wrong, which is the case as soon as one bit is wrong. For subscribers in the acknowledged mode retransmission is activated as soon as bits cannot be decoded. This maintains the link quality of the individual link (BER) but reduces the effective data throughput XE "throughput" , i.e. the RXQUAL value can still be good in case of bad link quality if retransmission is activated. Therefore it is necessary to consider besides RXQUAL and RXLEV also the retransmission rate or the BLER in order to be able to judge the quality of a radio link.

4.3.1 Reference Performance Point XE "Reference Performance Point"

With decreasing radio link quality the effective data throughput XE "throughput" is reduced due to retransmission. This dependency is shown schematically in Figure 13 for the coding schemes CS XE "CS" -1 and CS-2 in case of interference. It can be seen that a saturation effect occurs. E.g. for CS-1 the data throughput is 7kbit/s with a C/I ratio of 9dB. With an increasing C/I ratio the data throughput increases only little up to its maximum value of 8kbit/s (saturation point). According to ETSI the reference performance point XE "reference performance point" is defined by the point at which a BLER XE "BLER" of 10% is reached. This reference performance point is below the saturation point and is indicated schematically for both coding schemes in Figure 13. As soon as the C/I ratio is below this value, the retransmission significantly reduces the effective data throughput rate. A similar dependency is valid between the data throughput and the input level.

Depending on the fading environment (TU 50, HT 100) the reference performance point XE "reference performance point" is reached at different C/I and sensitivity thresholds. This context has been evaluated based on simulations, which are given more detailed in Annex B.

From the RNP point of view it is now reasonable to design the network with the optimum trade off between link quality requirements and data throughput XE "throughput" .

Thus the design target at the cell borders should be the radio conditions which are reached at the reference performance point XE "reference performance point" .

Figure 13: Definition of Reference Performance Point XE "Reference Performance Point"

As shown in Figure 14 the cell ranges are defined by the achievable data throughputs. With decreasing field strength level, four characteristic cell areas within the GPRS service area for the step 1 implementation (only CS XE "CS" -1 and CS-2) can be identified:

1. Area of CS XE "CS" -2 Saturation XE "Saturation" , i.e. data throughput XE "throughput" = 12kbit/s

2. Activation of retransmission due to insufficient coverage, i.e. reduced data throughput XE "throughput" until reference performance point XE "reference performance point" is reached [10.5kbit/s < throughput < 12kbit/s].

3. Performance at reference performance point XE "reference performance point" cannot be longer maintained, the BSS parameters should be set in a way that CS XE "CS" -1 is activated now => Area of CS-1 saturation, i.e. data throughput XE "throughput" = 8kbit/s

4. Activation of retransmission due to insufficient coverage in CS XE "CS" -1 area, i.e. reduced data throughput XE "throughput" until reference performance point XE "reference performance point" of CS-1 is reached [7kbit/s < throughput < kbit/s]. If the network is designed properly, the neighbour cell should provide sufficient level and reselection should be performed at the border of this area. Since no handover XE "handover" are implemented no cell overlap (as for CS designs) needs to be considered in the network design.

Figure 14: Dependency of data throughputs as a function of level (no interference)

Note, that the switch from CS2 to the more protected CS-1 mode gives higher reliability, since the operating point is in the more horizontal saturation area. But this is at the cost of reduced data throughput: Abrupt throughput reduction from 10.5 kbit/s to 8 kbit/s.

4.3.2 RNP Design Thresholds XE "Design Thresholds"

The C/I and sensitivity requirements at the reference performance point XE "reference performance point" has been specified by ETSI for various fading conditions:

4.3.2.1 C/I Requirements according to ETSI

For packet switched channels, the minimum interference ratio for which the reference performance for co-channel interference (C/Ic) shall be met is specified in table 2a in [9], according to the type of channel and the propagation condition. For standard RNP tasks, this table can be simplified.

Co-channel interference requirement at reference performance point XE "reference performance point" for PDTCH:

PDTCH C/I requirements at which reference performance point XE "reference performance point" is met

TU 3

TU 50

NO FH [dB]

Ideal FH [dB]

NO FH [dB]

Ideal FH [dB]

CS XE "CS" -1

13

9

10

9

CS XE "CS" -2

15

13

14

13

USF C/I requirement

USF CS XE "CS" -1

19

10

12

10

Table 4: C/I requirement according to ETSI for GPRS carriers

Non-hopping case: For slow moving mobiles the C/I requirements are significantly higher compared to a typical circuit switched design. Furthermore in order to maintain secure DL transmission of the USF (Uplink State Flag) for scheduling of UL TBFs, even 19dB are specified. This would require a drastic increase of the frequency reuse in the network design. However, it would not be reasonable, to design a non hopping network for a 19dB co-channel interference margin:

If the USF transmitted on the downlink can not be decoded, the UL TBF scheduling is postponed. This means, if we design our network for a 13 dB C/I margin only (CS-1 TU3 requirement) we would suffer from a degradation in the UL data throughput. However, this can be handled by the network, since the GPRS traffic is typically unsymmetrical and higher data rates are expected in the DL.

The introduction of frequency hopping reduces the C/I requirements significantly: For CS-1 almost the same thresholds as for the circuit switched case are valid. In the non hopping case for an optimised throughput a C/I of 13 dB is recommended.

Frequency hopping is recommended for the GPRS carrier.

The adjacent channel interference requirement is the same like for GSM circuit switched:

C/Ia = C/Ic - 18dB for all conditions.

4.3.2.2 Level/Sensitivity XE "Sensitivity" Requirements according to ETSI

For packet switched channels, the minimum input signal level for which the reference performance shall be met is specified in table 1a in [9], according to the type of channel and the propagation condition. The levels are given for a normal BTS. For standard RNP tasks, this table can be simplified to:

ETSI Level thresholds for PDTCH (for Alcatel specific UL threshold see below)

PDCH XE "PDCH" level requirements at which reference performance point XE "reference performance point" is met

TU 3

No FH

TU 50

No FH

CS XE "CS" -1

-104dBm

-104 dBm

CS XE "CS" -2

-104dBm

-100dBm

USF Level requirement

USF CS XE "CS" -1

-104dBm

-103dBm

Table 5: Level requirement according to ETSI for GPRS carriers

The thresholds are given for a normal BTS GSM 900. For other equipment, the levels shall be corrected by the following values:

- for DCS 1800 class 1 or class 2 MS : +2/+4 dB (normal/extreme conditions)

- for DCS 1800 class 3 MS :+2 dB

- for GSM 900 small MS :+2 dB (small MS:Not vehicle mounted, pwr. class 4=2W or 5=0.8W)

- for other GSM 900 MS and normal BTS : 0 dB

For the DL the upper ETSI requirements have to be considered, for the UL consideration in terms of sensitivity these values are dependent on the system supplier. As for CS XE "CS" traffic, where the performance of the Alcatel G3 BTS is 7dB better than required (-111dBm instead of -104dBm) accordingly for GPRS the G3 BTS performs also better. However according to simulations these 7dB are not always reached.

For an optimised throughput in the entire cell area an UL sensitivity of 109dBm for CS-1 and 105dBm for CS-2 should be considered in the GPRS link budget as long as no other values based on measurements are available.

Note: Not fulfilling these UL thresholds means a reduced data throughput at the cell border. If the NW is designed for 111dBm (as for G3 BTS circuit switched) instead of the upper values, the effect is that in UL direction the throughput is reduced. Since higher data rates are expected in downlink direction this could be handled by the system.

4.3.2.3 Comparision to thresholds of the circuit switched design

If the network shall not be optimised in terms of data throughput the same sensitivity thresholds can be applied as for the circuit switched design, accepting a throughput degradation at the cell border. Else 109dBm for CS-1 and 105dBm for CS-2 shall be applied for the G3 BTS.

The co-channel C/I requirement in the non-hopping case is significantly higher compared to the circuit switched case, especially for slow moving MS. Thus assigning the GPRS carrier on the BCCH carrier is recommended. A performance degradation in the UL will be expected, since the USF flag decoding requires a C/I of 19dB. But this can be handled since lower data rates are expected in uplink direction. However if frequency hopping is introduced, the same thresholds as for the CS designs can be applied without any degradation in the CS-1 performance. Therefore FH is recommended for the introduction of GPRS.

Initial GPRS Design XE "Initial GPRS Design"

4.3.3 GPRS Link Budget XE "Link Budget" and Cell Ranges XE "Cell Ranges"

Table 6 shows a GPRS link budget example, dimensioned for the G3 BTS: G3/900, 2 TRX, 0 ANy-Levels, 2 Antennas, air combining. The only difference to a circuit switched link budget are the used sensitivity values according to the previous chapter.

MS to BS

BS to MS

TX

Uplink

Downlink

Internal Power

33.0

dBm

45.4

dBm

Comb. + Filter Loss, Tol.

0.0

dB

1.6

dB

Output Power

33.0

dBm

43.8

dBm

Cable, Connectors Loss

3.0

dB

Body/Indoor Loss

4.0

dB

Antenna Gain

11.0

dBi

EIRP

29.0

dBm

51.8

dBm

RX

Uplink

Downlink

Rec. Sensitivity XE "Sensitivity"

-109 *)

dBm

-102/ -98**)

dBm

Body/Indoor Loss

4.0

dB

Cables, Connectors Loss

3.0

dB

2.0

dB

Antenna Gain

11.0

dBi

2.0

dBi

Diversity Gain

3.0

dB

Interferer Margin

3.0

dB

3.0

dB

Isotr. Rec. Power:

-117

dBm

-95.8 / -91.8

dBm

Max. Path Loss

146

dB

146.8 / 142.8

dB

Table 6: GPRS link budgets can be set up similar to CS XE "CS" link budgets

*) For more details on this threshold see previous section. This value is based on layer 1 simulations and is only preliminary as long as no measurements are available.

**) Valid for a slow moving MS in CS XE "CS" -1 mode / the second value is fora fast moving MS in CS-2 mode (TU 50) see Table 5.

Under static fading conditions the same sensitivity values can be applied in the link budget for CS XE "CS" -1 and CS-2. In case of fast moving MS, the sensitivities differ by 4dB.

The interferer margin considers the fact, that in case of coexistence of interference the system sensitivity is decreased. This has equally to be taken into account in GPRS designs. A margin of typically 3dB is applied

A fading margin can additionally be considered, if the path loss for cell range evaluations with a coverage probability above 50% shall be determined.

The cell ranges in coverage driven environments can then be estimated based on the analysis of the max. pathloss. According to Table 6 the max. pathloss for CS XE "CS" -1 = 146.8dB and for CS-2 = 142.8dB (fast moving mobiles = TU 50 ( sensitivity = 98dBm). Applying the Hata Formula with the parameters according to Table 7 the cell ranges and cell areas given in Table 8 can be determined for an omni configuration with a coverage probability of 90%.

Ant. Height BS

30.0

m

EIRP

51.8

dBm

Ant. Height MS

1.7

m

Max. Pthloss

146.8

dB

Frequency

900.0

MHz

Req. Cov.

0.9

Table 7: Hata parameters for cell range calculation

Clutter type

Cor [dB]

( [dB]

CS XE "CS" -2 Range [km]

CS XE "CS" -2 Area [km]

CS XE "CS" -1 Range [km]

CS XE "CS" -1 Area [km]

urban, flat

0

7

2.27

16.19

2.95

11.12

urban, hilly

0

14

1.35

5.75

1.76

3.94

suburban, flat

6

6

3.60

40.64

4.67

27.92

suburban, hilly

6

12

2.33

17.11

3.03

11.72

forest, flat

10

6

4.67

68.56

6.07

47.09

forest, hilly

10

10

3.52

38.96

4.57

26.72

open area, flat

25

6

12.46

487.24

16.18

334.68

open area, hilly

25

10

9.39

276.91

12.19

189.87

Table 8 : Achievable cell ranges in a coverage driven environment (Hata formula) for TU50

Accordingly 40% of the Cell area is covered by Coding Scheme 1 and 60% of the cell area is covered by Coding Scheme 2. Assuming that the BSS parameter for Coding Scheme Adaptation CS XE "CS" _LEV is set to 84dBm (=91.8dBm+8dB=-83.8dBm). A lognormal margin of 8dB is added to the isotropic received power of Table 6 to increase the coverage probability at the cell edge from 50% to 90%.

4.3.4 Frequency Planning XE "Frequency Planning"

The higher C/I requirements according to the previous chapter lead to the need of higher ARCS for planning GPRS frequencies. Thus it is recommended to assign the GPRS carrier always on the BCCH carrier, which has also higher C/I requirements, due to non availability of DTX, PC or hopping. A conservative BCCH frequency planning is thus additionally recommended. Disadvantage: On the BCCH carrier are only maximum 7 timeslots available.

4.3.5 Traffic XE "Traffic" dimensioning issues

Traffic XE "Traffic" in packet oriented systems is defined by its data throughput XE "throughput" [kbit/s]. Erlang B statistics are no longer applicable. The offered traffic per cell can be derived from the data throughput per timeslot, which is defined by the coding scheme performance. Which coding scheme is applied and what bit rates can be achieved due to retransmission depends on the radio link quality. Thus the offered cell capacity depends on radio link quality within the cell area (location dependent!) and on the amount of timeslots in GPRS mode.

The achievable capacity per user depends now on the amount of users sharing the same resource. Since in step 1 the best effort strategy is implemented, the amount of users can be roughly estimated if a traffic model with its user profiles are considered and the offered traffic is known:

4.3.5.1 TRAFFIC CALCULATION APPROACH 1:

According to the dimensioning example given in chapter 6.2.1 59% of the cell area is operated in CS XE "CS" -2 and 41% in CS-1 mode. The cell capacity can be estimated with a data rate of (0.41*8+0.59*12)kbit/s = 10.36 kbit/s per timeslot.

Assuming a traffic occurrence during busy hour according to [2]

45% Profile 1 traffic 21.6kbyte/hour

50% Profile 2 traffic 171 kbyte/hour

5% Profile 3 traffic 0.3 kbyte/hour

Total traffic occurrence per user during busy hour assuming a homogeneous traffic distribution: 0.45*21.6+0.5*171+0.05*0.3 = 95.2 kbyte/hour = 0.21 kbit/s

Thus maximum 10.36/0.21=49 users can be mapped on one timeslot with a profile according to the Alcatel GPRS Traffic XE "Traffic" Model XE "Traffic Model" . Since the traffic occurences are now inhomogeneous, collision probabilities occur and therefore a lower user amount is effectively possible.

However this approach can be used for a first approximation on maximal possible user amounts.

More accurate results in terms of offered traffic evaluation can be achieved based on fieldstrength prediction and the curves according to Figure 21 and Figure 19 (see chapter 6.3.4).

4.3.5.2 TRAFFIC CALCULATION APPROACH 2:

An alternative traffic dimensioning approach is given in [16]: The concept of GPRS allows the operator to share the radio resources between circuit switched and packet switched data. The Air interface dimensioning for GPRS is then very dependent of the allocation and desallocation mechanism for GPRS resources. In order to achieve the air dimensioning, the following inputs are needed:

description of the GPRS traffic model

number of GPRS users per cell per traffic model

number and duration of the GPRS transfers per user for each traffic model.

With this information, it is possible to calculate the mean number of MS, per multislot class, that are simultaneously engaged in a TBF XE "TBF" in a GPRS cell.

These values can be calculated for each traffic profile:

)

(

_

_

i

MS

Nb

Mean

with

[

]

Capacity

Multislot

Max

i

_

_

,

1

In the following equation, the TS needs are calculated, taking into account the total required GPRS bandwidth (Nb_TS):

The second step is to use this result as an input of a simulation tool which enables to measure the impact of GPRS on CS XE "CS" traffic with various cell configurations.

4.3.5.3 Impact on CS XE "CS" traffic

Allocating TS to GPRS traffic reduces the capacity within the circuit switched design. Depending on the amount of allocated timeslots for GPRS, the CS XE "CS" capacities based on a blocking probability of 2% are given in Table 9. According to chapter 0 the amount of timeslots allocated to GPRS is depending on the circuit switched and packet switched traffic. The maximum amount of PDCH XE "PDCH" is defined by MAX_PDCH_HIGH_LOAD in case of high BSC load indication and MAX_PDCH_GROUP otherwise.

Amount of TRX

Amount SDCCH

Amount TCH +PDCH XE "PDCH"

Amount PDCH XE "PDCH"

0

1

2

3

4

5

6

7

1 TRX

1

7

2.93

2.27

1.65

1.09

0.6

0.2

0.02

0

2 TRX

2

14

8.20

7.4

6.6

5.8

5.08

4.34

3.62

2.93

3 TRX

2

22

14.89

14.03

13.18

12.33

11.49

10.65

9.82

9.01

4 TRX

3

29

21.03

20.15

19.26

18.38

17.50

16.63

15.76

14.89

5 TRX

3

37

28.2

27.3

26.4

24.6

23.7

22.8

21.9

21.03

6 TRX

4

44

34.6

33.7

32.8

31.9

30.99

30.08

29.16

28.25

Table 9: Remaining Erlangs for circuit switched cell traffic (2%Blocking)

4.4 Detailed RNP design XE "Detailed RNP design" 4.4.1 Coverage XE "Coverage" Planning

The application of the coding schemes guarantees that areas covered for GSM services by a BTS also offer coverage for GPRS services, at least with the CS XE "CS" -1 coding scheme. The level and C/I requirements are a little higher, however this results in a reduced throughput XE "throughput" at the cell border. So if just the service has to be offered without the necessity of an optimised overall throughput, "GPRS-cells" can be planned as conventional GSM cells concerning coverage. If a fieldstrength prediction is performed, the areas should be represented in which the different coding schemes can be used. This means, that a check is performed for each pixel if the predicted level is above the threshold of CS-1 (minimum requirement), of CS-2, CS-3 or even CS-4. The coverage plot shows the coverage areas of the different coding schemes (see schematic representation in Figure 15).

Current planning tools can thus be applied: A coverage probability per coding scheme, best server plots etc. can be given. Note, that within the CS XE "CS" -1 coverage area of Figure 15 still areas with CS-2 service can occur due to interference. A real coding scheme map of course requires both inputs (level and interference).

Figure 15: GPRS coverage for different coding schemes (schematic)

4.4.2 Interference XE "Interference" Analysis

Annex B gives a detailed description of the impact of interference on the performance of the different coding schemes. The availability of the GPRS service is assured, at least with the CS XE "CS" -1 coding scheme for slow moving mobiles, if the GSM C/I constraints are fulfilled (see Table 4). But if one wants to take advantage of the higher throughput XE "throughput" offered by the coding scheme CS-2, resp. CS-3 and CS-4 (when implemented) or if full CS-1 data rates shall be offered to fast moving mobiles in the complete cell area, then the level AND the specified interference constraints have to be fulfilled. So, if both constraints are taken into account, service areas of the different coding schemes can be defined with a certain probability. This probability corresponds to the probability, that the interference ratio C/I is higher than the according threshold. Therefore, in a RNP tool, the calculation of the interference probability has to be executed separately for the GPRS carriers, taking into account the different thresholds per service class.

In order to achieve a maximum throughput XE "throughput" , the service areas should not be interference limited. That means, in the coverage areas described, the according C/I constraint should be fulfilled, too. This should be taken into account for frequency planning of the GPRS carriers. A solution could be to map the GPRS carrier on the BCCH carrier and to apply a conservative frequency reuse. The disadvantage is that on this carrier at most only 7 timeslots are available, since TS0 of the BCCH carrier cannot be used.

4.4.3 Coding scheme XE "Coding scheme" and Throughput XE "Throughput" density map

A more exact coding scheme map and a throughput XE "throughput" density map can be derived from a pixel wise analysis of the C/I and level predictions. These maps can be used for capacity evaluation (see next section) and further predict the throughput performance within the service area.

4.4.4 GPRS Capacity XE "Capacity"

A simplified approach for the evaluation of the maximum amount of GPRS users is described in chapter 6.2.3. A more accurate approach consists in using the mentioned coding scheme and throughput XE "throughput" density maps (chapter 6.3.3), which leads to the total throughput by integrating over the whole cell area and normalising. For each coding scheme, a constant throughput performance can be assumed with a little degradation at the cell borders. The mean offered data traffic capacity of one cell is equal to its total throughput and is given in kbit/s/cell. Considering the user profiles of the GPRS data traffic model as well as the amount of users and their distribution among these models. The amount of users per cell can be determined.

The traffic database containing the offered packet switched data traffic in the planning area has to be completely different from a GSM traffic database. It should contain the traffic density in kbit/s/pixel. It could be derived from a traffic budget, where the total occurring traffic in bytes per busy hour is elaborated for the uplink as well as for the downlink [2]. The limiting link is then selected to feed the traffic database.

4.5 GPRS RNP strategies 4.5.1 Migration to GPRS in case of existing GSM network

From the coverage point of view, the existing GSM cells can be reused, knowing that there will be a throughput XE "throughput" degradation at the cell borders (due to smaller sensitivity of GPRS equipment). However it might be necessary to revise the frequency plan for the GPRS carrier due to higher C/I constraints (13dB instead of 9dB for CS-1). In non hopping systems it is recommended to use the anyhow conservative planned BCCH carrier. The design values for the USF will still not be met since they are extremely high with 19dB, however this results only in a performance degradation in uplink direction. Frequency hopping improves the system performance under interference conditions significantly, so that the design values of the circuit switched designs can then be applied.

The expected development is, that in the beginning GPRS traffic is quite limited and one TRX will be sufficient. One and later more TS will be dedicated to GPRS traffic. But with increasing traffic for GPRS the CS XE "CS" capacity will be reduced and carrier upgrading will be necessary. In a bandwidth limited environment this can then lead to interference problems. According to marketing analysis the total traffic will increase strongly due to additional data services. NW expansion strategies like microcells promising the highest capacity gains will then play an important role.

4.5.2 Evaluation of capacity gains based on network expansion strategies

Achievable capacity gains in percent have been determined for the available network expansion strategies in the circuit switched case. However these values cannot be applied one to one for the GPRS traffic:

Introducing a network expansion strategy (e.g. SFH, concentric cells, microcells etc.) in a bandwidth limited environment first of all improves the network quality. Thus carrier upgrading becomes possible in the circuit switched case, since now frequencies can be reused more tightly.

If we think now on GPRS step 1 implementation, we can only switch one carrier to GPRS per cell, thus carrier upgrading (several GPRS TRX per cell) is not possible in the packet switched case. The upper limit is physically defined and the relationship between frequency reuse and capacity as it is given for CS XE "CS" traffic is not valid for packet switched designs. Therefore the achievable capacity gains have to be analysed from a different point of view.

The evaluation of capacity is also different: Erlang for circuit switched and kbit/s for packet switched systems.

Of course in GPRS there is a close relationship between radio link quality and throughput XE "throughput" , so introducing a network expansion strategy without carrier upgrading already improves the maximum throughput rates. This capacity improvement is further location dependant and the achievable gains depend very much on the individual design. Therefore it will be very difficult to predict for the 1 TRX GPRS configuration the achievable capacity gains for the classical network expansion strategies. Tool supported capacity analysis of the individual design will be required for capacity improvement predictions.

Summary

With Release B6.2 Alcatel introduces the step 1 implementation of GPRS. To have GPRS running in an Alcatel EVOLIUMTM BSS only software upgrade is needed. The software is remotely downloaded from the OMC-R. The PCU function is housed in the Alcatel Multi-BSS Fast Packet Server A935-MFS, which can be co located with the MSC and can be connected to up to 22 BSCs, thus allowing easy installation and maintenance.

GPRS is a bearer service providing end-to-end packet-switched data services at a rate of up to 96kbit/s per cell in step 1 (max. 12kbit/s per timeslot (CS-2 mode) and max. 1 TRX for GPRS per cell). It allows efficient usage of the radio resources and charging can be based on data volume transmitted. A radio connection is established only when a data packet has to be transmitted.

A timeslot allocated for GPRS is called a Packet Data Channel (PDCH). For the unidirectional packet data transfer, the data is transmitted on a Temporary Block Flow (TBF) and is identified by a Temporary Flow Identity (TFI). Several TBFs (=several users) can be multiplexed on one PDCH. One TBF can be transmitted on several PDCHs depending on the mobiles multislot capability. The mapping of logical channels, the user multiplexing and the TBF establishment scenarios for UL and DL as well as the PDCH dynamic channel allocation is described in chapter 5.2.

On the radio interface, data can be coded according to currently two (later four) different coding schemes (CS-1 to CS-4). These coding schemes offer different redundancy levels and thus different data throughput rates. Currently implemented in step 1 is CS-1 with a timeslot capacity of 8kbit/s and CS-2 with 12 kbit/s.

Radio link measurements are performed in uplink and in downlink direction. Level and quality is computed block wise in uplink. In downlink the MS measures the level of the serving cell and 6 strongest neighbours. However only level and quality of a link is reported, which will be processed by averaging in the BSS (C_VALUE and RXQUAL). Further the timing advance is determined similar to the CS method and is updated every 0.96 sec.

Power control is only implemented on the uplink in an open loop configuration.

Cell roaming is performed in form of cell selection and re-selection according to the circuit switched C1, C2 criterion or the GPRS specific C31, C32 criterion depending on the PDCH configuration (MPDCH yes or no). I.e. handovers are not implemented, if the mobile leaves a cell area in the packet transfer mode, the data transfer will be continued until the call is dropped. After cell re-selection the last TBF will be resent to the new cell.

Mobility management is performed by defining Routing Areas (RA) which are a subset of one Location Area (LA).

From the Radio Network Planning point of view, the design values in terms of C/I ratio and sensitivity are defined by the reference performance point. This point is specified by ETSI and defines the thresholds below which retransmission due to bad link quality significantly reduces the effective data throughput (BLER = 10%). Thus it should be the target of a network design to fulfil the specified thresholds at this point.

The co-channel C/I requirement in the non-hopping case is significantly higher compared to the circuit switched case, especially for slow moving mobiles (13dB for CS-1 and 15dB for CS-2). If a network design should be optimised for data throughputs, then these should be the design values. However if a throughput degradation can be handled in interfered areas it is recommended to use reuse factors of the circuit switched design and to assign the GPRS service on the BCCH carrier, which is usually planned with larger reuses. Note that this has the disadvantage that max. 7 TS are usable for data or speech traffic.

Applying this frequency planning strategy will have the effect, that the USF flag decoding will be interfered, since for the transmission of the USF at reference performance point a C/I ratio of 19dB is required. This threshold cannot be a design target value since it would require huge ARCSs. The USF is transmitted via the BTS and schedules the UL TBFs. The effect of USF decoding problems will result in reduced throughput in uplink direction.

However if frequency hopping is introduced, the same thresholds as for the CS designs can be applied without any degradation in the CS-1 performance. Therefore FH is recommended for the introduction of GPRS.

The downlink sensitivity for fast moving mobiles in CS-2 mode is 4dB lower than in the other cases (slow moving CS2 and CS1 mode). The uplink sensitivity is system supplier dependent. For the EVOLIUM BTS no measurements are available yet. According to simulation results the sensitivity is a little lower then in the circuit switched case (preliminary value 109dBm for CS-1 and 105dBm for CS-2). However, if the network shall not be optimised in terms of data throughput the same sensitivity thresholds can be applied as for the circuit switched design (-111dBm), accepting a throughput degradation at the cell border.

Considering these sensitivity thresholds, the link budgets and cell ranges can be determined accordingly. For the frequency planning in case of no hopping it is recommended to use the BCCH carrier for GPRS service as long as max. 7 timeslots are sufficient from the traffic point of view. The BCCH carrier is due to unavailability of PC, DTX or SFH usually planned with a large ARCS. Frequency hopping is recommended to use with GPRS, since it significantly improves the resistance. The same threshold as for the circuit switched case can be applied.

In order to determine the cell traffic capacity the easiest approach is to evaluate the mean data throughput based on an assumed CS-1 and CS-2 contribution proportional to each service area. In order to map this cell capacity on a capacity per user a GPRS peak hour traffic model needs to be defined. Based on this the mean bit rate per user during busy hour can be evaluated. Cell capacity divided by mean bit rate per users gives the maximum amount of users per timeslot according to the assumed traffic model.

For a detailed GPRS RNP design a tool supported approach is necessary by evaluating the data throughput per pixel. This can be done by mapping the predicted fieldstrength and C/I values to a bitrate. Such a datathroughput map allows to identify areas of bad and good data throughput. Further cell areas for CS-1 and CS-2 mode can be identified. By integrating and normalising over the throughput density map, the offered cell traffic can be evaluated more precisely. For a more accurate evaluation of the amount of users per cell an according mapping of the traffic model to the bitrates per user during peak hour is still missing.

ANNEX A Channel Coding and dynamic coding scheme adaption

In general, transmitted data are extremely error sensitive, so that an efficient error protection has to be assured. GPRS provides a combination of FEC (Forward Error Correction) and ARQ (Automatic Repeat reQuest). The latter consists of an error detection in the receiver, which is followed by a request for retransmission of the blocks containing errors or of an acknowledgement in case of successful transmission. For the "unacknowledged mode" (see 0), no ARQ error protection is applied because services are time critical. It has to be noted that this error protection is performed by the MAC/RLC layer. In order to guarantee the user QoS, higher layers provide additional error protection protocols.

Four different channel coding schemes, CS XE "CS" -1 to CS-4, are defined for the packet data traffic channels.

The first step of the channel coding procedure is to add a Block Check Sequence (BCS) for error detection. For CS XE "CS" -1 to CS-3, the second step consists of adding four tail bits and a half rate convolutional coding for error correction that is punctured to give the desired coding rate. For CS-4 there is no coding for error correction. It has to be noted that the initial GPRS implementation only implements CS-1 and CS-2.

The reference performance point XE "reference performance point"

Within ETSI, the performance of the different coding schemes has been examined intensively. The impact of interference has been studied in order to find the minimum required C/I level for the reference error performance, defined by a block error rate (BLER XE "BLER" ) of 10-1. Simulations were also ran without co-channel interferers, considering white noise as perturbation, in order to determine the required signal level at the mentioned reference error performance point. The simulation results are presented in Annex B.

Dynamic Coding Scheme Adaptation

The Alcatel GPRS solution contains a dynamic coding scheme adaptation mechanism. Depending on the average received level (RX_LEV) and the transmission quality (RX_QUAL), the appropriate coding scheme (either CS XE "CS" -1 or CS-2) is determined dynamically, both in the uplink and in the downlink. An O&M flag (DYN_CS_ENABLED) indicates whether this dynamic adaptation is activated in the cell or not.

Three thresholds allow to tune the dynamic coding scheme adaptation:

1. CS XE "CS" _LEV

constitutes the RX_LEV threshold used in radio link supervision, to change from CS1 to CS2 (and conversely)

Range: -110 to -47 dBm

Default value: -100 dBm

2. CS XE "CS" _QUAL_1_2

constitutes the RX_QUAL threshold used in radio link supervision, to change from CS1 to CS2

Range: 0 to 7

Default value: 2

3. CS XE "CS" _QUAL_2_1

constitutes the RX_QUAL threshold used in radio link supervision, to change from CS2 to CS1

Range: 0 to 7

Default value: 2

If a sufficient number of samples has been received to ensure a reliable quality, level and retransmission rate estimates, the decision whether to change or not the coding scheme can be made. The decision to change the coding scheme is made each time one of the following causes is checked:

Change from CS XE "CS" -1 to CS-2

AV_RXQUAL < CS XE "CS" _QUAL_1_2

and AV_RXLEV > CS XE "CS" _LEV_1

Change from CS XE "CS" -2 to CS-1

AV_RXQUAL >= CS XE "CS" _QUAL_2_1

OR

AV_RXLEV