EGPRS Radio Algorithms & Parameters Description B10.pdf

E-GPRS Radio Algorithms and Parameters Description B10 - Page 1All Rights Reserved © 2007, Alcatel-Lucent

All rights reserved © 2007, Alcatel-Lucent

EVOLIUM - E-GPRS Radio Algorithms and Parameters Description B10

EVOLIUME-GPRS Radio Algorithms and Parameters Description B10

TRAINING MANUAL

3FL11830ACAAWBZZA2Edition 02

Copyright © 2007 by Alcatel-Lucent - All rights reservedPassing on and copying of this document, use and

communication of its contents not permitted without written authorization from Alcatel-Lucent


All Rights Reserved © Alcatel-Lucent 2007

EVOLIUM

E-GPRS Radio Algorithms and Parameters Description

B10

2

Legal Notice

� Switch to notes view!Safety Warning

Both lethal and dangerous voltages are present within the equipment. Do not wear conductive jewelry

while working on the equipment. Always observe all safety precautions and do not work on the

equipment alone.

Caution

The equipment used during this course is electrostatic sensitive. Please observe correct anti-static

precautions.

Trade Marks

Alcatel and MainStreet are trademarks of Alcatel.

All other trademarks, service marks and logos (“Marks”) are the property of their respective holders

including Alcatel-Lucent. Users are not permitted to use these Marks without the prior consent of Alcatel

or such third party owning the Mark. The absence of a Mark identifier is not a representation that a

particular product or service name is not a Mark.

Copyright

This document contains information that is proprietary to Alcatel-Lucent and may be used for training

purposes only. No other use or transmission of all or any part of this document is permitted without

Alcatel-Lucent’s written permission, and must include all copyright and other proprietary notices. No

other use or transmission of all or any part of its contents may be used, copied, disclosed or conveyed to

any party in any manner whatsoever without prior written permission from Alcatel-Lucent.

Use or transmission of all or any part of this document in violation of any applicable Canadian or other

legislation is hereby expressly prohibited.

User obtains no rights in the information or in any product, process, technology or trademark which it

includes or describes, and is expressly prohibited from modifying the information or creating derivative

works without the express written consent of Alcatel-Lucent.

Alcatel-Lucent, The Alcatel-Lucent logo, MainStreet and Newbridge are registered trademarks of Alcatel-

Lucent. All other trademarks are the property of their respective owners. Alcatel-Lucent assumes no

responsibility for the accuracy of the information presented, which is subject to change without notice.

© 2007 Alcatel-Lucent. All rights reserved.

Disclaimer

In no event will Alcatel-Lucent be liable for any direct, indirect, special, incidental or consequential

damages, including lost profits, lost business or lost data, resulting from the use of or reliance upon the

information, whether or not Alcatel has been advised of the possibility of such damages.

Mention of non-Alcatel-Lucent products or services is for information purposes only and constitutes

neither an endorsement nor a recommendation.

Please refer to technical practices supplied by Alcatel-Lucent for current information concerning Alcatel-

Lucent equipment and its operation.



EVOLIUM


B10

3

Table of Contents

� Switch to notes view!1. Radio Algorithms

1. Principles

2. Radio Resource management

3. Radio Link Control

4. Algorithms Dynamic Behaviors

5. Appendix



EVOLIUM


B10

4

Table of Contents [cont.]

� Switch to notes view!

This page is left blank intentionally



EVOLIUM


B10

5

Course Objectives


Welcome to E-GPRS Radio Algorithms and Parameters Description B10

After successful completion of this course, you should understand :

� How to describe the main (E)GPRS mechanisms and concepts

� How to describe the radio algorithms and the related parameters

� How to estimate qualitatively the impact of a parameter change in order to solve typical

problems or enhance the GPRS performance



EVOLIUM


B10

6

Course Objectives [cont.]





EVOLIUM


B10

7

About this Student Guide

� Switch to notes view!Conventions used in this guide

Where you can get further information

If you want further information you can refer to the following:

� Technical Practices for the specific product

� Technical support page on the Alcatel website: http://www.alcatel-lucent.com

Note

Provides you with additional information about the topic being discussed.

Although this information is not required knowledge, you might find it useful

or interesting.

Technical Reference (1) 24.348.98 – Points you to the exact section of Alcatel-Lucent Technical

Practices where you can find more information on the topic being discussed.

WarningAlerts you to instances where non-compliance could result in equipment

damage or personal injury.



EVOLIUM


B10

8

About this Student Guide [cont.]





EVOLIUM


B10

9

Self-Assessment of Objectives

� At the end of each section you will be asked to fill this questionnaire

� Please, return this sheet to the trainer at the end of the training

�

Switch to notes view!

Instructional objectives Yes (or globally yes)

No (or globally no)

Comments

Contract number :

Course title :

Client (Company, Center) :

Language : Dates from : to :

Number of trainees : Location :

Surname, First name :

Did you meet the following objectives ?

Tick the corresponding box

Please, return this sheet to the trainer at the end of the training

��



EVOLIUM


B10

10

Self-Assessment of Objectives [cont.]


Instructional objectives Yes (or Globally yes)

No (or globally no)

Comments

Thank you for your answers to this questionnaire

Other comments

��

Section 1 � Module 1 � Page 1


3JK10863AAAAWBZZA Issue 02

Do not delete this graphic elements in here:

1�1All Rights Reserved © Alcatel-Lucent 2008

Module 1Principles


Section 1Radio Algorithms


3FL11830ACAAWBZZA2 Issue 02





EVOLIUM � E-GPRS Radio Algorithms and Parameters Description B10Radio Algorithms � Principles

1 � 1 � 2

Blank Page


First editionLast name, first nameYYYY-MM-DD01

RemarksAuthorDateEdition

Document History






1 � 1 � 3

Module Objectives

Upon completion of this module, you should be able to:

� Describe the main GPRS mechanisms and concepts






1 � 1 � 4

Module Objectives [cont.]







1 � 1 � 5

Table of Contents

Switch to notes view!Page

1 Service Overview 72 Alcatel-Lucent GPRS Architecture 133 Main Transactions 20






1 � 1 � 6









1 � 1 � 7

1 Service Overview






1 � 1 � 8

1 Service Overview

Data Transfer with GSM: Circuit Switching

Internet

GSM

Network

Air InterfaceAccess Node

Circuit switching

� Transaction is offered in connected mode.

� Allocation of a continuous radio resource UL/DL until the completion of the transfer.

� One circuit = one channel allocated per user.

� The traffic multiplexing is achieved inside the BSS, over the Ater interface.

High Speed Circuit Switched Data (HSCSD)

� A technology that allows the multislot allocation to one user.

� An important throughput can be achieved (up to 64 Kbit/s constant) but there is no optimization of the

use of the channel.

� (no dynamic allocation when data services are mainly carried in a bursty mode).

� The billing is time based like in GSM.

� Is likely to lead to an important congestion situation.

� Not offered by Alcatel-Lucent.






1 � 1 � 9

1 Service Overview

Data Transfer with GPRS: Packet Switching

GPRS

NetworkInternet

Air Interface

GPRS provides end-to-end packet-switched data transmission between MS users and fixed packet data

networks.

GPRS is a GSM feature.

GPRS provides an efficient use of the radio resources:

� multislot operation,

� Flexible sharing of radio resources between MSs,

� Resources are allocated only when data are transmitted.

Charging is based on the volume of data transmitted, not on connection time.






1 � 1 � 10

1 Service Overview

MS Classes

� Three MS classes are defined:

� class A:

� simultaneous GPRS and GSM traffic (different from Class A DTM MS)

� class B:

� simultaneous GPRS and GSM attach but not simultaneous traffic

� an MS can be paged for a GSM call while performing a GPRS transfer

� class C:

� only GPRS

B10

Do not confuse with the multislot class of an MS, or the MS capacity, which characterizes the number of

TSs one MS can monitor in the UL and the DL simultaneously.

For the detail of the MS multislot class, see the next slide.

The main traffic class available on the market is Class B.

Class C is preferred for PDA devices

A new feature called Dual Transfer Mode (DTM) has been introduced in 3GPP Standard and in the Alcatel-

Lucent B10 release.

Contrary to classical Class A MSs, Class A DTM MSs do not need two independent transmitter/receiver.






1 � 1 � 11

1 Service Overview

MS Multislot Class B10

MS Type

� Type 1 are simplex MSs, i.e., without duplexer: they are not able to transmit and receive at the same time.

� Type 2 are duplex MSs, i.e., with duplexer: they are able to transmit and receive at the same time.

Rx

The maximum number of received time slots that the MS can use per TDMA frame. The received TSs shall be

allocated within window of size Rx, but they do not need to be contiguous. For SIMPLEX MS, no transmitted TS shall

occur between received TSs within a TDMA frame. This does not take into account the measurement window (Mx).

Tx

The maximum number of transmitted time slots that the MS can use per TDMA frame. The transmitted TS shall be

allocated within the window of size Tx, but they do not need to be contiguous. For SIMPLEX MS, no received TS shall

occur between transmitted TSs within a TDMA frame.

SUM

The maximum number of transmitted and received time slots (without Mx) per TDMA frame.

The meaning of Ttb, Tra et Trb changes according to the MS type.

� For SIMPLEX MS (type 1):

� Ttb is the minimum time (in time slot) necessary between the Rx and Tx windows.

� Tra is the minimum time between the last Tx window and the first Rx window of the next TDMA in order to

be able to open a measurement window.

� Trb is the same as Tra without opening a measurement window.

� For DUPLEX MS (type 2):

� Ttb is the minimum time necessary between 2 Tx windows belonging to different frames.

� Tra is the minimum time necessary between 2 Rx windows belonging to different frames in order to be able

to open a measurement window.

� Trb is the same as Tra without opening a measurement window.

EDA

The Extended Dynamic Allocation, is a new B10 feature that allows higher throughput in uplink for type 1 MS.

Through the support of more than two radio TSs (e.g. MS multislot class 11)






1 � 1 � 12

1 Service Overview

MS Multislot Class [cont.]

� New Multislot classes supported in B10

� Activation with the flag EN_MULTISLOT_CLASS_30_33

B10

Multislot

class Rx Tx Sum Tta Ttb Tra Trb PTM DTM

30 5 1 6 2 1 1 1 (5+1)

31 5 2 6 2 1 1 1 (5+1), (4+2) (4+2)

32 5 3 6 2 1 1 1

(5+1), (4+2),

(3+3)/EDA

(4+2),

(3+3)/EDA

33 5 4 6 2 1 1 1

(5+1), (4+2),

(3+3)/EDA,

(2+4)/EDA

(4+2),

(3+3)/EDA,

(2+4)/EDA

34 5 5 6 2 1 1 1

(5+1), (4+2),

(3+3)/EDA,

(2+4)/EDA,

(1+5)/EDA

• Note that DTM multislot classes 30 and 34 do not exist

• PTM multislot class 34 can be supported as multislot class 33

• Multislot 30-33 MS can support bi-directional transfer configurations with up to 5 DL TS.

• EDA necessary to fully support multislot classes 32 and 33

• PTM high multislot classes defined in 3GPP release 5 and,

• DTM high multislot classes defined in 3GPP release 6

• Main benefits :

� More downlink throughput

� PTM mode : up to 5 DL PDCHs instead of 4 (5+1 instead of 4+1)

� DTM mode : up to 3 DL PDCHs instead of 2 (4+2 instead of 3+2)






1 � 1 � 13

1 Service Overview

MS Multislot Class [cont.]

� PTM MultiSlot Class 11 (4+1):

� DTM MultiSlot Class 11 (2+3):

� MultiSlot Class 11 (2+3) EDA:

� High MultiSlot Class (5+1):

Rx

Mx

0 1 2 3 4 5 6 7

DL

Tx UL

Ttb Tra

Rx Rx Rx

Rx Mx

0 1 2 3 4 5 6 7

DL

Tx

UL

Ttb Tra

Rx Rx

Rx Rx

B10

Multislot class 11:

EDA shall be used in UL for the 1+3, 2+3 and 1+2(1hole) configurations.

DA (Dynamic Allocation) shall be used for all the other configurations (4+1, 3+2, 3+1, 2+2, 2+1, 1+2 and

1+1).






1 � 1 � 14

2 Alcatel-Lucent GPRS Architecture






1 � 1 � 15

HLR

Gn

DNS/DHCP

IPBackbone

ChargingGateway

BorderGateway

GGSNInternetIntranet

Gi

PSTNMSC/VLR

MFS

SGSN

Gs Gr

BTS

BTS

Radio Access Network

AbisAterMux

Gb

TC

BSC

GSM Core Network

GPRS & EGPRS Core Network


General Architecture

The BSS is used for both circuit-switched and GPRS services.

A GPRS core network (also called GSS, an IP backbone) offers the interconnection between the PDN and

the BSS.

The BSS has 2 clients:

� The MSC, for circuit-switched services (through the A interface).

� The GPRS backbone network, for packet-switched services (through the Gb interface).

The A interface is unchanged.

PDN X25 is not supported anymore by 3GPP.






1 � 1 � 16

A

Gn

Signaling + data

Pure Signaling

Mobile GPRS

MSCSMS-

GMSC

Gd

Um

GPRS Network

Gb

BSS

Gc

HLR

GsGr

Gi

SGSN

SGSN GGSN


Main Entities - Interfaces

PDN

GPRS network = IP network

Note: Additional IP routers might be used to route the information between the GSNs (intra-PLMN

backbone network). All the elements connected to this backbone have private permanent IP addresses.

Signaling protocols:

� MAP/TCAP/SCCP/MTP on Gr, Gd and Gc (through the SGSN for the latter),

� GTP/UDP/IP on Gn, BSSAP+/SCCP/MTP on Gs,

� GMM/SM/LLC on Gb/Um.

Gc: for Network-Requested PDP contexts Activation (the GGSN asks the HLR for SGSN Routing

Information).

Gs: defines the Network Mode of Operation I. It allows to perform LA + RA combined Location Update,

and PS and CS Paging Coordination.

Gr: exchange of Subscription Information at Attachment Phase.

Additional interfaces:

� Gf (to the EIR).

� Gd to deliver the SMS to the mobiles via the GPRS network (SGSN option and subscriber feature).






1 � 1 � 17

� A Packet Control Unit is defined by the GSM standard:

� It handles RLC/MAC functions

� It may be in the BTS, in the BSC or, in the SGSN

� Alcatel-Lucent choice:

� PCU in a network element called the MFS

� smooth and cost-effective introduction of the GPRS

BTS

BTS

BSC

BSC

MFS SGSN

Gb

AterMux


PCU

The standard specifies that the PCU function shall be implemented in one of the 3 following entities:

� BTS,

� BSC,

� after the BSC (in the SGSN for instance)

The implementation of the PCU functions determines the position of the Gb interface.

Alcatel-Lucent chooses the MFS integration in order to offer a faster implementation inside the BSS as

well as an easier maintenance and supervision.

MFS: Multi BSS Fast packet Server.






1 � 1 � 18


M-EGCH principle

� An M-EGCH organizes the transmission resources (pool of Abis nibbles and Ater nibbles) given to a TRX for its PS traffic

� 3 types of nibbles are considered

� BASIC nibbles shareable between all TREs of a given cell

� BONUS and EXTRA nibbles shareable between all TREs of a given BTS

TRX 3

TRX 2

TRX 1

Abis Ater

M-EGCH 3

M-EGCH 2

M-EGCH 1

BasicExtraBonus

Basic Abis Nibbles

One Basic Abis nibble is initially linked to one RTS. The number of Basic Abis nibble in a given cell

corresponds to the number of RTS. They have the following characteristics:

� They carry either CS or PS traffic.

� They are shareable at cell level (among all the TREs).

Bonus Abis Nibbles

The basic Abis nibbles whose RTSs are currently used for BCCHs and static SDCCHs are called bonus Abis

nibbles. They have the following characteristics:

� They only carry PS traffic.

� They are shareable at BTS level (among all the TREs).

Extra Abis Nibbles

Considering one (E)GPRS capable BTS, the Operator can defined additional GCHs: the Extra Abis nibbles.

They have the following characteristics:

� They only carry PS traffic.

� They are shareable at BTS level (among all the TREs).

The multiplexing capacity of the GCH link on the AterMux belongs to the granularity chosen. 1 Ater TS

GPRS dedicated = 1 GCH when allocated.

The number of 64 Kbit/s time slots assigned to PS traffic and signaling is configured by the Operator from

the OMC-R, with the following granularity: 4, 8, 15, 22, and 29 per PCM.






1 � 1 � 19

MS BTS MFS SGSN

Um Abis/Ater(mux) Gb

SNDCP

GSM RFGSM-RF

RLC

LLC

SNDCP

Application

IP

L2-GCHRelay

L1bis

L2-GCH

RLC

NS

LLC

MACBSSGP

MAC

L1-GCH L1-GCH L1bis

NS

BSSGP

Relay


Transmission Plane

For the exact purposes of the tracing, please refer to “Introduction to GPRS & E-GPRS Quality of Service

Monitoring”.

It can be said from this protocol stacks diagram that after allocation of a GCH by the BSC to the MFS, the

data carried over the GCH are transparent for the BSC.

The RLC function defines the procedures for segmentation and reassembly of LLC PDUs into RLC/MAC

blocks and, in RLC acknowledged mode of operation, for the Backward Error Correction (BEC) procedures

enabling the selective retransmission of unsuccessfully delivered RLC/MAC blocks. In RLC acknowledged

mode of operation, the RLC function preserves the order of higher layer PDUs provided to it.

The RLC function also provides link adaptation.

In EGPRS, in RLC acknowledged mode of operation, the RLC function may provide Incremental

Redundancy (IR).

The MAC function defines the procedures that enabled multiple mobile stations to share a common

transmission medium, which may consist of several physical channels. The function may allow a mobile

station to use several physical channels in parallel, i.e., use several time slots within the TDMA frame.

For the mobile station originating access, the MAC function provides the procedures, including the

contention resolution procedures, for the arbitration between multiple mobile stations simultaneously

attempting to access the shared transmission medium.

For the mobile station terminating access, the MAC function provides the procedures to queue and

schedule access attempts.






1 � 1 � 20


Signaling Plane

MS BTS BSC

Um Abis Ater

GSM RFGSM-RF

LLC

GMM/SM

L2-RSLRelay

L1-RSL L1-GSL

L2-GSL

BSCGP

RR/RRM

L2-RSL

L1-RSL

Relay

MFS

RR

Gb

L1bis

NSL2-GSL

RLC

BSCGP

L1-GSL

RRMRelay

BSSGP

GTTP RR/GTTP

B10

GTTP: GPRS Transparent Transport Protocol:

� MS-BSS protocol

� Allows GMM signaling between MS and SGSN, without TBF establishment

� With GTTP, GMM messages are conveyed on main DCCH between MS and BSC, and on BSCGP between

BSC and MFS.

� GTTP possible only if :

� DTM capable MS in dedicated mode (but not in DTM)

� DTM enabled in the cell

� Signalling LLC PDU to be exchanged (Tbit set to “signaling”)

� LLC PDU length does not exceed MAX_LAPDm x (length of a LAPDm frame)






1 � 1 � 21

3 Main Transactions






1 � 1 � 22

3 Main Transactions

Session Management: Attach Procedure

� Aim: to access GPRS services, the MS must first make its presence known to the network by performing a “GPRS attach” procedure with the SGSN

� Results:

� a logical link between the MS and the SGSN is created

� the MS is in Standby state and may activate a PDP context

� the MS location is known (RA accuracy)

� the MS is available for PS paging via the SGSN

� A combined GPRS and IMSI attach is possible for class A/B MS

Each signaling procedure taking place between the MS and the GSS involves UL and DL TBFs. Each

GMM/SM message triggers the establishment of a TBF. The complete attachment procedure in GPRS may

involve up to 7 TBFs (4 UL and 3 DL).

These TBFs are usually short (a few RLC blocks), thus the optimization of the TBF establishment time is

very important, as well as the TBF establishment success rate.






1 � 1 � 23

3 Main Transactions

Session Management: PDP Context Activation

� Aim:

� in order to send and receive GPRS data, the MS must activate the PDP address it wants to use

� Results:

� the MS is known in the corresponding GGSN (the GGSN knows the SGSN where the MS is located) and data transmission with external data network can begin

The PDP context activation procedure is fairly close to the call establishment of the GSM including the

CCCH and the SDCCH phases (between the Channel Request message of the MS on RACH to the Alert

message).

It is important to control the overall PDP context activation duration for a good overview of the GPRS

QoS as it involves MS, MFS, SGSN, GGSN. The duration is longer than the GSM call establishment.






1 � 1 � 24

3 Main Transactions

Mobility Management: Definitions

� For GPRS, as paging is more frequent than in GSM, RAs have been defined smaller than Las.

� An RA is a subset of one and only one LA.

� The MS location in Standby state is known in the SGSN at the RA level.

� The MS is paged in its RA when MT traffic arrives at the SGSN.

� One RA is served by only one SGSN.

2 types of PCH/PPCH use shall be kept distinct for PS procedures:

� DL transfer establishment for MS in Ready State: mapping of the DL Immediate Assignment message

� DL transfer establishment for MS in Standby State (PS Paging Procedure): mapping of the Packet Paging

Request message

The first type will obviously be used quite often but does not affect the dimensioning of the RA (the

message is sent in one cell only). The second type of paging message is sent over the RA and will occur

more often that CS Paging as the transfer mode of most external servers is bursty, so the rate of arrival

of the PDU inside the SGSN is irregular. This is the reason why an RA shall be dimensioned smaller than

an LA if we want to achieve a battery use relative to GPRS Paging procedures equivalent to the GSM one.

For Mobility Management constraints in the CN, it is not recommended to split one RA over 2 LAs (the

routing information for a CN originated Paging Message being the BSC, a given BSC shall belong to a

unique LA as well as a unique RA).






1 � 1 � 25

3 Main Transactions

Mobility Management: MS States

"Idle"

GPRS Attach

Packet

Idle Mode

Packet

TransferMode

MS MM states

MS RR states

GPRS Detach

T_READY expiry PDU Transmission

"Ready"

"Standby"

Idle: the MS is not attached to the network: paging is not possible.

Standby:

� the MS is attached to the network: paging is possible.

� the MS location is known in the CN with the RA accuracy.

Ready:

� the MS location is known with the cell accuracy.

� timer T_READY keeps the MS in the Ready state just after data transfer.

Packet Idle Mode:

� no Temporary Block Flow exists. Upper layers can require the transfer of an LLC PDU which, implicitly, may trigger the

establishment of TBF and transition to packet transfer mode.

� the MS listens to the PBCCH and to the paging sub-channel for the paging group the MS belongs to in idle mode. If PCCCH is not

present in the cell, the mobile station listens to the BCCH and to the relevant paging sub-channels.

Packet Transfer Mode:

In packet transfer mode, the mobile station is allocated radio resource providing a Temporary Block Flow on one or more physical

channels. Continuous transfer of one or more LLC PDUs is possible. Concurrent TBFs may be established in opposite directions.

Transfer of LLC PDUs in RLC acknowledged or RLC unacknowledged mode is provided.

� When selecting a new cell, the mobile station leaves the packet transfer mode, enters the packet idle mode where it switches to

the new cell, reads the system information and may then resume to packet transfer mode in the new cell.

� The timers regulating the transition between states are SGSN timers, not tunable in the BSS.

Caution: Idle mode in GPRS and Idle mode in GSM are two different states.

� A GSM MS in Idle mode is attached to an MSC and can be paged.

� A GPRS MS in Idle mode is NOT attached to an SGSN, so it cannot be paged but can monitor the GPRS information broadcast in the

SI13 of the BCCH.

Standby is the closest GPRS MS state to Idle GSM.

The MS state in the SGSN shall be considered apart from the Packet Transfer Mode in the BSS:

� MS in Standby mode can be in Packet Transfer Mode.

� MS in Ready mode can be in Packet Idle Mode.

The detach procedure is usually triggered by the MS. Three other types of detach are triggered by the CN:

� HLR Detach,

� SGSN Detach upon SGSN overload,

� SGSN Detach upon timer.






1 � 1 � 26

� Dual Transfer Mode (DTM) enables a user to run Circuit Switched (CS) and Packet Services (PS) services simultaneously.

B10

So how was your holiday?

Very nice i’m sending you some pictures

I got it! You were in Lannion with your parents?

Yes it was very nice

� Like a class A mobile, but without the need of a dual receiver.

� Particular interest on 2G/3G dual mode handsets, for 3G to 2G continuity of service purpose.

3 Main Transactions

Simultaneous CS and PS services: DTM

3 GPP:

� Feature part of Rel.99 specifications

� Optional for MS, BSS, CN

� Impacts on MS, BSS, MSC

� Improvements in Release 4, 6 and 7

B10 implementation aligned with Release 4 definition of the feature. Release 6 and 7 improvements not

implemented in B10.






1 � 1 � 27

3 Main Transactions

Dual Transfer Mode Activation

� Activation

� The activation of the DTM is done by the cell parameter (optional):

� EN_DTM = enabled

� Defaut value = Disabled

� Activation is possible:

� In Evolium cells

� In cells with a minimum number of resources

� MAX_PDCH_HIGH_LOAD > 1 (at least 2 TSs are needed for DTM resource allocation)

� In non-extended cells (to simplify implementation)

� Gs interface is highly recommended (but not strictly mandatory) for paging coordination.

B10






1 � 1 � 28

Class A (DTM)Class B

Packet Transfer

Release Last TBF (1)

Idle/Packet Idle

Dedicated

Cs Setup (2)

Dual Transfer

Packet

Request (3)

Release Last TBF (6)

Packet Request(5)

Cs Release (7)

Cs and Ps Release (4)

Scenario 1

Scenario 2

Scenario 3

3 Main Transactions

DTM Principles B10

Note that the system does not support a transition directly between packet transfer mode an DTM in

either direction.

Scenario 1: Entering the DTM mode for a Ms already in Packet Transfer receiving a Cs call request

1. Release of the TBF.

2. CS establishment.

3. A new TBF is established, the Ms enters into the DTM mode.

Scenario 2: Releasing the CS connection in DTM

1. Both CS and PS services are released.

2. New TBF is re established in Packet Transfer mode by the MS or by the network.

Scenario 3: Releasing the PS connection in DTM, then releasing the CS connection

1. When the last TBF is released for an Ms in DTM, the CS call continues in dedicated mode.

2. After releasing the CS call, the MS enters the idle mode.






1 � 1 � 29

3 Main Transactions

DTM Multi slot Classes B10

0 1 2 3 4 5 6 7

UL

0 1 2 3 4 5 6 7

Ttb Tra

DL

P T

P T

0 1 2 3 4 5 6 7

UL

0 1 2 3 4 5 6 7

Ttb Tra

DL

P T

P T P

P: PDCH

T: TCH

P: PDCH

T: TCH

� Class 5 (2+2): 1 TCH + 1 PDCH

� Class 9 (3+2): 1 TCH + 2 PDCH

The usage of Extended Dynamic Allocation in UL together with DTM is only allowed in the case of a DTM

[E]GPRS multislot Class 11.

All the other configurations don’t need usage of the Extended Dynamic Allocation

Note: C means « configurations with contiguous TSs ».

Class 11 (2+3) with EDA: 1 TCH + 2 PDCH






1 � 1 � 30

3 Main Transactions

Mobility Management: MS Location Management

The MS enters a new

cell

New cell inside the current

RA

MS in Ready state

Cell update

New cell belongs to a new

RA

RA update

New cell belongs to a new LA

RA/LA update

Only in NMO I

When the MS is in Ready State, it performs a “Cell Update”.

� The MS sends any LLC frame in the new cell with its TLLI in the header.

� The Cell and RAC information is added by the BSSGP at the programming of the BSSGP frame.

RA Update:

� The MS sends an RA Update Request message containing the identity of the MS, the old RAI and the

Update Type. The update type is either enter a new RA or periodical RA update.

� The BSS adds the cell global Identity when transferring the message into a BSSGP frame towards the

SGSN.






1 � 1 � 31

3 Main Transactions

Interactions Between the SGSN and the MSC/VLR

� The Gs interface between the MSC and the SGSN is needed

� Following actions are possible:

� IMSI attach/detach via the SGSN

� LA update via the SGSN

� CS paging via the SGSN

The Gs interface carries signaling between the P-VLR (SGSN) and the VLR (MSC). Whether Gs is provided

or not, it does NOT belong to the BSS release as it is a CN feature.

The presence of the Gs interface is given by the NMO information inside the SI13 and the CN feature in

the PSI1.

The Gs interface carries only MAP signaling between the P-VLR and the VLR of the MSC.






1 � 1 � 32

3 Main Transactions

Network Mode of Operation

PCH PCH no Gs interfaceII

PCH NA no Gs interface

PCH PPCH no Gs interface

III

Packet idle mode

Packet transfer mode

Mode CS paging channel PS paging channel Remarks

PCH PCH Gs interface

PACCH NA Gs interface

I

Packet idle mode

Packet transfer mode

B10

NA Main DCCH Gs interface

NA NA Gs interface

Dedicated mode

Dual transfer mode

All the possible combinations with the MPDCH are:

� NMOIII,

� NMOI with MPDCH.

According to the NMO offered and the packet mode of the MS (Packet Transfer Mode or Packet Idle

Mode), the routing of the PS paging and the CS paging changes.

The NMO setting is done from the OMC-R via the NETWORK_OPERATION_MODE parameter.

In case of DTM activation, Gs interface is highly recommended (but not strictly mandatory) for paging

coordination.

When the DTM feature is enabled, and when NETWORK_OPERATION_MODE = NMO I, the BSS shall use the

following channels to page a DTM MS engaged in a CS or PS connection:

� The PACCH to page a DTM mobile for a CS connection if this mobile has already a PS connection

ongoing. In this case, the MS is in Packet Transfer Mode.

� The main DCCH (FACCH or SDCCH) to page a DTM mobile for a PS connection if this mobile has already

a CS connection on-going. In this case, the MS is in dedicated mode.






1 � 1 � 33

3 Main Transactions

CS/PS Paging During a Packet Transfer/Call

� CS paging during packet transfer (class B MS):

� according to the GSM standard, a class B MS may or may not (implementation dependent) listen to the PCH channel during a packet transfer

� if the MS listens to the PCH channel:

� some RLC blocks are lost

� the MS receives all the CS paging messages

� the MS can start a CS call

� at the end of the call, the MS triggers an RA updating procedure

� Refer to Suspend/Resume for Class B MS

The Class A and Class C GPRS mobiles do not face these problems. The former is able to handle

simultaneously CS and PS traffic (no traffic disruption), whereas the latter is not reachable in one

domain while it is attached to the other domain.

The Class B situation is the most complex one and shall be considered since Class B GPRS mobiles are the

most popular for the manufacturers. It shall be remembered that the most important service in the GSM

network where GPRS is activated is the speech. Therefore, an MS shall be given the opportunity to listen

to any CS paging. It is then up to the operator and to the MS to decide whether or not the CS call shall be

answered.






1 � 1 � 34

3 Main Transactions

CS/PS Paging During a Packet Transfer/Call [cont.]

� Suspend/Resume for Class B MS:

1. Suspend (TLLI, RAI)2 Suspend

6 Resume (TLLI, RAI, SRN)

5 Suspend Ack

8 Resume Ack

3 Suspend (TLLI, RAI, suspend cause)

7 Resume

11 Routing Area Update Request

10 Channel Release

9 Resume Ack

MS BSC MFS SGSN

4 Suspend Ack (TLLI, RAI, SRN)

End of on-going TBF

DL LLC PDUs are discarded

Normally, no more paging

messages sent by the SGSNEnd of the GSM call

The MS leaves the GSM dedicated mode

The MS listens to “Channel Release” message content

The MS enters in

GSM dedicated

Mode (Ongoing GPRStransfer or not)

Suspend Reference Number (SRN)

1) The GPRS suspension procedure is initiated by the MS by sending an RR Suspend (TLLI, RAI, suspension cause) message to the BSC.

2) The BSC sends a Suspend (TLLI, RAI, suspension cause) message to the MFS, via the GSL link. The BSC shall store TLLI and RAI in

order to be able to request the SGSN (via the MFS) to resume GPRS services when the MS leaves dedicated mode.

3) The MFS sends a Suspend (TLLI, RAI) message to the SGSN.

4) The MFS receives a Suspend Ack from the SGSN, in which there is a Suspend Reference Number which will have to be used in the

resume step.

5) The MFS sends a suspend acknowledgement to the BSC, with the Suspend Reference Number information.

6) The BSC determines that the circuit-switched radio channel shall be released (typically upon circuit-switched call completion). If

the BSC is able to request the SGSN to resume GPRS services (i.e., the suspend procedure succeeded and the BSC received the

Suspend Reference Number, no external handover has occurred), the BSC shall send a Resume (TLLI, RAI, Suspend Reference

Number) message to the MFS.

7) Upon receipt of a Resume message from the BSC, the MFS sends a Resume (TLLI, RAI, Suspend Reference Number) message to the

SGSN.

8) The MFS receives a Resume Ack from the SGSN.

9) Upon receipt of the Resume Ack from the SGSN, the MFS sends a Resume Ack message to the BSC.

10) The BSC sends an RR Channel Release (GPRS Resumption) message to the MS and deletes its suspend/resume context.

11) The MS resumes GPRS services by sending a Routing Area Update Request message in the following cases:

� reception of a Channel Release with GPRS Resumption = NOK

� reception of a Channel Release without GPRS Resumption IE

� T3240 expiry

Alcatel-Lucent BSS does not support the suspend/resume procedure in case of inter-BSC reselection. In this case, MS shall resume

the GPRS service by sending a Routing Area Update Request message to the SGSN.

When a mobile station operating in class B mode of operation suspends its GPRS activity during circuit-switched activities, RRM

forces the release of the on-going TBF(s). Once the TBF(s) has(ve) been released, the MS context is kept for T_MS_Context_Lifetime

seconds before being deleted.

T_MS_Context_Lifetime is an MFS parameter. The default value is 120s and it cannot be set at OMC-R level.






1 � 1 � 35

3 Main Transactions


� The Suspend/Resume feature is used for a Class A DTM MS

in the following cases:

� When a Class A DTM MS is handed over to a cell not supporting DTM

� After the Handover procedure is completed, the MS sends a suspend message with the suspension cause “DTM not supported in the cell”

� When a GPRS attached MS is in a cell that does not support DTM and a Cs service is initiated

� After the CS access to the cell, the MS sends a suspend message with the suspension cause “DTM not supported in the cell”

B10

In case two, the GPRS suspension procedure is initiated by the mobile station by sending a GPRS suspend

message with the suspension cause set to “DTM not supported in the cell”. This can be done as early as

possible after access but shall be done afetr sending a CLASSMARK CHANGE message.






1 � 1 � 36

3 Main Transactions


� CS paging during packet transfer (class B MS):

� if the MS does not listen to the PCH channel:

� CS paging messages are lost if the duration of the packet transfer is higher than the duration of the repetition of CS paging messages

� if Gs interface is available:

� CS paging messages are sent via the PACCH channel

� PS paging during a GSM call (class B MS):

� the MS does not receive PS paging messages

If CS paging repetitions are viewed as a GSM QoS problem, the situation shall be avoided.

A missed PS Paging during a CS call is less of a problem as the traffic in GPRS is under the influence of

the GSS capacity to store the PDU originated from an external Packet Data Network. When the delay of

transfer is not a requirement for the service, there is no direct impact of a missed MS Paging.






1 � 1 � 37

3 Main Transactions

PS Paging for DTM MS in dedicated mode [cont.]

� PS paging during packet transfer (class A DTM MS):

B10

MS BSC MFS SGSN

MS in dedicated mode

PS PAGING

DTM Packet notification

DTM Packet notification

Ack

Packet notification

GPRS Information[UL LLC PDU]

DTM GPRS Information[UL LLC PDU]

UL LLC PDU

Start T_Wait_DTM

Stop T_Wait_DTM

Signalling carried on GTTP if possible

(1)

(2)

(3)(4)

(5)

(6)

1. The DTM MS is in dedicated mode.

2. As the MS is in GMM Standby state, the SGSN shall first page the MS before sending Data.

3. Upon receipt of the PS paging, the MFS checks wether the concerned MS is in dedicated mode. If it is

the case, the MFS sends to the BSC a BSCGP DTM Packet notification message providing the BSC with

the reference of the MS (i.e. IMSI and BSC reference) and the contents of the 44.018 Packet

Notification message and starts T_Wait_DTM, to monitor the reception of acknowledgement from BSC.

The BSC forwards the Packet Notification message to the MS on the main DCCH, and sends a DTM

Packet Notification Ack message to the MFS. The MFS stops T_Wait_DTM and delete the paging (paging

is considered as being successfully sent to the MS).

4.to 6. On receipt of the Packet Notification message, the MS shall answer to the notification with a cell

update procedure, I.e. by sending an LLC Frame.

� If the LLC frame does not convey user data information and can be transmitted through Gprs

Transparent Transport Protocol (GTTP) in less than MAX_LAPDm frames, the MS uses the GTTP

to send the UL LLC frame to the the SGSN.

� Otherwise the MS requests the establishment of an UL TBF with a 44.018 DTM request

message and sends the UL LLC PDU on the established UL TBF






1 � 1 � 38

B103 Main Transactions

Support of GTTP [cont.]

� GTTP allows GMM signaling between MS and SGSN, without TBF establishment

� With GTTP, GMM messages are conveyed on main DCCH between MS and BSC, and on BSCGP between BSC and MFS.

� GTTP is possible if:

� DTM capable MS in dedicated mode (but not in DTM)

� DTM enabled in the cell

� Signaling LLC PDU to be exchanged (i.e. “RA update”)

� LLC PDU length does not exceed MAX_LAPDm x (length of a LAPDm frame)

� Description of the GTTP:

While in dedicated mode or in DTM, upper layers in the mobile station or in the BSS may request the

transport of GPRS information transparently over the radio interface.

This procedure is possible only if:

� the information from upper layers is signalling information

� the GTTP length of the message is below the maximum indicated by the network, iI.e. it can fit in less

than MAX_LAPDm frames.

In other case, the MS or the BSS shall initiate the establishment of a TBF (I.e. enter DTM mode).

The information from upper layers shall be carried inside the GTTP Information message. The GTTP

Information message contains:

� TLLI of the MS

� LLC PDU






1 � 1 � 39

3 Main Transactions

TBF Establishments

Data Transfer establishment

UL TBF establishment DL TBF establishment

MS in PIM MS in PTM MS in PIMMS in PTM

MS in MM Ready state MS in MM Standby stateMS in MM Ready state

UL TBF running

DL TBF running

T3192 running

PS Paging

on CCCH

2-Phase

1-Phase

on CCCH

Non-DRX

DRX

There are 3 types of UL TBF establishment:

1. 1-Phase access on CCCH when the MS is in Packet Idle Mode and the MS does not need more than 1

PDCH and wants to transfer blocks in RLC acknowledged mode.

2. 2-Phase access on CCCH: when the MS is in Packet Idle Mode and the MS needs more than 1 PDCH or

wants to transfer blocks in RLC unacknowledged mode.

3. During a DL TBF: when the MS is in Packet Transfer Mode in the DL.

There are 4 types of DL TBF establishment:

1. DRX mode on CCCH: when the MS is in Packet Idle Mode and the MS is listening to all PCH channels of

its CS paging group.

2. Non-DRX mode on CCCH: when the MS is in Packet Idle Mode and the MS is listening to all AGCH

channels.

3. During a UL TBF: when the MS is in Packet Transfer Mode in the UL.

4. When T3192 is running: when a DL TBF has been released at the MS side and before the previously

used radio resources are released (at T3192 expiry).






1 � 1 � 40

Self-assessment on the Objectives

� Please be reminded to fill in the formSelf-Assessment on the Objectivesfor this module

� The form can be found in the first partof this course documentation






1 � 1 � 41

End of ModulePrinciples






Module 2Radio Resource management









EVOLIUM � E-GPRS Radio Algorithms and Parameters Description B10Radio Algorithms � Radio Resource management

1 � 2 � 2

Blank Page




Document History






1 � 2 � 3

Module Objectives


� Describe the algorithms of Resource Management and the related parameters






1 � 2 � 4








1 � 2 � 5

Table of Contents


1 (E)GPRS Channels 72 Autonomous Packet Resource Allocation 223 TBF Radio Resources Allocation and Re-Allocation 574 TBF Release Routine 78






1 � 2 � 6








1 � 2 � 7

1 (E)GPRS Channels






1 � 2 � 8

1 (E)GPRS Channels

Overview

PDCH Slave PDCH

PTCCH

PDTCH

PACCH

PTCH

physical channel

control channel

traffic channel

signaling associated control channel

logical channel category

logical channel






1 � 2 � 9

1 (E)GPRS Channels

GPRS Physical Channel, PDCH

0 7 0 7 0 7

1 TDMA frame = 4.615 ms

0 1 2 49 50 51

The 52-multiframe= 240 ms

X X Block

Frame0 4 8 1213 17 21 25 26 30 34 3839 43 47 51

PTCCH

B0 B1 B2 B3 B4 B5 B9 B10 B11B6 B7 B8

� Packet Data Channel (PDCH): a physical channel which carries GPRS logical channels

PDCH frame:

� Made up of 52 TSs of the same rank belonging to 52 consecutive TDMA frames.

� The 52 TSs are divided into blocks of 4 consecutive TSs.

� 12 blocks are created and 4 single TSs:

� TS12 and TS38 for the Timing Advance,

� TS25 and TS51 are pseudo Idle TS for transmission purposes (synchronization with the occurrences of

SACCH on the GSM 26-multiframe).

According to the PDCH design, a maximum of 8 PDCHs can be created with one TRX.

A PDCH can be entirely allocated to a single user, which is close to the principle of circuit in GSM.






1 � 2 � 10

1 (E)GPRS Channels

GPRS Physical Channel, PDCH [cont.]

� 1 RLC PDU uses 1 PDCH block

� Except in MCS-7 to MCS-9 where 2 RLC PDUs use 1 PDCH block

� 2 kinds of physical channel PDCH

� master PDCH (MPDCH)

� A PDCH which carries the PCCCH and PBCCH logical channels

� For signaling purpose

� slave PDCH (SPDCH)

� A PDCH which carries PTCH logical channels

� For traffic purpose

The MPDCH has been introduced with the B7 release (details are given in the annex).

The use of GSM CCCH for the GPRS traffic offer can lead to QoS problem in GSM (PCH use more specifically).






1 � 2 � 11

1 (E)GPRS Channels

(E)GPRS Logical Channels

� Different GPRS logical channels mapped on PDCH, which are shared on a block basis:

� PTCH: PDTCH and PACCH

� Packet Timing advance Control Channel (PTCCH)






1 � 2 � 12

1 (E)GPRS Channels

(E)GPRS Logical Channels [cont.]

� Packet Traffic Channel (PTCH):

� used for user data and associated signaling transmission

� Packet Data Traffic Channel (PDTCH):

� unidirectional channel used for user data transmission

� mapped on one PDCH

� up to 8 PDTCHs may be allocated to an MS on different PDCHs with the same frequency parameters

Issue: the network shall control the multiplexing of several users on a unique UL PDCH avoiding collision

occurrence. This is achieved by the RLC/MAC functions and the use of USF, RRBP and TFI fields of the

RLC/MAC header.

The number of PDTCHs allocated to one MS belongs to:

� The MS capabilities (multislot class),

� The traffic in the BSS,

� Operator configuration of the BSS parameters.

NB: max number of PDTCHs to one MS = 8 because the MS has to be allocated TSs on a unique TRX, and one

TRX can support 8 PDCHs max.






1 � 2 � 13

1 (E)GPRS Channels


� Packet Traffic Channel (PTCH):

� Packet Associated Control Channel (PACCH):

� A bidirectional channel used to transmit control and acknowledgement messages

� Mapped on one PDCH:� if a single PDTCH is allocated to an MS, the PACCH is allocated on the PDCH carrying the PDTCH

� if multiple PDTCHs are allocated to an MS, the PACCH is allocated on one of the PDCH carrying the PDTCHs (Alcatel BSS)

Caution: PACCH blocks are used to carry the BSS signaling but not the GSS signaling.

The scheduling of PACCH blocks in the UL and the DL is monitored by the MFS. The most frequent use of the

PACCH blocks is for “Packet Ack/Nack” messages.

It can be used as well for CS Paging message when Master Channels are not available.

It is necessary for the MS to update the PSI13 on a regular basis in order to achieve proper RLSs and Power

Control mechanisms. The PSI13 content can be sent to the MS in Packet Transfer Mode via a PACCH.






1 � 2 � 14

1 (E)GPRS Channels


� Packet Timing advance Control Channel (PTCCH):

� bidirectional channel (DL: TA messages; UL: Access Burst for TA calculation) used by the continuous timing advance mechanism

� the PTCCH of one MS is carried by the PDCH carrying the PACCH

� Timing Advance Index (TAI), used for the scheduling of the AB, is part of the radio resources allocated to an MS.

� The TAI is a PDCH parameter

� The TAI takes 16 values

MFS

Access

Burst

TA Mes

sage

The Access Burst in the UL and the Timing Advance Messages in the DL are scheduled in time manner on

TS12 and TS38.

The TAI is part of the GPRS radio resources allocated by the MFS to the MS. Each mobile needs to have a

TAI.

The TAI range value is a limitation to MS multiplexing on a same TS, as both MS in the UL transfer and MS in

the DL transfer send their AB in the UL and receive their TA value in the DL.

16 values for TAI means that each MS sends an AB every 1.96 s, when the content of the TA Messages is

updated every 480 ms (every 4 occurrences of TAM).

IdemTAM 112N+24

idemTAM 438N+715

………………

TAM 038N+13

TAM 012N+12

TAM 038N1 4 repetitions of the 16 TA values (4 TA values updated)

TAM 012N0

TA MessageOn PTCCH TS

numberA.B. scheduled for

MF-51TAI value






1 � 2 � 15

1 (E)GPRS Channels

Multiplexing of (E)GPRS Logical Channels

� Temporary Block Flow (TBF): a unidirectional flow of data between the MS and the MFS for the transfer of one or more LLC PDUs (refer to GSM 04.60)

� Several TBFs can be transmitted on one PDCH (TFI 5)

�One TBF can be served on several PDCHs (TFI 17 & 24)

� A TBF is identified by a Temporary Flow Identity (TFI)

PDCH 1

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

PDCH 2

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

PDCH 3

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

TBF with TFI = 5 TBF with TFI = 17 TBF with TFI = 24

Temporary Flow Identity (TFI): Each TBF is assigned a TFI by the MFS.

Important: It is possible to establish 32 TBFs per TRX.

TBF: a group of blocks dynamically allocated to one MS for one transfer of RLC blocks in one direction inside

one cell.

A Temporary Block Flow is a temporary, unidirectional physical connection across the Um interface,

between one mobile and the BSS. The TBF is established when data units are to be transmitted across the

Um interface and is released as soon as the transmission is completed.

There is still a 3 RTS shift between Rx and Tx, on the TDMA frame.






1 � 2 � 16

� Downlink PDTCH and PACCH blocks multiplexing

� Uplink PDTCH and PACCH for a UL TBF multiplexing

1 (E)GPRS Channels

Multiplexing of (E)GPRS Logical Channels [cont.]

PDTCH

TFI24

USF = 5

PDTCH

TFI17

PACCH

TFI24

TFI UL

PDTCH/

PACCH

+1

DL PDCH N°2

UL PDCH N°2

Downlink PDTCH and PACCH blocks multiplexing:

� The multiplexing of the different MSs is performed thanks to the TFI which is present in the RLC block

header.

� An MS decodes all the blocks of all its allocated PDCHs and keeps the blocks carrying its TFI in the RLC

header.

Uplink PDTCH and PACCH for a UL TBF:

� At UL TBF establishment, an MS receives a USF (Uplink State Flag, 8 values, MAC header) per allocated

PDCH.

� If the MS receives its USF on the downlink block n of PDCH i, it can transmit in uplink using the block n+1

of PDCH i.

� This is the principle of the Dynamic Allocation, which allows a maximum of two TSs in Uplink

NB: the values of the USF are entirely dedicated to PDTCH and PACCH transfers. See further (MPDCH and

RRBP)

The TFI is used in the UL as well: each mobile shall put its TFI in the UL header of the UL blocks during a UL

TBF, as well as in the RLC header of the UL PACCH blocks of a DL TBF.

So we can say that the de-multiplexing of the blocks is achieved by the use of a TFI.






1 � 2 � 17

1 (E)GPRS Channels

Multiplexing of (E)GPRS Logical Channels [cont.]

� Uplink PACCH for a DL TBF scheduling:

S/P

≠≠ ≠≠false

PDTCH

TFI24

Packet DL

Ack/NAckmessage

PACCH

TFI24

USF =

000

PACCH

TFIXX

USF = 5

PDTCH

TFI17

RRBP = +3

Ø

TFI UL

PDTCH/

PACCH

DL PDCH N°2

UL PDCH N°2Exercise

RRBP: Relative Radio Block Period

Allocation of a PACCH block for the sending of acknowledgements in the UL of blocks received in the DL:

� The MS has no USF because it is involved in a DL TBF

� Use of the RRBP field transmitted in the downlink (MAC header) in association with the TFI of the DL TBF

in the RLC header.

� At the exact occurrence of the RRBP, a special USF value is used for the UL TBF taking place on the same

PDCH: USF=no emission.

It is a semi-boolean parameter. The RRBP field of an RLC/LAC block is checked each time by the MS whose

TFI is written in the RLC header.

� When S/P is false, no UL PACCH is scheduled.

� When the RRBP field is valid, the value gives the number of blocks to wait before sending its PACCH block

in the UL.

S/P is false means the MS has to send an acknowledgement message to the MFS.






1 � 2 � 18

1 (E)GPRS Channels

Extended Dynamic Allocation B10

� EDA allows higher throughput in uplink through the support of more than two TSs

� Principles

� Same basic principles as Dynamic Allocation (i.e. based on USF)

� Differences

� A mobile station detecting its assigned USF value on one assigned PDCH is allowed to transmit on that PDCH and all higher numbered assigned PDCHs

� The mobile station needs not to monitor all the downlink PDCH corresponding to its uplink PDCH allocated

B10 MR2

The EDA feature is optional for the network.






1 � 2 � 19

1 (E)GPRS Channels

Extended Dynamic Allocation [cont.] B10

� USF scheduling for

EDA mode:

DL TS0 TS1 TS2 TS3

BP n

UL

TS0 TS1 TS2 TS3

DL TS0 TS1 TS2 TS3

BP n+1

UL

TS0 TS1 TS2 TS3

DL TS0 TS1 TS2 TS3

BP n+2

UL

TS0 TS1 TS2 TS3

DL TS0 TS1 TS2 TS3

BP n+3

UL

TS0 TS1 TS2 TS3

DL TS0 TS1 TS2 TS3

BP n+4

UL

TS0 TS1 TS2 TS3

DL TS0 TS1 TS2 TS3

BP n+5

UL

TS0 TS1 TS2 TS3






1 � 2 � 20

1 (E)GPRS Channels

Extended Dynamic Allocation Usage [cont.] B10

� Uplink Throughput increase:

� Class 11 (2+3):

0 1 2 3 4 5 6 7

UL

0 1 2 3 4 5 6 7

Ttb Tra

DL

Tx Tx Tx

Rx Rx Mx

� Class 12 (1+4):

0 1 2 3 4 5 6 7

UL

0 1 2 3 4 5 6 7

Tta Trb

DL

Tx Tx Tx

Mx

x Rx

Tx






1 � 2 � 21

1 (E)GPRS Channels

Extended Dynamic Allocation Usage [cont.] B10

� EN_EDA: disabled (default Alcatel value).

� Changeable at OMC-R on Cell level

� EDA_MS_ACTIVATION_LEVEL: activation level

� Gives an artificial mean to limit the activation domain (inter-operabilityissues)

� Possibility is given by the parameter ALLOW_DTM_EDA_COMBINATIONto allow DTM and EDA combination

� (i.e. support of 2+3 configuration for DTM multislot class 11)

EDA_MS_ACTIVATION_LEVEL and ALLOW_DTM_EDA_COMBINATION are relevant if EN_EDA = 1.

EDA_MS_ACTIVATION_LEVEL=0 (Alcatel default value) and it can be set at the OMC-R (BSS Level)

ALLOW_DTM_EDA_COMBINATION = 0 (Alcatel default value) and it can be set at the OMC-R ( BSS Level)

Note : none of the combination of the three paramters allows activating EDA only in case of DTM operations

The Ms is supporting EDA mode depending on the following PCC flag:

EDA_Allowed_GPRS (respectively EDA_Allowed_EGRPS) which is set to true if:

� The cell belongs to an Evolium BTS

� The cell is not an extend cell

� EN_EDA parameter is activated

� EDA is supported by the MS in GPRS (respectively EGPRS)

� EDA is enabled in the BSS ( EDA_MS_ACTIVATION_LEVEL parameter)

� The N_UL_BIAS_FOR_EDA last observed biases since the transfer start are uplink

� There is no RT PFC associated to the MS






1 � 2 � 22

2 Autonomous Packet Resource Allocation






1 � 2 � 23


Definitions: SPDCH Group

� SPDCH group = time slots usable for PS traffic

� Only 1 SPDCH group per TRX

� contains TS

� belonging to the same TRX

� having the same frequency configuration

� without hole

� MPDCHs are not part of the SPDCH group

� Up to 16 PDCH groups per cell

� A TRX is PS/CS capable only if TRX_PREF_MARK = 0

� TRX_PREF_MARK <> 0 then the TRX is only CS capable

Any TRX should possibly support one SPDCH group except for one case: Concentric cell or multi-band cell

design, an SPDCH group can NOT belong to the inner zone.

An SPDCH group can be supported by both hopping and non-hopping TRXs.

Only one Mobile Allocation (MA) is supported in a cell.






1 � 2 � 24


Definitions: SPDCH Group [cont.]

� Example: BBH is used, NB_TS_MPDCH=0

0 1 2 3 4 5 6 7

TRX1

TRX2

TRX3

BCCH SDCCH

SDCCH

TRX4

TRX_PREF_MARK

0

1

0

1

SPDCH group1

SPDCH group2

SPDCH group3

SPDCH group4

SPDCH group2 Null

SPDCH group4 Null

CAUTION: animated slide.






1 � 2 � 25


Definitions: 8-PSK High Power Capability

� GMSK

� constant envelope(amplitude) modulation

� one bit per modulated symbol over the radio path

� 8-PSK

� Envelope modulation not constant

� 3 bits per modulated symbol over the radio path dB

(147 bits)

PN

0

-20

dB

t

(147 bits)

542.8 µs

PN

dB

t

(147 bits)

542.8 µs

PN

Q

I

1,1,1

0,1,1

0,1,0

0,0,0

0,0,1

1,0,1

1,0,0

1,1,0

GMSK = the Gaussian Minimum Shift Keying belongs to a subset of phase modulations

8-PSK = 8-state Phase Shift Keying

� 8-PSK is not a constant envelope modulation. Part of the information is conveyed by the amplitude of the carrier

which varies over time.

� An 8-PSK signal carries three bits per modulated symbol over the radio path, which allows to triple the data

transmission rates.

Modulation Gross Bit Rate

� The normal burst is divided into 156.25 symbol periods. A normal burst has a duration of 3/5.2 seconds (577 µs).

(3GPP TS 05.02).

� For GMSK modulation, a symbol is equivalent to a bit (3GPP TS 05.04).

� A GMSK burst is composed of 156.25 bits (6 tail bits + 26 training sequence bits + 116 encrypted bits + 8.25 guard

period (bits)).

� Modulation gross bit rate = (156.25 bits) / (3/5.2 seconds) = 270 Kbit/s

� For 8-PSK modulation, one symbol corresponds to three bits (3GPP TS 05.04).

� An 8-PSK burst is composed of 156.25 x 3 = 468.75 bits (18 tail bits + 78 training sequence bits + 348 encrypted

bits + 24.75 guard period (bits)).

� Modulation gross bit rate = (468.75 bits) / (3/5.2 seconds) = 810 Kbit/s.

Amplitude variesconstantCarrier envelope

EGPRSGPRS / EGPRSPacket radio service

810 Kbit/s270 Kbit/sGross bit rate per carrier

200 KHz200 KHzChannel spacing

Phase modulationFrequency modulationModulation type

8-PSKGMSK






1 � 2 � 26


Definitions: 8-PSK High Power Capability [cont.]

�G4 TRE characteristics

30 W

44,77 dBm

25 W

43,98 dBm

30 W

44,77 dBm

12 W

40,79 dBm

30 W

44,77 dBm

25 W

43,98 dBm

30 W

44,77 dBm

15 W

41,76 dBm

8-PSK output power

60 W

47,78 dBm

35 W

45,44 dBm

60 W

47,78 dBm

45 W

46,53 dBm

GMSK output power

1800900Frequency

band

High powerMedium powerHigh powerMedium powerType

TADHE

EDGE+

TADHTRADE

EDGE+

TRADTAGHE

EDGE+

TAGHTRAGE

EDGE+

TRAG

G3 TREs are not able to handle the 8-PSK modulation. Only G4 TREs (also called TRA) are EDGE capable.

The TRA sensitivity is as follows:

� GMSK: - 111 dBm.

� 8-PSK: - 108 dBm for MCS5, - 99 dBm for MCS9.






1 � 2 � 27


Definitions: 8-PSK High Power Capability [cont.]

� The GMSK output power is:

� The minimum value among the maximum TRE output power in:

� One cell

� One frequency band

� The maximum output power in the cell

� The 8-PSK output power

� Is given for one TRE by the modulation_delta_power

� = [Maximum output power in the cell] – [8-PSK TRE output power]

� And a possible attenuation, in order not to exceed the GMSK power in the cell (case of BS_TXPWR_MAX <> 0 dB)

� 8-PSK High Power Capability

� G4 HP TRE if the modulation_delta_power < 3dB

� G4 MP TRE if the modulation_delta_power > 3dB

� The 8-PSK High Power Capability is used in the TRX ranking process

CAUTION: do not confuse “MP” and “HP” given as a TRE hardware characteristic and “MP” and “HP”

defined by the 8-PSK High Power Capability.

� The first ones represent the maximum power a TRE can transmit (in GMSK or 8-PSK).

� The second ones represent the capability of a TRE to emit with an 8-PSK power close to the maximum

GMSK power in a given cell.






1 � 2 � 28

� Input / Output


Definitions: TRX Ranking

TRX Ranking

function

8-PSK High Power Capability (G4 HP, G4 MP, or G3)

TRX Ranking Table

DR TRE capability (FR or DR)

Inputs Outputs

TRX_PREF_MARK

PS_PREF_BCCH_TRX

Radio cell configuration (nb of SPDCHs per TRX)

TRX id (0,1,…,14, or 15)

TRE

Characteristics

TRX or cell

Characteristics

TRX Rank

E-GSM, P-GSM, GSM850 or DCS TRX

Once the TRXs are mapped to the TREs (after a possible TRX adjustment), the TRX Ranking function is

computed. This function consists in ranking the TRXs in order to ensure that the CS and PS allocations will

be consistent.

HP means there is less than 3 dB of difference between the maximum power of the GMSK TRXs and the

maximum power of the 8-PSK TRXs in the cell. On the contrary, MP means the difference is more than 3 dB.






1 � 2 � 29

� Ranking criteria (from the highest preference for PS allocation to the lowest)

� Concerns only PS capable TRX: TRX_PREF_MARK = 0

� PS_PREF_BCCH_TRX

� = 0 (no specific preference) then the BCCH TRX is managed as a non-BCCH TRX

� = 1 then the BCCH TRX has the highest preference

� = 2 then the BCCH TRX has the lowest preference (B9 MR4)

� HW TRE capability: G4 HP > G4 MP > G3

� DR TRE capability: FR > DR

� E-GSM TRX preference: E-GSM TRXs > P-GSM/GSM850/DCS TRXs

� TRX(s) having the maximum SPDCH group size

� TRX having the lowest TRX index


Definitions: TRX Ranking [cont.]

Exercise

The TRX with the lowest Rank value has the highest preference for PS allocations.

Step 01: Separation of non-PS and PS capable TRXs

The first step consists in removing the non-PS capable TRXs, i.e., the TRXs having a non-null TRX_PREF_MARK. Only the PS capable

TRXs are then kept for the TRX ranking.

Step 02: Ranking of the PS capable TRXs

� First criterion: PS allocations preferred on the BCCH TRX

� If the BCCH TRX is a PS capable TRX and the parameter PS_PREF_BCCH_TRX is set to ‘2’, then the BCCH TRX has the highest

rank in the TRX Ranking Table. This means that this TRX is selected last for PS allocations.

� If the BCCH TRX is a PS capable TRX and the parameter PS_PREF_BCCH_TRX is set to ‘1’, then the BCCH TRX has the lowest

rank (i.e., Rank 1) in the TRX Ranking Table. This means that this TRX is selected first for PS allocations.

� If the parameter PS_PREF_BCCH_TRX is set to ‘0’, then the rank of the BCCH TRX is determined by the remaining criteria.

� Second criterion: HW TRE capability

� The TRXs mapped on a G4HP TRE are ranked first, then they are followed by the TRXs mapped on a G4MP TRE, finally by the

TRXs mapped on a G3 TRE.

� Third criterion: DR TRE capability

� Among the TRXs having the same HW TRE capability, select first the FR TRX, then the DR TRX.

� Fourth criterion:E-GSM TRX preference

� Among the TRXs having the same HW TRE capability and the same DR TRE capability, select first the E-GSM TRXs, then the P-

GSM/GSM850/DCS TRXs.

� Fifth criterion: Maximum SPDCH group criterion

� Among the TRX having the same HW TRE capability, the same DR TRE capability and the same E-GSM TRX preference, rank first

the TRX having the maximum number of consecutive SPDCHs per TRX. Note that this is a static information given by the O&M

configuration of the TRX.

� Sixth criterion: TRX identity

� Among the remaining TRXs, select first the TRX having the lowest TRX id. This criterion aims at having a deterministic criterion

at the end of the TRX Ranking function.

Note:

The TRX ranking function does not take into account the current traffic load.






1 � 2 � 30

� Allocated PDCH: MIN_PDCH = 1

� Established TRX: MIN_PDCH = 1 and EN_FAST_INITIAL_GPRS_ACCESS = enabled


BSS Resources Reservation

MFSBSC

PDCH

BTS

Air

MFSBSC

M-EGCHPDCH M-EGCH

BTS

Air Abis AterMuxTRX

The 3 following states are handled, for a PDCH:

� allocated TS: timeslots allocated to the MFS.

� not-allocated TS: timeslots allocated to the BSC

� de-allocating TS: transitory state, TS allocated to the MFS but must be given back to the BSC






1 � 2 � 31

� If EN_FAST_INITIAL_GPRS_ACCESS = enabled

� 1 PDCH is always available for (E)GPRS traffic

� This PDCH is located

� on an established TRX

� with at least N_GCH_FAST_PS_ACCESS (=1) GCHs allocated on this TRX

� Mandatory rule:

� MIN_PDCH ≥ NB_TS_MPDCH + 1

� Since MIN_PDCH is the number of master and slave PDCHs that are permanently allocated

� Otherwise, the rule is MIN_PDCH ≥ NB_TS_MPDCH


Fast Initial PS Access

The “first TRX of the cell having some allocated RTSs” is identified thanks to the latest RR Allocation

Indication message received from the BSC. When considering the sorted TRX list in the latest RR Allocation

Indication message, it is the TRX supporting the first RTS allocated to the MFS, according to the

SPDCHs_Allocation bitmap.






1 � 2 � 32


Principle

� The BSC indicates regularly to the MFS in the RR Allocation Indicationmessage:

� The number of TSs allocated for PS traffic

� MAX_SPDCH_LIMIT

� The location of those SPDCHs on the PS capable TRXs

� Using the SPDCHs_Allocation bitmap

� The MFS sends to the BSC the RR Usage Indication message to:

� Confirm the allocated / de-allocated TSs to the BSC (Acknowledgement)

� Indicate the PS usage state of those TSs

� This message is sent:

� Periodically

� After the reception of the RR allocation Indication from the BSC

All the usable PS capable radio time slots are explicitly allocated to the MFS through the “RR Allocation

Indication” message.

A coordination is performed between the MFS and the BSC to allocate radio time slots for the PS traffic.






1 � 2 � 33


Principle [cont.]

� MFS / BSC Synchronization

BSC MFS

RR Allocation Indication

RR Usage Indication

RR Usage Indication

TCH_INFO_PERIOD (5s)

TCH_INFO_PERIODRR Allocation Indication

RR Usage Indication

RR_ALLOC_PERIOD (2)

x

TCH_INFO_PERIOD

B10

The RR Allocation Indication message:

� Is sent from the BSC to the MFS to provide the MFS with the location of the allocated SPDCH.

� Is transmitted periodically every RR_ALLOC_PERIOD * TCH_INFO_PERIOD seconds = 2 * 5 seconds.

� Contains the SPDCH_Allocation bitmap which indicates whether available time slots in the cell are allocated or not to

the MFS.

The RR Usage Indication message:

� Is sent from the MFS to the BSC periodically (every TCH_INFO_PERIOD seconds) or in response to a “Radio Resource

Allocation Indication” message.

� Contains 4 bitmaps:

� The SPDCHs_Confirmation bitmap. The role of this bitmap is to indicate the status of each SPDCH from the point

of view of the MFS and also to acknowledge the allocation of SPDCH newly granted by the BSC and the

deallocation of SPDCH given back to the BSC. The value of each bit in the SPDCHs_Confirmation bitmap has the

following meaning:

� 1: this SPDCH is allocated to the MFS (SPDCH allocation state is “allocated” or “de-allocating”).

� 0: this SPDCH is not allocated to the MFS (SPDCH allocation state is “not allocated”).

� The SPDCHs_Usage bitmap:

� 1: this SPDCH is allocated to the MFS and “used” (one TBF, RT PFC or one UL block has some radio

resources allocated on it, and/or if its basic Abis nibble is being used by a GCH channel or is still switched

to an Ater nibble in the BSC).

� 0: this SPDCH is either allocated to the MFS and “unused” (*), or is not allocated to the MFS.

� The SPDCHs_RadioUsage bitmap:

� 1: this SPDCH is allocated to the MFS and there is at least one TBF allocated on it.

� 0: this SPDCH is either allocated to the MFS and there is no TBF allocated on it, or is not allocated to the

MFS.

� The SPDCHs_DTM_TCH_List bitmap:

� 1: this SPDCH is used as TCH for DTM.

� 0: this SPDCH is used as No TCH for DTM.

The bitmaps are present for all the available PS capable TRXs of a cell, even if no SPDCH is allocated to the MFS for a

given TRX.






1 � 2 � 34

Computation of MAX_SPDCH_LIMIT


Functional Entities

AV_USED_CS_TS

AV_USED_PS_TS

AV_UNUSED_TS

MAX_SPDCH_LIMIT

CS/PS Margin

Load Evaluation

Thresholds computation

NB_TS

LOAD_EV_PERIOD_GPRS

HIGH_TRAFFIC_LOAD_GPRS

THR_MARGIN_PRIORITY_PS

MARGIN_PRIORITY_CS

MARGIN_PRIORITY_PS

NB_TS_DEFINED

NB_TS_SPDCH

MAX_PDCH_HIGH_LOAD

MAX_PDCH

MIN_PDCH

NB_TS_MPDCH

NB_USED_CS_TS

NB_USED_PS_TS

MAX_SPDCH_HIGH_LOAD

MIN_SPDCH

MAX_SPDCH

B10

The BSC peridically computes the number of Slave PDCHs that it can provide to the MFS. This number of

SPDCHs is denoted MAX_SPDCH_LIMIT. In order to evaluate the value of this parameter, the BSC requires to

compute a certain number of other parameters.

This different steps to evaluate MAX_SPDCH_LIMIT are the following ones:

Computation of MIN_SPDCH, MAX_SPDCH and MAX_SPDCH_HIGH_LOAD from O&M parameters and from

NB_TS which has a changeable value (in case of TRX failure for example, seen later)

� Computation of MARGIN_PRIORITY_CS and of MARGIN_PRIORITY_PS

� Computation of MAX_SPDCH_LIMIT






1 � 2 � 35


Thresholds Computation


AV_USED_CS_TS

AV_USED_PS_TS

AV_UNUSED_TS

MAX_SPDCH_LIMIT

CS/PS Margin

Load Evaluation


NB_TS

LOAD_EV_PERIOD_GPRS



MARGIN_PRIORITY_CS

MARGIN_PRIORITY_PS

NB_TS_DEFINED

NB_TS_SPDCH

MAX_PDCH_HIGH_LOAD

MAX_PDCH

MIN_PDCH

NB_TS_MPDCH

NB_USED_CS_TS

NB_USED_PS_TS

MAX_SPDCH_HIGH_LOAD

MIN_SPDCH

MAX_SPDCH






1 � 2 � 36


Thresholds Computation [cont.]

� MAX_SPDCH = Roundup [Min(MAX_PDCH – NB_TS_MPDCH ; NB_TS_SPDCH) x Availability_TS_Ratio]

� MAX_SPDCH_HIGH_LOAD = Roundup [Min(MAX_PDCH_HIGH_LOAD –NB_TS_MPDCH ; NB_TS_SPDCH) x Availability_TS_Ratio]

� MIN_SPDCH = Roundup [(MIN_PDCH – NB_TS_MPDCH) x Availability_TS_Ratio]

� Where Availability_TS_Ratio(k) = NB_TS(k) / NB_TS_DEFINED

� evaluated at instant tk, every RR_ALLOC_PERIOD x TCH_INFO_PERIOD (10s)

� NB_TS(k) takes into account the possible TRX failures

NB_TS_SPDCH is the total number of TCH/SPDCH time slots in the cell, without taking into account the

possible TRX failure. This parameter can be retrieved by the BSC from the O&M configuration of the cell.

NB_TS_DEFINED is the total number of pure TCH, TCH/SDCCH, or TCH/SPDCH time slots, without taking into

account the possible TRX failure. This parameter can be retrieved by the BSC from the O&M configuration

of the cell.

NB_TS is the total number of pure TCH, TCH/SDCCH, or TCH/SPDCH time slots (evaluated every 10s), taking

into account possible TRX failures.

A TCH/SPDCH is a TS which can be allocated for either CS traffic or PS traffic (i.e. mapped on a PS capable

TRX and inside the SPDCH group).

A pure TCH is a TS which can be allocated only for CS traffic (i.e. outside the SPDCH group).

A TCH/SDCCH is a dynamic SDCCH (can be allocated as either a TCH or an SDCCH).






1 � 2 � 37


Load Evaluation


AV_USED_CS_TS

AV_USED_PS_TS

AV_UNUSED_TS

MAX_SPDCH_LIMIT

CS/PS Margin

Load Evaluation


NB_TS

LOAD_EV_PERIOD_GPRS



MARGIN_PRIORITY_CS

MARGIN_PRIORITY_PS

NB_TS_DEFINED

NB_TS_SPDCH

MAX_PDCH_HIGH_LOAD

MAX_PDCH

MIN_PDCH

NB_TS_MPDCH

NB_USED_CS_TS

NB_USED_PS_TS

MAX_SPDCH_HIGH_LOAD

MIN_SPDCH

MAX_SPDCH

B10






1 � 2 � 38


Load Evaluation [cont.]

� The BSC computes 3 averaged values

� AV_USED_CS_TS(k), AV_USED_PS_TS(k) and AV_UNUSED_TS(k)

� For each cell

� At sampling instant tk, every RR_ALLOC_PERIOD x TCH_INFO_PERIOD (10s)

� Using a sliding window, LOAD_EV_PERIOD_GPRS

� Same formula for AV_USED_PS_TS(k) and AV_UNUSED_TS(k) based on NB_USED_PS_TS(k) and NB_UNUSED_TS(k)

� NB_UNUSED_TS(k) = NB_TS(k) – NB_USED_CS_TS(k) – MAX[MIN_SPDCH(k) ; NB_USED_PS_TS(k)]

∑1- RIOD_GPRSLOAD_EV_PE

0 =i

i)-(k _TSNB_USED_CS RIOD_GPRSLOAD_EV_PE

1 = _TS(k)AV_USED_CS

B10

NB_USED_CS_TS(k): number of available time slots handled by the BSC and carrying CS traffic in the cell at

sampling instant tk. A time slot is taken into account in the evaluation of NB_USED_CS_TS(k) if:

� SPDCH allocation state = not allocated,

� Occupancy state = used.

� SPDCH allocation state = allocated or de-allocating

� DTM mode = DTM TCH

NB_USED_PS_TS(k): number of available time slots used for PS traffic in the cell at sampling instant tk. A

time slot is taken into account in the evaluation of NB_USED_PS_TS(k) if:

� SPDCH allocation state = allocated or de-allocating,

� Occupancy state = used.

� DTM Mode = DTM TCH

These variables are updated every TCH_INFO_PERIOD (5s).






1 � 2 � 39


Load Evaluation [cont.]

TCH_INFO_PERIOD = 5s

AV_USED_CS_TS(k)

AV_USED_PS_TS(k)

AV_UNUSED_TS(k)

NB_USED_CS_TS(k)

NB_USED_PS_TS(k)

NB_USED_TS(k)

NB_UNUSED_TS(k)

kk-1k-2

LOAD_EV_PERIOD_GPRS = 3

k+1 k+2

AV_USED_CS_TS(k+2)

AV_USED_PS_TS(k+2)

AV_UNUSED_TS(k+2)

RR_ALLOC_PERIOD * TCH_INFO_PERIOD

B10






1 � 2 � 40


CS/PS Margins


AV_USED_CS_TS

AV_USED_PS_TS

AV_UNUSED_TS

MAX_SPDCH_LIMIT

CS/PS Margin

Load Evaluation


NB_TS

LOAD_EV_PERIOD_GPRS



MARGIN_PRIORITY_CS

MARGIN_PRIORITY_PS

NB_TS_DEFINED

NB_TS_SPDCH

MAX_PDCH_HIGH_LOAD

MAX_PDCH

MIN_PDCH

NB_TS_MPDCH

NB_USED_CS_TS

NB_USED_PS_TS

MAX_SPDCH_HIGH_LOAD

MIN_SPDCH

MAX_SPDCH






1 � 2 � 41


CS/PS Margins [cont.]

�MARGIN_CS = MAX(MARGIN_PRIORITY_CS(k), AV_UNUSED_TS(k) / 2)

� Where:

� MARGIN_PRIORITY_CS = THR_MARGIN_PRIO_CS x (NB_TS –MAX_SPDCH_HIGH_LOAD)

� THR_MARGIN_PRIO_CS = 100% - HIGH_TRAFFIC_LOAD_GPRS

�MARGIN_PRIORITY_PS = THR_MARGIN_PRIO_PS x MAX_SPDCH_HIGH_LOAD

� Where:

� THR_MARGIN_PRIO_PS = 10%

� These 2 margins are used to ensure that a certain number of TSs is kept available for the arrival of new calls / transfers between 2 RR Allocation Indication messages

B10

These new margins, one for CS traffic and one for PS traffic are here introduced to guarantee that a certain

number of timeslots are kept available for the arrival of new calls between two transmissions of RR

Allocation Indication messages.

MARGIN_PRIORITY_CS : dedicated to CS traffic.

MARGIN_PRIORITY_PS : dedicated to PS traffic.

These margins are re-evaluated every RR_ALLOC_PERIOD * TCH_INFO_PERIOD, before the computation of

MAX_SPDCH_LIMIT.






1 � 2 � 42


MAX_SPDCH_LIMIT Computation


AV_USED_CS_TS

AV_USED_PS_TS

AV_UNUSED_TS

MAX_SPDCH_LIMIT

CS/PS Margin

Load Evaluation


NB_TS

LOAD_EV_PERIOD_GPRS



MARGIN_PRIORITY_CS

MARGIN_PRIORITY_PS

NB_TS_DEFINED

NB_TS_SPDCH

MAX_PDCH_HIGH_LOAD

MAX_PDCH

MIN_PDCH

NB_TS_MPDCH

NB_USED_CS_TS

NB_USED_PS_TS

MAX_SPDCH_HIGH_LOAD

MIN_SPDCH

MAX_SPDCH






1 � 2 � 43


MAX_SPDCH_LIMIT Computation: Functional Entities

MAX_SPDCH_LIMITComputation of MAX_SPDCH_LIMIT

MIN_SPDCH

MARGIN_PRIORITY_PS

Computation of

MAX_SPDCH_LIMIT_CS

Computation of

MAX_SPDCH_LIMIT_PS

AV_USED_CS_TS

AV_UNUSED_TS

AV_USED_PS_TS

MARGIN_PRIORITY_CS

MAX_SPDCH

MAX_SPDCH_HIGH_LOAD

MAX_SPDCH_LIMIT_CS

MAX_SPDCH_LIMIT_PS

The basic idea for the evaluation of MAX_SPDCH_LIMIT is to start from the number of unused time slots and

to share them between CS and PS traffic, taking into account 2 margins (one for CS, one for PS traffic)

defined to guarantee a certain number of time slots available to serve incoming calls/transfers.






1 � 2 � 44


MAX_SPDCH_LIMIT Computation: Formulas

�MAX_SPDCH_LIMIT is equal to:

� MAX_SPDCH_LIMIT_CS under normal load

� MAX_SPDCH_LIMIT_PS under high CS load

�MAX_SPDCH_LIMIT_CS

� Represents the maximum number of SPDCHs that can be allocated to the MFS ensuring the CS allocation is not degraded

� = RoundDown[NB_TS – AV_USED_CS_TS – MARGIN_CS]

�MAX_SPDCH_LIMIT_PS

� Represents the minimum number of SPDCHs that can be allocated to the MFS ensuring the PS allocation is not degraded

� = RoundUp[AV_USED_PS_TS + MARGIN_PRIORITY_PS]

OR

B10






1 � 2 � 45


MAX_SPDCH_LIMIT Computation: Formulas [cont.]

� Choice between MAX_SPDCH_LIMIT_CS and MAX_SPDCH_LIMIT_PS

� If MAX_SPDCH_LIMIT_CS >= MAX_SPDCH_HIGH_LOAD then

� MAX_SPDCH_LIMIT = Min[MAX_SPDCH_LIMIT_CS ; MAX_SPDCH]

� Else

� If MAX_SPDCH_LIMIT_CS > MAX_SPDCH_LIMIT_PS then� MAX_SPDCH_LIMIT = Min[MAX_SPDCH_LIMIT_CS ; MAX_SPDCH]

� Else � MAX_SPDCH_LIMIT = Min[MAX_SPDCH_LIMIT_PS ; MAX_SPDCH_HIGH_LOAD]

�MAX_SPDCH_LIMIT is always between MIN_SPDCH and MAX_SPDCH

B10






1 � 2 � 46


MAX_SPDCH_LIMIT Computation: Example

NB_TS=14 MAX_SPDCH=14 MIN_SPDCH=1

MAX_SPDCH_HIGH_LOAD=2 HIGH_TRAFFIC_LOAD_GPRS=80%

Exercise

2

4

6

8

10

12

14

High CS load

Without PS traffic

CS

Traffic

Normal

load

CS

Traffic

PS

Traffic

High PS

load

PS

Traffic

High CS load

with PS traffic

CS

Traffic

PS

Traffic

Case “Normal Load”: The capacity is shared between CS and PS taking into account the associated margins.

Case “High CS load”: When the CS traffic increases, MAX_SPDCH_LIMIT is reduced down to

MAX_SPDCH_HIGH_LOAD to ensure a minimum capacity for the PS traffic.

Case “Very high CS load”: When the CS traffic increases and in the same time there is no PS traffic,

MAX_SPDCH_LIMIT is decreased down to MIN_SPDCH to ensure the maximum capacity for the CS.

Case “High PS load”: MAX_SPDCH is not the only criterion taken into account to limit the PS capacity, the

CS_MARGIN is applied too.






1 � 2 � 47


PS Zones: Definition

� MAX_SPDCH = 12, MIN_SPDCH = 1, MAX_SPDCH_LOAD = 8

� MAX_SPDCH_LIMIT = 7

TRX3 TRX1

BCCH SDCCHPS PS PS PS CS CS CS

Non pre-emptable zone

MAX_SPDCH_HIGH_LOAD zone

MAX_SPDCH_LIMIT zone

PS traffic zone

B10

MAX_SPDCH_HIGH_LOAD zone: this zone corresponds to the MAX_SPDCH_HIGH_LOAD consecutive PS capable time slots that are preferred for PS allocation. In this zone, allocated TBFs cannot be preempted. If

the value of MAX_SPDCH_HIGH_LOAD is not modified, this zone remains unchanged.

For a DTM request, the MFS will always allocate DTM TCH in this zone, and try to select the leftmost

timeslot as DTM TCH.

Non pre-emptable PS zone: this zone is always inside the MAX_SPDCH_HIGH_LOAD zone. In this latter zone, we search for the rightest time slot allocated to the MFS and used, or allocated to the MFS, not use, TCH

for DTM. Then, all time slots situated at its left define this non pre-emptable PS zone.

MAX_SPDCH_LIMIT zone: this zone corresponds to the MAX_SPDCH_LIMIT consecutive PS capable time slots that are preferred for PS allocation.

PS traffic zone: this zone corresponds to the larger zone between the non pre-emptable PS zone and the MAX_SPDCH_LIMIT zone.






1 � 2 � 48


PS Zones: Selection of MAX_SPDCH_LIMIT TS

� The MAX_SPDCH_LIMIT timeslot are selected within:

� the non pre-emptable PS zone� It contains all the TS that can be or are allocated to the MFS

� the MAX_SPDCH_LIMIT zone� If the number of selected TS is still lower than MAX_SPDCH_LIMIT, the process continues outside this zone until this number reaches MAX_SPDCH_LIMIT.

�Once MAX_SPDCH_LIMIT timeslot have been selected, the remaining TS are allocated to the BSC

� The TS carrying the TCH of a MS in DTM remains allocated to the MFS:

� Even if they are outside the MAX_SPDCH_LIMIT zone

� It only concerns the TCH of the MS, not the TBF resources

B10

Caution: animated slide.

The BSC shall not de-allocate a DTM TCH timeslot carrying CS traffic, even if the DTM TCH is outside of

MAX_SPDCH_HIGH_LOAD value and MAX_SPDCH_LIMIT zone (in case of TRX failure, the BSC will adjust

MAX_SPDCH_HIGH_LOAD value, so it is possible the DTM TCH timeslot is outside of non-emptable zone) until

the DTM CS connection is released. When the DTM CS connection is released, the BSC notifies it to the MFS,

the MFS changes the DTM TCH timeslot from “TCH for DTM” to “no TCH for DTM” and indicates this

information to the BSC through the message RR Usage Indication, only then the BSC can de-allocate it.






1 � 2 � 49


PS Zones: Impact on CS Call

� If EN_RETURN_CS_ZONE_HO = enabled

� AND a CS call is inside both

� The Non pre-emptable zone and

� The MAX_SPDCH_LIMIT_ZONE then

� An intra cell HO cause 30 is triggered

TRX1

CS CSBCCH SDCCHPS PS PS PS CS

HO cause 30

PS PS

TRX3


MAX_SPDCH_HIGH_LOAD zone


PS traffic zone

B10


The BSC shall not de-allocate a DTM TCH timeslot carrying CS traffic, even if the DTM TCH is outside of

MAX_SPDCH_HIGH_LOAD value and MAX_SPDCH_LIMIT zone.

Cause 30 is not applied in case of TS carrying a DTM TCH. SPDCH allocation state is equal to “TCH for DTM”






1 � 2 � 50


SPDCH Allocation: Conclusion

� Example of an SPDCHs_Allocation bitmap

� MAX_SPDCH_LIMIT = 7

TRX3 TRX1

BCCH SDCCHPS PS PS PS CS CS



1 1 1 0 1 1 11 0 0 0 0 0 0 0 0

B10

CS*

DTM MS

CS* : DTM TCH


The algorithm of selection of MAX_SPDCH_LIMIT TCH/SPDCH time slots to allocate to the MFS and of

building of the SPDCHs_Allocation bitmap is performed as follows:

� Initialization:

� Definition of the non pre-emptable PS zone and of the MAX_SPDCH_LIMIT zone,

� nb_selected = 0,

� Begin with the time slot having the lowest index and situated on the TRX having the lowest rank in the

TRX Ranking Table.

� Process on each TCH/SPDCH time slots available in the cell:

� First, begins with the non pre-emptable PS zone: analysis of all time slots within this zone. At the end

of this zone:

� If nb_selected >= MAX_SPDCH_LIMIT: stop the process and the remaining time slots are

allocated to the BSC,

� Otherwise: continue the process in the MAX_SPDCH_LIMIT zone. At the end of this zone:

� If nb_selected = MAX_SPDCH_LIMIT: stop the process and the remaining time slots are

allocated to the BSC,

� Otherwise: continue the process outside the MAX_SPDCH_LIMIT zone. When nb_selected =

MAX_SPDCH_LIMIT: stop the process and the remaining time slots are allocated to the BSC.






1 � 2 � 51


SPDCH De-allocation: Principle

� A preemption is triggered in the MFS

� At the reception of an RR Allocation Indication message from the BSC

� In case there is one or several PDCH to give back to the BSC

� It concerns only used PDCHs where:� Its mapped basic Abis nibble is used in an M-EGCH link

� At least 1 TBF is allocated on it

� CS preemption takes place in two main steps:� Soft preemption

� Fast preemption

� SPDCH de-allocation is based on GCH inactivity timers

� Timers T_GCH_INACTIVITY and T_GCH_INACTIVITY_LAST are used to de-allocate the unused GCHs

OR

B10






1 � 2 � 52


SPDCH De-allocation: Preemption Mechanism

All T1 requests

played successfully

No TBF candidate for T1 or

T_PDCH_Preemption expiry

Fast preemption

Soft preemptionT1 re-allocation

Unlock PDCHs

Arrival of RR

Allocation

indication message

Lock PDCHsOn TRXs with all its basic Abis

nibbles impacted by CS

preemption

Step 1: lock the PDCHs of the TRXs which are highly impacted by the CS preemption

� This step is only applicable to Evolium BTSs

� The basic Abis nibbles of the time slots preempted by the BSC may be currently used by some GCHs in the cell. Thus, the TRXs

(or M-EGCHs) using these GCHs will be impacted by the preemption, even if the preempted TSs are not on these TRXs. That’s

why the TRXs of the cell for which the 2 following conditions are fulfilled have all their PDCHs locked by the preemption

process, it means that they cannot accept anymore traffic:

� Nb_GCH_Impacted_By_CS_Preemption > 0

� Established_Nb_GCH – Nb_GCH_Impacted_By_CS_Preemption <= Min_Nb_GCH_GBR

� Where:

� Nb_GCH_Impacted_By_CS_Preemption is the number of established GCHs in the M-EGCH link using a basic Abis

nibble mapped onto a radio time slot that is being deallocated by the BSC.

� Established_Nb_GCH is is the number of GCHs that are activated in the M-EGCH link and that will be kept activated

during a certain time.

� Min_Nb_GCH_GBR is an estimation of the minimum number of GCHs necessary in a given M-EGCH link for the GBR

(Guaranteed Bit Rate) traffic supported by the TRX in both directions (UL and DL).

Step 2: soft preemption process

� 2 kinds of TBF are candidate for the T1 re-allocation process:

� The TBFs whose PACCH is supported by a preempted time slot.

� The TBFs for which it will no longer be possible to serve their on-going max allowed (M)CS because of the subsequent

reduction of the M-EGCH link size of their TRX.

� In addition, in the case where the number of TBFs established on a TRX will become too high according to the remaining

number of GCHs in the M-EGCH link of the TRX, then some TBFs will be released.

Step 3: fast preemption process

� At T_PDCH_Preemption timer expiry:

� The TBFs which could not be T1 reallocated are released.

� The locked PDCHs which do not carry PACCH are released (TBF throughput reduction).

� The GCHs using the basic Abis nibbles of the preempted radio TSs are released.

Step 4: unlock the PDCHs:

� If some PDCHs were “locked” in step 1, they are unlocked. Indeed, the CS preemption process is over, so the existing M-EGCH

links will no longer be disturbed by the release of some of their GCHs due to the fast preemption.






1 � 2 � 53


SPDCH De-allocation: Preemption Mechanism [cont.] B10

Fast preemption

Soft preemption

Unlock PDCHs

Arrival of RR

Allocation

indication message

Lock PDCHsOn TRXs with all its basic Abis

nibbles impacted by CS

preemption

No T1 reallocation will be

triggered during

T_PDCH_Preemption

TBF resources of the MSs are

released.

TCH ressources shall not be

deallocated

� Case of resources allocated in DTM mode

The preemption of a PDCH in the “TCH state” is controlled by the BSC Through arrival of RR Allocation

Indication message

That situation is rare as the TSs used for an MS in DTM mode are only allocated in the “non preemptable PS

zone”

A PDCH supporting a BE TBF for an MS in DTM mode can be preempted by the BSC. That situation can occur:

� if a RR Usage Indication message indicating the TSs used by the BE TBFs crosses a RR Allocation Indication

message sent by the BSC which preempts some of those TSs.

� or due to a change of the “non preemptable PS zone” of the cell

In such situation the following algorithm shall be applied:

� tep 1: Identical to non DTM case

� Step 2: The same conditions as non DTM ones apply

If the DTM TBF resources are impacted by the CS preemption , the a notification message will be

sent to RRM-PPCC. This notification message will be ignored by RRM-PCC and no T1 reallocation

will be triggered by RRM-PCC.

� Step 3: fast preemption process

The same conditions as non DTM ones apply

RRM-PCC will release the TBF resources (both UL and DL) of the MSs. The TCH resources shall not

be deallocated (TCH resource deallocation will happen on BSC-Shared-DTM-Info-Indication

reception in the nominal case)

� Step 4: unlock the PDCHs:

Identical to non DTM case






1 � 2 � 54


SPDCH De-allocation: Soft Preemption

� The MFS locks all the PDCHs impacted by the CS preemption:

� i.e. The de-allocated PDCH indicated in the RR Allocation Indication message sent by the BSC (the PDCHs to be preempted)

The MFS locks also all the PDCHs of the TRX which have a GCH mapped on a RTS impacted by preemption

and only in case of all the GCHs of the TRX are impacted by the preemption (rare case).






1 � 2 � 55


SPDCH De-allocation: Soft Preemption [cont.]

� 2 kinds of TBF are candidate for the T1 re-allocation process:

� The TBFs whose PACCH is supported by a preempted time slot.

� The TBFs for which it will no longer be possible to serve their on-going max allowed (M)CS because of the subsequent reduction of the M-EGCH link size of their TRX.

TRX1

TRX2

TRX3

TRX4

RSL

OML

31

TRX1

TRX3

BCCH SD

TBF1 – CS2

PDCHs to

be

preemptedBonus Bonus

.

.

.TBF2 is candidate for

T1 re-allocation

Abis

TBF2 – MCS7






1 � 2 � 56


SPDCH De-allocation: Fast Preemption

� Fast preemption:

� At expiry of T_PDCH_PREEMPTION, the MFS de-allocates the TS before TBF ending, having the following impacts:

� The TBFs whose PACCH is impacted (the corresponding PDCH is marked) are released.

� The MFS sends a “Packet TBF release” message with polling (i.e acknowledgement is requested)

� The TBFs whose PACCH is not impacted are not released but have a throughput reduction.

� The MFS sends a “Packet PDCH release” message indicating the preempted PDCHs

� T_PDCH_PREEMPTION = TCH_INFO_PERIOD –1 = 4s

� T_PDCH_PREEMPTION cannot be set at the OMC-R anymore

The PACCH blocks are the most important blocks to monitor. Many GPRS features ensure that PACCH blocks

are always monitored by the MS:

� The PTCCH is carried by the same PDCH as the PACCH.

� The RXLEV measurements for the power control and CS adaptation are made on the PDCH that carries the

PACCH blocks.

� Some RLS mechanisms are based on whether or not the MS is able to send or listen to PACCH blocks.






1 � 2 � 57

3 TBF Radio Resources Allocation and Re-Allocation






1 � 2 � 58

� ASAP: used for BE TBF establishment, T1, T2 and T4 reallocation

� Its goal is to serve the request as soon as possible.

� OPTIMAL: used for T3 reallocation

� Its goal is to ensure that a significant bandwidth will be offered to the MS upon T3 reallocation, even if it takes some time to establish all the necessary GCHs

� Nb_GCH_For_TBF_Estab:

� Minimum number of GCHs, which are required on the TRX to serve the request.


Definition of the 2 TBF Allocation Policies

1 to 5 (Max_Allowed_(M)CS of concurrent TBF)OptimalT3 TBF reallocation (MS in PTM or in DTM

1ASAPDTM allocation request

1 to 2 (Max_Allowed_CS of concurrent TBF)ASAPT4 TBF reallocation

1ASAPT1 TBF reallocation

1 to 5 (Max_Allowed_(M)CS of concurrent TBF)ASAPTBF establishment (with concurrent)

1ASAPTBF establishment (without concurrent)

Nb_GCH_For_EstabPolicyType of request

New B10






1 � 2 � 59

� Depends on the type of TBF (GPRS / EGPRS) and of the direction of the TBF (UL / DL)

� Is based on the number of established GCHs:

� Established_Nb_GCH – Nb_GCH_Impacted_CS_Preemption

� Is limited by:

� GPRS / EGPRS TRX capability,

� MAX_GPRS_CS and MAX_EGPRS_MCS


Determination of the Max_Allowed_(M)CS of a TBF

CS-4≥ 2

CS-2 (UL) / CS-1 (DL)1

Max_Allowed_CSNb_GCH

MCS-9≥ 5

MCS-74

MCS-63

MCS-52

MCS-2 (UL) / MCS-1 (DL)1

Max_Allowed_MCSNb_GCH

Established_Nb_GCH is is the number of GCHs that are activated in the M-EGCH link and that will be kept

activated during a certain time.






1 � 2 � 60


Functional Entities

Candidate TS allocation

Best-effort TBF allocation/reallocation request (received from RRM-PCC or dequeued from an Li list)

TRX list computing

Cf. session 2.2

Best candidate allocation computation

No candidate TS allocation

RADIO RESOURCE ALLOCATION/REALLOCATION ALGORITHM

TBF ESTABLISHMENT PROCESS

Cell Transmission Equity

“Enough GCHs” “Not enough GCHs”

“ALLOC OK” case

“ALLOC FAILED” case

Test if enough GCHs

Available_Nb_GCH_With_Equity

TRX list

Transmission Resource Availability

DSP congestion state

TRX list sorted by the BSC

Available_Nb_GCH

Transmission resource reservation

n_MS_requested, n_MS_requested_concurrent

Multislot class, Bias, Traffic type

Number of radio TSs

determination

Type of the TBF request

PDCH capacity/TFI/TAI/USF allocation

- rejected request - or L4 queuing - or L5/L6 queuing - or L7 queuing) - or try to change TBF mode (EGPRS case)

L4

� The list of DL TBFs which are still not served (due to a lack of resources).

� This list is sorted according to the PDU lifetime of the first DL LLC PDU of each queued request.

� On PDU lifetime expiry, the request is removed from the queue.

L5

� The list of MSs with UL bias which are candidate for resource re-allocation due to trigger T3.

� This list is sorted according to the following criteria:

� The candidate MSs are processed according to a FIFO order: the first request posted in the list is the

first processed by RRM PRH.

� In case it is not possible to reallocate resources to a candidate MS, the MS is put back at the end of

the list.

L6

� The list of MSs with DL bias which are candidate for resource re-allocation due to trigger T3.

� This list is sorted in the same way as L5.

L7

� The list of MSs which are candidate for resource re-allocation due to trigger T4.

� The candidate MSs are processed according to a FIFO order: the first request posted in the list is the first

processed by RRM PRH.

� In case it is not possible to reallocate resources to a candidate MS, the MS is put back at the end of the

list.






1 � 2 � 61




Best -effort TBF allocation/reallocation request (received from RRM -PCC or dequeued from an Li list)

TRX listcomputing

Cf. session 2.2







“ALLOC OK” case


Test if enough GCHs


TRX list



TRX list sorted

by the BSC

Available_Nb_GCH


n_MS_requested,n_MS_requested_concurrent

Multislot class,Bias,Traffic type

Numberof radio TSsdetermination



- rejected request- or L4 queuing

- or L5/L6 queuing

- or L7 queuing )- or try to change TBF mode (EGPRS case)

B10

The transmission resource Availibility step determines the total number of new GCHs which can be established:

� with free Abis and Ater resources (only the non-CS pre emptable Abis ressources are considered for RT PFC)

� with inter-cell GCH preemptions

This number is:

� Available_Nb_GCH in case of Best Effort TBF allocation

� Available_Nb_GCH_for_GBR in case of RT PFC allocation






1 � 2 � 62


Transmission Resource Availability: Example

� 3 cells with 2 TRXs:

� Max_PDCH_High_Load = 4 => a maximum of 4 non CS preemptable basic Abis nibbles can be established in the cell

� MAX_PDCH = 8, MAX_EGPRS_MCS = MCS-9, MAX_GPRS_CS = CS-4

� Max_SPDCH_Limit = 6 =>a maximum of 6 basic Abis nibbles can be established in the cell

CS

radioCell A

Cell B

Cell C

Abis

Basic

nibbles

6x4

extra

nibbles






1 � 2 � 63


Transmission Resource Availability: Example [cont.]

� First EDGE MS1, class 8 in cell A -> Target_Nb_GCH = 18

� GCH Allocation for MS1

� 4 basic Abis nibbles in the Max_PDCH_High_Load zone

� 14 extra Abis nibbles

� Total = 18 GCHs

Cell A

Cell B

Cell C

basic

extra

CSMS1

CSCell A






1 � 2 � 64



� Second EDGE MS2, class 8 in cell B -> Target_Nb_GCH = 18

� GCH Allocation for MS2� 4 basic Abis nibbles in the Max_PDCH_High_Load zone

� 10 free extra Abis nibbles

� 2 basic Abis nibbles outside the Max_PDCH_High_Load zone

� Inter GCH preemption between Cell A and Cell B:� 14 extra Abis in cell A, 10 extra Abis in cell B

� 2 extra Abis nibbles are preempted from Cell A to Cell B

� Total = 18 GCHs

basic

extra

Cell A

Cell B

Cell C

CSMS1MS2

CSCell ACell B






1 � 2 � 65



� After GCH allocation of MS2 in cell B

� In Cell A: Established_Nb_GCH = 16

�Next Periodical GCH process in Cell A: Equity process

� 2 basic Abis nibbles outside the Max_PDCH_High_Load zone

� Total = 16 + 2 = 18 GCHs

basic

extra

CSMS1MS2

CSCell ACell B

Cell A

Cell B

Cell C






1 � 2 � 66


Transmission Resource Availability: High Ater Usage

� Each time some radio resources are allocated on a new PDCH of a TRX

� No radio resources were formerly allocated on this PDCH for some TBFs

� If the Ater is in “high” usage

� Percentage of used Ater nibbles, in a GPU > Ater_Usage_Threshold

� Then, the number of GCHs targeted per PDCH is reduced

� Target_Nb_GCH_per_PDCH x GCH_RED_FACTOR_High_Ater_Usage

� In case of DTM:

� Ater_Usage_Threshold is not taken into account

B10

For other cases, each time some radio resources (for a TBF or for an RT PFC) are allocated on a new PDCH

of a TRX (“new PDCH” means that no radio resources were formerly allocated on this PDCH for some TBFs

or for some RT PFCs), then a value called alpha_HiAter shall be associated to this PDCH.

The value of alpha_HiAter to be associated to a “newly-used” PDCH is computed as follows:

� If the Ater usage of the GPU is “normal” (nominal case):

� alpha_HiAter = 1,

� If the Ater usage of the GPU is “high” (“high Ater usage” situation):

� alpha_HiAter = GCH_RED_FACTOR_High_Ater_Usage,

� with GCH_RED_FACTOR_High_Ater_Usage the O&M parameter value.

The value of alpha_HiAter is used in the Target_Nb_GCH computing.

If the Ater usage of the GPU is “high”, this will have the effect of applying a GCH reduction factor

(GCH_RED_FACTOR_High_Ater_Usage) to the number of GCHs targeted per PDCH, when “opening” new

PDCHs.

Some consequences of this mechanism are:

� If the Ater usage of the GPU becomes “high”, then the Ater consumption will increase more slowly from

this moment,

� If the Ater usage of the GPU stays “high” for a long time, and if the PDCH “closure” and “(re)opening”

rate is sufficient in the GPU, then the GCH reduction factor (GCH_RED_FACTOR_High_Ater_Usage) will tend

to be applied on many of the TRXs managed by the GPU, which will lead to a “pseudo-equity” in the usage

of the Ater resources of the GPU.






1 � 2 � 67


Transmission Resource Availability: High Ater Usage [cont.]

� Example:

� Considering an M–EGCH link of a TRX supporting one 4-TS best-effort EGPRS TBF

� MAX_EGPRS_MCS = MCS-9

� GCH_RED_FACTOR_High_Ater_Usage = 0.75

� Target_Nb_GCH = 1*4.49+1*4.49+1*4.49+1*4.49 = 18 GCHs if Ater usage of the GPU is “normal”

� Target_Nb_GCH = 0.75*4.49+0.75*4.49+0.75*4.49+0.75*4.49 = 14 GCHs if Ater usage of the GPU is “high”

For other cases, each time some radio resources (for a TBF or for an RT PFC) are allocated on a new PDCH

of a TRX (“new PDCH” means that no radio resources were formerly allocated on this PDCH for some TBFs

or for some RT PFCs), then a value called alpha_HiAter shall be associated to this PDCH.

The value of alpha_HiAter to be associated to a “newly-used” PDCH is computed as follows:

� If the Ater usage of the GPU is “normal” (nominal case):

� alpha_HiAter = 1,

� If the Ater usage of the GPU is “high” (“high Ater usage” situation):

� alpha_HiAter = GCH_RED_FACTOR_High_Ater_Usage,

� with GCH_RED_FACTOR_High_Ater_Usage the O&M parameter value.

The value of alpha_HiAter is used in the Target_Nb_GCH computing.

If the Ater usage of the GPU is “high”, this will have the effect of applying a GCH reduction factor

(GCH_RED_FACTOR_High_Ater_Usage) to the number of GCHs targeted per PDCH, when “opening” new

PDCHs.

Some consequences of this mechanism are:

� If the Ater usage of the GPU becomes “high”, then the Ater consumption will increase more slowly from

this moment,

� If the Ater usage of the GPU stays “high” for a long time, and if the PDCH “closure” and “(re)opening”

rate is sufficient in the GPU, then the GCH reduction factor (GCH_RED_FACTOR_High_Ater_Usage) will tend

to be applied on many of the TRXs managed by the GPU, which will lead to a “pseudo-equity” in the usage

of the Ater resources of the GPU.






1 � 2 � 68


Number of Radio TS Determination



TRX list computing

Cf. session 2.2







“ALLOC OK” case


Test if enough GCHs


TRX list




Available_Nb_GCH




Number of radio TSs

determination









1 � 2 � 69


Number of Radio TS Determination [cont.]

� The number of radio TSs is determined for:

� The direction of the request (n_MS_requested)

� The concurrent direction (n_MS_requested_concurrent)

� Taken into account

� The MS multislot class, (or DTM (E)GPRS multislot class)

� Maximum number of PDCH allocated to a single (E)GPRS connection:

� MAX_PDCH_PER_TBF

� The direction mainly used (in terms of throughput) by the on-going application: the bias

� The MFS counts and averages the number of bytes transferred in both directions

� By default bias = DL� Except when the first TBF establishment is a “UL 2 phase access”

� Extended Dynamic Allocation allowed or not (EN_EDA)

� The traffic type

� Only 1 TS is allocated in case of GMM/SM signaling traffic

B10

When allocating resources to an MS, both n_MS_requested and n_MS_requested_concurrent are considered, even when only one TBF is

established or being established, to take into account a potential future concurrent TBF, except in case of UL TBF establishment

without concurrent DL TBF (Immediate UL TBF establishment), where only n_MS_requested is taken into account to allocate resources

for the UL TBF.

General purpose of the bias determination:

�GPRS MSs are often involved in consecutive UL/DL transfers for a unique service.

� The Bias determination shall identify the direction of the main flow of data (based on the quantity of data exchanged at a specific

moment) in order to prioritize:

� The initial allocation on the biased direction.

� The re-allocation process on the main direction (likely to carry the useful data).

Each time n_received_octets_UL = N_BIAS_DETERMINATION or n_sent_octets_DL = N_BIAS_DETERMINATION (whichever is reached first),

RRM determines the bias as follows:

� if Av_n_received_octets_UL > Av_n_sent_octets_DL, the transfer is deemed uplink biased,

� else the transfer is deemed downlink biased.

� RRM then resets to 0 the counters n_received_octets_UL and n_sent_octets_DL.

Where:

� n_received_octets_UL represents the number of octets received in the UL.

� n_sent_octets_DL represents the number of octets sent in the DL.

� Av_n_received_octets_UL represents an average number of octets received in the UL.

� Av_n_sent_octets_DL represents an average number of octets sent in the DL.

� Av_n_received_octets_UL (new) = WEIGHT_BIAS_DETERMINATION * Av_n_received_octets_UL (old) + (1 -

WEIGHT_BIAS_DETERMINATION) * n_received_octets_UL.

� Av_n_sent_octets_DL (new) = WEIGHT_BIAS_DETERMINATION * Av_n_sent_octets_DL (old) + (1 - WEIGHT_BIAS_DETERMINATION) *

n_sent_octets_DL.

� N_BIAS_DETERMINATION and WEIGHT_BIAS_DETERMINATION are MFS (DLS) parameters used to tune the bias determination. The

weighting factor is used in particular to avoid changing too quickly the bias of a transfer so that this determination takes into account

the number of octets exchanged in the past.

Default values:

� N_BIAS_DETERMINATION = 3 KB

� WEIGHT_BIAS_DETERMINATION = 0,7






1 � 2 � 70


Best Candidate Allocation Computation



TRX list computing

Cf. session 2.2







“ALLOC OK” case


Test if enough GCHs


TRX list




Available_Nb_GCH




Number of radio TSs

determination




B10






1 � 2 � 71

�GPRS

� EGPRS

� The same states and parameters are used

� But, only EGPRS TBFs are taken into account

� Specific state for GPRS UL TBF allocation: EGPRS PDCH� PDCH used in the DL by an 8-PSK capable EGPRS TBF

� TCH State: If a TCH has been allocated by the MFS on this PDCH for a DTM-capable MS


SPDCH States

MAX_XX_TBF_SPDCH

PDCH

FULL

PDCH

ACTIVEPDCH

ALLOCATED

PDCH

TCH State

B10

Following states are defined for a PDCH:

� Allocated: the PDCH is an SPDCH which has been indicated as usable for PS traffic by the BSC.

� Active: an allocated PDCH is active if it supports at least one radio resource allocated for a TBF or for an

RT PFC.

� Full: an allocated PDCH is full in a given XL (XL = UL or DL) direction if and only if:

� for GPRS Best Effort TBF: Nb_RT_PFC_XL + Nb_BE_TBF_XL ≥ MAX_XL_TBF_SPDCH

� for EGPRS Best Effort TBF: Nb_RT_PFC_XL + Nb_BE_EGPRS_TBF_XL ≥ MAX_XL_TBF_SPDCH

� for RT resource allocation: Nb_RT_PFC_XL + Nb_BE_TBF_XL ≥ MAX_XL_TBF_SPDCH

� This is the same definition as in B8 release except that the concepts of RT PFC and best effort TBF

are introduced.

� EGPRS: An allocated PDCH is in the “EGPRS” state if some radio resources are allocated in the DL

direction, for an EGPRS TBF or an EGPRS RT PFC. This state is only used when running the radio resource

allocation/reallocation algorithm in GPRS mode and when considering the UL direction of the candidate

TBF allocations.






1 � 2 � 72


Available Throughput Computation

� For a GPRS TBF, in case of only BE TBFs with the same priority

� available_throughput_candidate_XL = R_AVERAGE_GPRS *

� NB_TBFPDCHi represents the number of already allocated GPRS and EGPRS TBFs on the PDCH i

� For an EGPRS TBF, in case of only BE TBFs with the same priority

� available_ throughput _candidate_XL = R_AVERAGE_EGPRS *

� NB_TBFPDCHi represents the number of already allocated EGPRS TBFs on the PDCH i

∑= +

n

1i PDCHi 1NB_TBF

1

∑= +

n

1i PDCHi 1NB_TBF

1

Appendix






1 � 2 � 73


Candidate TS Allocation Sorting

� The criteria of the TBF radio resource allocation/reallocation algorithm are “throughput-based”:

� ALPHA/ For “ASAP” policy only: the candidate time slot allocations, which are on some TRXs for which (Established_Nb_GCH - Nb_MPDCH) is greater than Nb_GCH_For_TBF_Estab” are preferred

� A/ For UL GPRS TBF establishment / reallocation only: the candidate time slot allocations, which have the lowest number of PDCHs in the “EGPRS” state are preferred

� B/ the candidate time slot allocations, which have the highest available throughput in the direction of the bias are preferred

� C/ the candidate time slot allocations, which have the highest available throughput in the direction opposite to the bias are preferred

� D/ the candidate time slot allocations, which are on the TRX with the highest priority, are preferred

� E/ for EGPRS TBFs establishments only: the candidate time slot allocations, which have the lowest number of GPRS TBFs in the direction of the bias, are preferred

� F/ combinations with the PDCHs that have the lowest index are preferred

Exercise1 Exercise2

When evaluating criterion [F], the concurrence constraints imposed by the MS multislot class (if it is known)

or by the default multislot class (if the MS multislot class is not known) shall be taken into account. This

will indeed avoid unnecessary subsequent T2 TBF reallocations (after having established an incoming TBF

without concurrent TBF).




Criterion [DTM A]: The TS of a candidate timeslot allocation supporting the TCH shall not be taken into account in the evaluation of this criterion.

This criterion is only relevant for:

� GPRS DTM allocation request

� T3 GPRS TBF reallocation request for an MS in DTM mode when considering the UL direction of a

candidate timesolt allocation.

Criterion [DTM TCH 1]: Nb_BE_TBF_XL indicates the total number of BE TBFs (GPRS or EGPRS) which have some radio resources allocated on the considered PDCH in a given direction.

Criterion [DTM TCH 2]: This criterion is only applicable for an MS whose DTM [E]GPRS multislot class is 11.

This criterion shall not be applied for an MS whose DTM [E]GPRS multislot class is 5 or 9.

This criterion has for goal to allow (if possible) a future T3 reallocation of the UL TBF of the DTM MS from

DA to EDA mode

Example :

Criterion [DTM TCH 3]: For an MS whose DTM [E]GPRS multislot class is 11, criterion [DTM TCH 2] will have precedence over this criterion, thus preventing (if possible) the usage of the “leftmost” TS of the TRX of a

TCH allocation.



1 � 2 � 74


Candidate TS Allocation Sorting [cont.]

� The criteria of the TBF radio resource allocation/reallocation algorithm in case of DTM mode are “throughput-based”:

� DTM A/ The candidate time slot allocations, which have the lowest number of PDCHs in the “EGPRS” state are preferred

� DTM B/ the candidate time slot allocations, which have the highest available throughput in the direction of the bias are preferred

� DTM C/ the candidate time slot allocations, which have the highest available throughput in the direction opposite to the bias are preferred

� DTM TCH 1/ the candidate time slot allocations for which (Nb_BE_TBF_DL + Nb_BE_TBF_UL)

� Is the lowest on the TS supporting the TCH are preferred

� DTM D/ the candidate time slot allocations, which are on the TRX with the highest priority, are preferred

� DTM TCH 2/ for an MS class 11, the candidate time slot allocations whose TCH is located on TS(n), with n>=1 and with TS(n-1) valid to support TBFs resources in both UL and DL are preferred

� DTM TCH 3/ the candidate time slot allocations whose TCH has the lowest TS index (on a given TRX) is prefered

B10

If only TS0 and TS1 are allocated, a DTM

request for a MS class 11 will be reserved on

TS0 (DL and UL), TCH on TS1. If a third TS is

allocated, a T3 reallocation of the UL TBF to

EDA will be possible






1 � 2 � 75


Re-allocation

� 4 different TBF reallocations are permanently activated:

� T1: reallocation to maintain a TBF alive despite the CS preemption of someRTSs or of some GCHs in the cell

� T2: reallocation of an on-going TBF when establishing a concurrent TBF

� T3: reallocation useful to� Establish a new M-EGCH link for one of the TRXs of the cell

� Perform a “radio de-fragmentation” process

� Provide a higher throughput, if it is possible, to a TBF

� Reallocate the UL TBFs established in the cell� From DA to EDA mode, or from EDA to DA mode

� T4: reallocation to move a UL GPRS TBF sharing one PDCH with a DL EGPRS TBF onto PDCHs which do not support a DL EGPRS TBF. It concerns only GPRS TBFs

� In case of DTM mode, reallocations are forbidden (MR1)

B10

To be candidate to subsequent resource re-allocation (T3 and T4), the following conditions have to be met:

� the TBF established in the biased direction is marked with “subsequent allocation”.

� more than N_CANDIDATE_FOR_REALLOC bytes have been transferred for the TBF in the biased direction.

� T3192 is not running (specific to UL TBF re-allocation when T3192 is running for the DL TBF).

Note: the case where an UL TBF is established using EDA, the MS bias become equal DL and the DL TBF

(concurrent of the UL TBF) is released is neglected. In this case, as there is no TBF established in the

direction of the bias (DL), the MS will not be candidate for T3 reallocation, and the reallocation from EDA

to DA will not operate for the UL TBF, even if the usage of EDA is no longer valid for this UL TBF.

The subsequent re-allocation is done whether or not the TBF corresponds to the bias transfer direction of

the MS.

N_CANDIDATE_FOR_REALLOC default value = 200 Bytes, it cannot be set at OMC-R level.






1 � 2 � 76


Re-allocation [cont.]

� T3 TBF reallocation:

� A T3 TBF reallocation is based on the following principles:

� Computing of a THROUGHPUT_RATIO (= Allocated_Throughput / Optimal_Throughput) to know “how sub-optimal a TBF allocation is”

� A T3 TBF reallocation will only be allowed if a significant THROUGHPUT_RATIO gain is reached. The minimal gain is set by the system parameter: MIN_THROUGHPUT_GAIN (= 40%)

THROUGHPUT_RATIO:

� for each MS which is candidate for a T3 TBF reallocation, a “throughput ratio” is calculated.

� this “throughput ratio” is useful to:

� validate the candidate TBF allocations when playing the radio resource reallocation algorithm.

� sort the T3 TBF reallocation requests within the L5 and L6 lists: low value of the “throughput ratio”

means high priority of the request.

� THROUGHPUT_RATIO = ALLOCATED_THROUGHPUT / OPTIMAL_THROUGHPUT

� ALLOCATED_THROUGHPUT is the throughput currently allocated to the TBF in the direction of the bias

and is equal to potential_throughput_PDCH * available_capacity_candidate_XL.

� OPTIMAL_THROUGHPUT is the optimal throughput that could be potentially allocated to the TBF in the

direction of the bias by considering its multislot class and is equal to potential_throughput_PDCH *

n_MS_requested.

Best candidate allocation computation:

a candidate TBF allocation shall fulfill the following conditions:

� NEW_THROUGHPUT_RATIO min(1, (1+MIN_THROUGHPUT_GAIN ) * CURRENT_THROUGHPUT_RATIO).

� NEW_THROUGHPUT_RATIO is the “throughput ratio” of the candidate TBF allocation.

� CURRENT_THROUGHPUT_RATIO is the “throughput ratio” of the current TBF allocation.

� MIN_THROUGHPUT_GAIN is an O&M parameter.






1 � 2 � 77


Re-allocation: Example

� Initial situation:

� 3 MSs (MSa, MSb, and MSc), all GPRS and (4+1)

� MSC is the most impacted by the multiplexing in terms of throughput

� Final Situation:

� MSC is candidate for T3 reallocation

� A new TRX will be established (cf. “Optimal” policy) and MSc will then be reallocated on this new TRX

0 1 2 3 4 5 6 7

DL

UL

MSaMSaMSaMSa MSbMSbMSbMSb

MScMScMScMSc

MSc MSbMSa

0 1 2 3 4 5 6 7

DL

UL

0 1 2 3 4 5 6 7

DL

UL

MSaMSaMSaMSa MSbMSbMSbMSb

MSbMSa

0 1 2 3 4 5 6 7

DL

UL

MScMScMScMSc

MSc






1 � 2 � 78

4 TBF Release Routine






1 � 2 � 79


Justification

�Wap access to home page without any feature

�With Delayed DL TBF Release (B7)

�With Delayed DL TBF TBF Release & (I) Extended UL TBF Mode (B10)

� Typical gain on Wap access to home page: � ~8 seconds with Delayed DL TBF Release

� + ~2 seconds with Extended UL TBF Mode

� New B10 feature to enhance MS battery saving

UL

DL

B10






1 � 2 � 80


Delayed DL TBF Release

(1): There is no more DL LLC PDU stored for the MS. The BSS sends the last segment of the last useful RLC

block. This block contains the last segment of the last useful DL LLC PDU, completed by a dummy LLC

PDU in order to maintain the DL TBF alive. The Final Block Indicator is not set (FBI=0) and a polling is

requested to the MS. T_Delayed_DL_TBF_Pol_Initial is activated (default value = 100 ms).

(2): When the MS has acknowledged the last useful DL RLC block, T_Delayed_DL_TBF_Rel is activated

(duration of the delayed DL TBF release phase).

(3): At T_Delayed_DL_TBF_Pol_Initial expiry, a new RLC block containing one or more dummy LLC PDUs is

sent to the MS. This RLC block contains a polling indication so that the MS can request a UL TBF

establishment, if required. T_Delayed_DL_TBF_Pol (default value = 200 ms) is activated.

(4): At T_Delayed_DL_TBF_Rel expiry, a last DL RLC block is sent with the Final Block Indicator set (FBI = 1),

indicating the end of the DL TBF. The normal release procedure then applies.






1 � 2 � 81


Delayed DL TBF Release [cont.]

� Artificial extended DL TBF duration aiming at coping with jerky DL traffic from the CN

� Procedure:

� the last DL RLC blocks are marked with FBI=0

� the TBF state goes from Active to Delayed

� periodical Dummy DL RLC blocks in polling (S/P=1) sent by the MFS to trigger acknowledgement from the MS (FAI=0)

� when a new DL LLC PDU arrives at the MFS, the useful RLC Block transfer is resumed

� the TBF state goes from Delayed to Active

� The MS does not take into account Dummy LLC PDU during the delayed release phase

Jerky LLC PDU delivery at MFS due to buffer capacities of servers, SGSN and MFS. A TCP segment can

generate up to 3 LLC PDUs. Also called “Bursty traffic”. HTTP and WAP services are likely to benefit from

this feature.

FBI: Final Block Indicator (RLC header)

FAI: Final Acknowledgement Indicator

S/P: triggers polling (packet Ack/Nack message) when set to 1

Periodical sending of DL RLC Blocks = polling period calculation: the MFS takes into consideration T3190

(guarding timer between 2 valid data received from the Network) in addition to the requirement of

receiving at least one block every 360 ms (78 TDMA frames). (T3190n = 5 s (Alcatel recommended value), it

cannot be set at OMC-R level).

T3190n = Timer used in the procedure “DL TBF abnormal release”: when the DL TBF is cut due to the radio

link quality or loss of the MS, the TFI and TAI cannot be reallocated during T3190n. The default value is 5s

and it cannot be set at the OMC-R level.

The UL Delayed TBF release (scheduling of additional USF) is only possible for Rel-4 MS.






1 � 2 � 82



� End of delayed released period when T_Delayed_DL_TBF_Rel expires

� T_Delayed_DL_TBF_Rel = T_NETWORK_RESPONSE_TIME

� The MFS sends a Dummy UI command marked with FBI=1, S/P=1

� Acknowledged mode:

� The MS sends the last Packet DL Ack/Nack message (FAI=1)

� Non-Acknowledged mode:

� The MS sends the last Packet Control Ack message

� T3192n and T3192 are triggered (Fast DL re-establishment)

T_NETWORK_RESPONSE_TIME corresponds to the time difference between a command sent to the SGSN and

the response received at the MFS. The default value is 700ms but it can be set at OMC-R level and can be

tuned according to Gb traces.






1 � 2 � 83



� RRM periods on MFS side:

� T_DELAYED_DL_TBF_POL_INITIAL (=60ms): the time that the MFS shall wait before sending the first RLC data block containing only LLC Dummy UI

� 2 timers are used to define the period between 2 DL Dummy UIs sent to the MS:

� T_MIN_POLL (=60ms), in case of the MS is “alone” on the PDCH which carries the PACCH

� T_DELAYED_DL_TBF_POL (=200ms) in case of the MS is “multiplexed” on the PDCH which carries the PACCH

� T_DELAYED_DL_TBF_POL_UL (=2s): the period between 2 DL Dummy UIs sent to the MS, when there is an on-going UL TBF

Upon each expiry of T_DELAYED_DL_TBF_POL (the timer reaches T_DELAYED_DL_TBF_POL_INITIAL or

T_DELAYED_DL_TBF_POL), a new Dummy UI command is inserted and T_DELAYED_DL_TBF_POL is restarted.

T_DELAYED_DL_TBF_POL_INITIAL = 60 ms

T_DELAYED_DL_TBF_POL = 200 ms

T_MIN_POLL = 60 ms

T_DELAYED_DL_TBF_POL_UL = 2000 ms

All these timer values are default ones and they cannot be set at OMC-R level.

T_DELAYED_DL_TBF_POL is used to:

� give the opportunity to the MS to request a UL TBF through the Packet DL Ack/Nack acknowledging the

polling, without too much disturbing the other TBFs (data transfer for the DL and USF scheduling for the UL)

multiplexed on the same PDCH (--> not too short period for Dummy UI commands).

� maintain the DL TBF at MS level (re-activation of T3190 timer (5s) in the MS) (--> period for Dummy UI

commands < 5s).

T_MIN_POLL is applied when the PDCH corresponding to the PACCH of the considered DL TBF is not shared

with another enabled DL TBF and when there is no activated UL TBF on this PDCH (ie, no UL TBF in enabled

or extended mode, for which USF may be scheduled).






1 � 2 � 84


DL TBF Extension

(1): The DL TBF is in delayed DL TBF release phase. Periodically Dummy UI Command messages are sent to

the MS with polling indication to give to the MS the opportunity to send a UL TBF request.

(2): The MS uses a Packet DL Ack/Nack to request a UL TBF. The UL TBF is established and

T_Delayed_DL_TBF_Rel is stopped (the DL TBF remains in delayed DL TBF release state, during the UL

TBF).

(3): At UL TBF release, the timer T_Delayed_DL_TBF_Rel is re-activated.

(4): When a DL LLC PDU is received, the first DL RLC data block can be immediately sent and

T_Delayed_DL_TBF_Rel is stopped.

(5): T_Delayed_DL_TBF_Rel is stopped, when the first UL RLC data block is received.






1 � 2 � 85


DL TBF Extension [cont.]

� When a UL TBF is established during a delayed DL TBF release:

� T_Delayed_DL_TBF_Rel is stopped and reset

� The delayed DL TBF release state is maintained during the UL TBF

� DL Dummy UIs are sent every T_DELAYED_DL_TBF_POL_UL

� The aim of this timer is just to maintain the DL TBF at MS level (must be lower than T3190)

� Therefore T_DELAYED_DL_TBF_POL_UL > T_DELAYED_DL_POL

� At the end of the UL TBF, T_Delayed_DL_TBF_Rel is restarted

Exercise

The aim of this function is to avoid the case where a UL TBF is established on the PACCH of a DL TBF which

is at the end of its delayed DL TBF release phase.

Indeed in this case, the subsequent DL LLC PDU (i.e., corresponding to a server response) may be received

after the release of the DL TBF.






1 � 2 � 86


Fast DL TBF Re-establishment

� After DL TBF release, the following timers are considered

� MS side

� T3192 started after sending of final Packet DL Ack/Nack message (FAI=1)

� during T3192 the MS listens to the PDCH carrying the PACCH blocks of its last DL TBF

T3192 = 500 ms (default value which can be set at OMC-R level).






1 � 2 � 87


Fast DL TBF Re-establishment [cont.]

� MFS side

� T3192n started after reception of final Packet DL Ack/Nack message

� Wait for reuse of MS radio resources (PDCHs, TAI, TFI)

� If a DL LLC PDU is received by the MFS, a fast DL TBF re-establishment is triggered on the PACCH

� T3192n = T3192 – (T_Round_Trip_Delay + T_Fast_DL_Margin)

� During T3192n, a UL TBF establishment is not possible

During the on-going of T3192n, no UL TBF establishment procedure can be proceeded. This is a limitation to

fast switching from DL TBF to UL TBF during the MS-GSS signaling procedure (location update for example).

In order to avoid a too long duration of these procedures, the MFS anticipates the UL TBF establishment by

starting the procedure before the end of the DL TBF release.

T3192n takes into account the trip time needed for “Packet DL Ack/Nack” message from the MS to the MFS

(½T_Round_Trip_Delay + ½T_Fast_DL_Margin) AND trip time needed for “Packet DL Immediate Assignment

” message from the MFS to the MS (½T_Round_Trip_Delay + ½T_Fast_DL_Margin).

Round_Trip_Delay (MFS-MS) = 160 ms (default value which can be set at OMC-R level).

T_Fast_DL_Margin = 50 ms (default value which cannot be set at OMC-R level).

Note: during T3192n seconds, the Timing Advance is monitored. Even so, if TAI occurs, the MS must send its

Access Burst for the Timing Advance calculation by the BTS. The MS shall listen to the TA Messages in the

DL.






1 � 2 � 88


Non-DRX Mode after Packet Data Transfer

� Discontinuous Reception (DRX mode):

� Used in GSM to increase the battery autonomy on an MS as in GSM CS: the MS listens only to its Paging Group

� Downlink TBF establishment through PCH long as compared with the TBF duration

� The MFS establishes a DL TBF on the first available PCH message of an MS Paging group

� Non-DRX:

� Continuous monitoring of AGCH messages by the MS

� The MFS establishes a DL TBF on the first available AGCH block (without MPDCH) or the first PPCH occurrence (with MPDCH)

This feature outlines one of the major differences between the GPRS service (non connected mode) and the

GSM service (connected mode). The DRX mode is highly recommended in GSM to save the cell battery when

it may be a handicap in GPRS (where the paging is likely to occur more frequently).






1 � 2 � 89


Non-DRX Mode after Packet Data Transfer [cont.]

� Non-DRX mode:

� Non-DRX period

� Min (NON_DRX_Timer; DRX_TIMER_MAX)

� Non-DRX period computed by the MFS and sent in “Packet DL Immediate Assignment” message

� The DRX mode of an MS is evaluated each time the MFS receives DL LLC PDU from the SGSN

� The MFS keeps the MS context until expiry of:

� DRX_TIMER_MAX if NON_DRX_Timer unknown for the MS

� Non-DRX period otherwise (provided within the DL LLC PDU)

� DRX_TIMER_MAX limited to 4 seconds, broadcast on SI13

NON_DRX_Timer is unknown for the MFS when after the release of an uplink TBF, no DL concurrent TBF was

Established, or after the release of a downlink TBF when the DL LLC PDUs do not convey the DRX

parameters.

When the MFS assesses that the MS returns to the DRX mode during the transmission of the assignment

message, the message is sent to the PCH or PPCH channel. The MFS shall then take into account the 95%

AGCH or PPCH queuing time (about 400 ms) in addition to the round trip time delay measured at RRM level

(about 160 ms).

Assuming a Non-DRX period of 2 seconds, this means that the downlink LLC PDU shall be received within 1.4

second to speed up the establishment of the DL TBF.

DRX_TIMER_MAX = 2 s (Alcatel recommended value) and it can be set at OMC-R level.






1 � 2 � 90


DL TBF Release: Summary

DL TBF

T_NETWORK_RESPONSE_TIME

TBF delayed

release

Fast DL TBF establishment via PACCH

TBF active

DL TBF

T3192

Fast DL TBF establishmentvia AGCH or any PPCH

DL TBF

T3192

Non-DRX mode

DRX_TIMER_MAX

DL TBF

T3192

Non-DRX mode

DRX_TIMER_MAX

DL TBF establishment

via PCH or PPCH

of MS paging group

The tables below indicate examples of the expected DL TBF establishment duration with or without the

feature.






1 � 2 � 91


Delayed UL TBF Release: Without Extended Mode

� The UL TBF release is delayed when:

� T_DELAYED_FINAL_PUAN <> 0

� No concurrent DL TBF:

� Is established

� Is being established

� Is being Released

� During the delayed UL TBF release:

� The MFS can establish a DL TBF on the PACCH/DL of the UL TBF

� The DL TBF establishment is speeded up

The UL Delayed TBF release (scheduling of additional USF) is only possible for a Rel-4 MS.






1 � 2 � 92


Delayed UL TBF Release: Without Extended Mode [cont.]

� No DL LLC PDU is received during the delayed final PUAN phase

T3180n

T3180

P. UL Ack/Nack (FAI=0, SSN = n) (3)

RLC/MAC block (BSN=n, CV=0) (1)

(4)

Packet Control Ack (6)

final P. UL ACK/NACK + polling (5)

MS MFS SGSN

LLC PDU (last LLC) (2)

T_DELAYED_FINAL_PUAN

(7)

(1): The last UL LLC PDU is received by the RLC. The RLC maintains the UL TBF alive since a normal end is

not allowed.

(2): The UL LLC PDU is sent to the RRM. A flag is set by the RLC, in order to notify the RRM that it is the last

UL LLC PDU of the TBF.

(3): A Packet UL Ack/Nack is sent to the MS, without polling, with FAI=0.

This message does not acknowledge the last block (n).

(4): Upon receipt of the last UL LLC PDU, the RRM starts an instance of the timer T_DELAYED_FINAL_PUAN.

The UL LLC PDU is forwarded to the SGSN.

(5): At T_DELAYED_FINAL_PUAN expiry, the RRM requests the RLC to release the UL TBF, by sending a

PCC_RLC_Activate-req allowing the RLC to perform a normal TBF end.

The RLC sends the final packet UL Ack/Nack to the MS and waits for Packet Control Acknowledgement.

T3180n is activated (T3180n = T3180 - Round Trip Delay - T_DELAYED_FINAL_PUAN).

(6): The MS acknowledges the receipt of the final PACKET UL Ack/Nack message.

(7): The RLC notifies the RRM about the release of the UL TBF. The resources of the UL TBF are released.

The RRM informs the BSSGP layer about the release of the UL TBF, and the BSSGP de-allocates the

corresponding throughput. T3180n is stopped.






1 � 2 � 93


Delayed UL TBF Release: Without Extended Mode [cont.]

� Delayed final PUAN, with DL TBF establishment

T3180nT3180

P. UL Ack/Nack (FAI=0, SSN = n) (3)

RLC/MAC block (BSN=n, CV=0) (1)

(5)

(4)


final P. UL ACK/NACK + polling

LLC PDU (last LLC) (2)


PACKET DOWNLINK ASSIGNMENT (6)


MS MFS SGSN

DL LLC PDU

(8)

(10)

(1): The last UL LLC PDU is received by the RLC. The RLC maintains the UL TBF alive since normal end is not

allowed.

(2): The UL LLC PDU is sent to the RRM. A flag is set by the RLC, in order to notify the RRM that it is the last

UL LLC PDU of the TBF.

(3): A Packet UL Ack/Nack is sent to the MS, without polling, with FAI=0.

This message does not acknowledge the last block (n).

(4): Upon receipt of the last UL LLC PDU, the RRM starts an instance of the timer T_DELAYED_FINAL_PUAN.

The UL LLC PDU is forwarded to the SGSN.

(5): A DL LLC PDU is received by the MFS while the timer T_DELAYED_FINAL_PUAN is running.

(6): Radio resources are requested to RRM-PRH. Upon RRM-PRH response, the RRM-PCC stops the timer

T_DELAYED_FINAL_PUAN, delays the release of the UL TBF until the completion of the DL TBF

establishment procedure, establishes the DL TBF on the PACCH/DL of the UL TBF.

� If T_DELAYED_FINAL_PUAN expires before the response of the RRM-PRH, then the RRM-PCC requests the

RLC to release the UL TBF, by sending a PCC_RLC_Activate-req allowing the RLC to perform a normal

TBF end. In this case, at RRM-PCC response, the DL TBF will be established on CCCH, at the end of the

UL TBF release.

(7): The MS acknowledges the receipt of the assignment message and listens to the DL resources. The BSS

sends a Packet Power Control and Timing Advance and then begins the DL data transfer.

(8): Upon receipt of the PACKET CONTROL ACKNOWLEDGEMENT, the RRM immediately requests the RLC to

release the UL TBF, by sending a pcc-rlc-activate-req primitive allowing the RLC to perform a normal TBF

end. The RLC sends the final PACKET UL Ack/Nack to the mobile station and waits for a PACKET CONTROL

ACKNOWLEDGEMENT.

(9): The MS acknowledges the receipt of the final PACKET UL Ack/Nack message.

(10): The RLC notifies the RRM about the release of the UL TBF. The RRM stops T3180n. The resources of

the UL TBF are released. The RRM informs the BSSGP layer about the release of the UL TBF, and the

BSSGP de-allocates the corresponding throughput.






1 � 2 � 94


Delayed UL TBF Release: with Extended Mode

� Principle

� At the end of the active phase (CV=0)

� The MFS continues to schedule some USFs

� To allow the MS to send some dummy RLC blocks

P. UL Ack/Nack (FAI=0, SSN = n+1)

RLC/MAC block (BSN=n, CV=0)

Packet Control Ack

final P. UL ACK/NACK + polling

LLC PDU (last LLC)

T_MAX_EXTENDED_ULUSF

Dummy RLC data block

MS MFS SGSN

USF


...

start

expiry

The aim of this feature is to extend the duration of the UL TBF in order:

� To quickly restart data transmission in UL if higher layers in the MS deliver new data, without having to

re-establish a new UL TBF, after the countdown procedure has started.

� To maintain the UL TBF established, some time after the last block (CV=0) has been acknowledged by the

network.

This feature allows to improve the access time to the GPRS network. It also improves the throughput in

some cases.

The feature is described in 3GPP TS 44.060 - V4.18.0. It applies for R4 MS.






1 � 2 � 95


Delayed UL TBF Release: with Extended Mode [cont.]

� The UL transfer can resume at any time during the extended phase



LLC PDU (last LLC)

T_MAX_EXTENDED_UL

USF


MS MFS SGSN

USF


USF

RLC/MAC block (BSN=n+1)

start

stop

USF

RLC/MAC block (BSN=n+2)

...

Active UL data transfer

Extended UL data transfer

Active UL data transfer

LLC PDU






1 � 2 � 96



� Conditions to operate

� EN_EXTENDED_UL_TBF = enabled

� Rel-4 MS

� The MS has to know if the BSS supports the feature

� The BSS capability (NW_EXT_UTBF) is broadcast on BCCH (SI13)

� The BSS has to know if the MS supports the feature

� The MS radio access capability is� received at downlink TBF establishment in the first downlink PDU, or

� retrieved through the Radio Access Capability Update which involves the SGSN

� If EN_RA_CAP _UPDATE = enabled

� If one of these 4 conditions is not fulfilled, the normal UL TBF release is performed, as in B8 (T_DELAYED_FINAL_PUAN is applied)

TLLI context retrieval procedure

� This procedure has been developed in B8 for the feature full intra PDU re-routing.

� The TLLI is known in the SGSN and by one GPU in the MFS. The TLLI retrieval procedure allows to retrieve

information linked to the TLLI, among which the Radio Access Capability from another GPU.

� This procedure can be triggered at uplink TBF establishment, as soon as the contention resolution is

completed.

Radio Access Capability Update

� Enabled or disabled the Radio Access Capability update on Gb by flag, EN_RA_CAP _UPDATE. It is

recommended to enabled this flag if EN_EXTENDED_UL_TBF is enabled and Radio Access Capability

update is supported by the SGSN.

� At UL TBF establishment, immediately after the “contention resolution” procedure, the “radio access

capability update” procedure is triggered in the BSS. The BSS requests an MS’s current Radio Access

capability and/or its IMSI by sending to an SGSN an RA_CAPABILITY_UPDATE, which includes the TLLI of

the MS and a Tag. Then it starts timer T5_RA_CAP_UPDATE (value = 5s). In case of timer expiry, the BSS

shall repeat the request up to RA_CAPABILITY_UPDATE_RETRIES times (value = 3).

� The SGSN shall respond by sending an RA_CAPABILITY_UPDATE_ACK, which includes the TLLI of the MS,

the Tag received in the corresponding RA_CAPABILITY_UPDATE.

� When the SGSN answers, the MS Radio Access capability is updated and the Extended UL feature can be

used if the “GERAN Feature Package 1” bit is set. Otherwise, the MS does not support the extended uplink

feature.






1 � 2 � 97



� The way to schedule the USFs in extended UL TBF mode:

� Scheduled only on the PDCH which carries the PACCH

� IF the PDCH supports uplink TBFs which are all in extended mode AND EN_FAST_USF_UL_EXTENDED = enabled THEN� the throughput in radio blocks is equally shared between MSs

� So USFs are scheduled as follows:� One MS in extended mode on PACCH: USF scheduled every 20ms

� Two MSs in extended mode on PACCH: USF scheduled every 40ms

� n MSs in extended mode on PACCH: USF scheduled every n x 20ms

� ELSE (if EN_FAST_USF_UL_EXTENDED = disabled OR if the PDCH supports at least one MS which is in UL transfer)� A period T_EXTENDED_UL_TBF_POL (200 ms) is used to schedule the USFs for all theMSs in extended mode

� The remaining bandwidth is used for MSs in transfer

� RRBP mechanism has priority over USF scheduling

� This means the USF scheduling may be shifted if RRBP request from the RLC






1 � 2 � 98


Delayed UL TBF Release: Improved Extended Mode

� Principle: make optional the MS answers to USF

� BSS sends first USF at each block or every T_EXTENDED_UL_TBF_POL

� In case of no MS answer, N3101 is incremented

� When N3101 reaches N3101_POLLING_THR, BSS sends polling (PUAN with polling request) every T_UL_RLS_EUTM

� In case of no answer to the polling, N_POLLING_EUTM is incremented

�When N_POLLING_EUTM_LIMIT is reached, Tbf is abnormally released

� Regular USF sent allows MS to resume UL traffic at any time (but N3101 is no more incremented when no MS answer)

B10

Principle is to make optional MS answers to USF in Extended UL phase

When an MS enters Tbf UL releasing phase, BSS knows only if MS supports Extended UL (but can not know if

the MS supports Improved or Normal Extended UL mode)






1 � 2 � 99


Delayed UL TBF Release: Improved Extended Mode [cont.]

� UL transfer can resume at any time during the improved extended phase



Packet Control Ack

P. UL ACK/NACK + polling

LLC PDU (last LLC)

T_MAX_EXTENDED_UL

USF

MS MFS SGSN

USF

...

start

NO ANSWER N3101 ++

NO ANSWER N3101 ++

N3101_POLLING_THR

P. UL ACK/NACK + polling

T_UL_RLS_EUTM

USF

USF

USF

Packet Control Ack

B10

When an MS enters Tbf UL releasing phase, BSS knows only if MS supports Extended UL (but can not know if

the MS supports Improved or Normal Extended UL mode)






1 � 2 � 100


UL TBF Release: Summary

� Delayed final PUAN (without extended mode)

� Extended UL TBF mode

UL TBF

release


TBF active

UL TBF

release

T_MAX_EXTENDED_UL

TBF active TBF extended

The UL TBF can be released before the expiry of the timer T_MAX_EXTENDED_UL: in case of a concurrent

DL TBF is present and after the completion of DL delayed phase, the expiry of T3192 triggers the release of

the UL TBF.

Consequently, the uplink TBF in extended mode is released when either T3192 expires or

T_MAX_EXTENDED_UL expires.






1 � 2 � 101

5 Exercises






1 � 2 � 102

5 Exercises

(E)GPRS Channels

� UL PDTCH and PACCH multiplexing on SPDCH:

� UL transfer? DL transfer?Downlink UplinkBlock number

TFI USF RRBP

Block n

Block n+1

Block n+2

Block n+3

Block n+4

Block n+5

Block n+6

TFI a USF j

TFI b USF k

TFI a USF j + 3

TFI b USF k

TFI b ???

TFI b USF j

TFI a USF k

false

false

false

false

false

false

PDTCH / PACCH a

PDTCH / PACCH b

PDTCH / PACCH a

PDTCH / PACCH b

PDTCH / PACCH b

PDTCH / PACCH a

PDTCH / PACCH b

RLC header MAC header

?

?

?

?

?

?

?

Block Content?

Back






1 � 2 � 103

5 Exercises

Autonomous Packet Resource Allocation

� Considering one cell with the following configuration:

� 4 TRXs in the DCS1800 band

� PS_PREF_BCCH_TRX = 0

� NB_TS_MPDCH = 0

� In the next slide, find the rank of each TRX

Time allowed:

10 minutes






1 � 2 � 104

5 Exercises

Autonomous Packet Resource Allocation [cont.]

0 1 2 3 4 5 6 7

BCCH SDCCHTRX1

TRX2 SDCCH

TRX3

TRX4

TRX_PREF_MARK

0

1

0

1

TRE G3

TRE G4 MP

TRE G3

TRE G4 MP FR

DR

FR

DR

RANK= ?

RANK= ?

RANK= ?

RANK= ?

Back






1 � 2 � 105

5 Exercises

Autonomous Packet Resource Allocation [cont.]

� Inputs:

� NB_TS = 14, AV_USED_CS=0, AV_USED_PS=11

� MIN_SPDCH=0

� MAX_SPDCH=14

� MAX_SPDCH_HIGH_LOAD=2

� HIGH_TRAFFIC_LOAD_GPRS=80%

� THR_MARGIN_PRIO_PS=10%

� Find the value of:

� MARGIN_CS, MARGIN_PRIORITY_PS

� MAX_SPDCH_LIMIT_CS, MAX_SPDCH_LIMIT_PS

� MAX_SPDCH_LIMIT

� Any Remark?

� Define a rule (relationship between some parameters) to fix this non-optimal setting

Time allowed:

30 minutes

Back






1 � 2 � 106

5 Exercises

TBF Radio Resources Allocation and Re-allocation

� GPRS allocation on BCCH TRX (7 PDCHs are allocated)

� UL TBFs 1 to 11 are established one after the other

� It is assumed that a concurrent DL TBF is established after the UL, and before the next UL

� All transfers are deemed DL biased

� Inputs:

� BCCH TRX, 1 unique SPDCH group

� NB_TS_MPDCH = 0

� MAX_XX_TBF_PER_SPDCH = 5

� En_Fast_Initial_GPRS_Access = disabled

� The MS GPRS multislot class is given below:

Time allowed:

20 minutesBack






1 � 2 � 107

5 Exercises

TBF Radio Resources Allocation and Re-allocation [cont.]

� Fill in the initial TBF allocation TS mapping UL/DL

DL

UL






1 � 2 � 108

5 Exercises


� GPRS/EGPRS allocation case:

� Cell with 2 PS capable TRXs:

� TRXa: EGPRS capable, SPDCH group = TS0 to TS7

� TRXb: non-EGPRS capable, SPDCH group = TS0 to TS7

� Ordered TRX list: TRXa > TRXb

� En_Fast_Initial_GPRS_Access = enabled

� MIN_PDCH = 1

� NB_TS_MPDCH = 0

� MAX_GPRS_CS = CS-3;

� MAX_EGPRS_MCS = MCS-6

� Default MS multislot class = 4 + 1

� All the MSs are 4 + 1 MSs

Time allowed:

20 minutesBack






1 � 2 � 109

5 Exercises


� MS constraints:

� Consecutive establishments:

� MSa: GPRS UL TBF followed by DL TBF

� MSb: EGPRS UL TBF followed by DL TBF

� MSc: GPRS UL TBF followed by DL TBF

� MSd: EGPRS UL TBF followed by DL TBF

� Bias = DL

Time allowed:

20 minutes

Rx Mx

0 1 2 3 4 5 6 7

DL

TxUL

0 1 2 3 4 5 6 7

Ttb Tra

Rx Rx Rx






1 � 2 � 110

5 Exercises


� Fill in the initial TBF allocation TS mapping UL/DL

DL

0 1 2 3 4 5 6 7PDCH

UL

UL

DL

TRXa

EGPRS

TRXb

GPRS

Rank

1

2






1 � 2 � 111

5 Exercises

DL TBF Release Routine

� DL TBF routine in acknowledged mode

� Fill in the blanks in the diagram of a DL TBF displayed on the next slide:

� name of the timers “T_?????”

� states of the DL TBF? ACTIVE, DELAYED, RELEASED

Time allowed:

10 minutes






1 � 2 � 112

5 Exercises

DL TBF Release Routine [cont.]

Back






1 � 2 � 113









1 � 2 � 114

End of ModuleRadio Resource management






Module 3Radio Link Control









EVOLIUM � E-GPRS Radio Algorithms and Parameters Description B10Radio Algorithms � Radio Link Control

1 � 3 � 2

Blank Page




Document History






1 � 3 � 3

Module Objectives


� Describe the algorithms of Radio Link Control and the related parameters






1 � 3 � 4








1 � 3 � 5

Table of Contents


1 GPRS CS Adaptation 72 EGPRS MCS Adaptation 223 RLC Blocks Retransmission 374 UL Power Control 465 NC0 Cell Selection and Reselection 506 NC2 Cell Reselection 637 Flow Control at the Gb Interface 898 Radio Link Supervision 969 Exercises 106






1 � 3 � 6








1 � 3 � 7

1 GPRS CS Adaptation






1 � 3 � 8


Introduction – GPRS and EGPRS

� The MFS decides the UL and DL changes of coding scheme

� CS adaptation is enabled by means of 2 parameters:

� EN_CS_ADAPTATION_ACK

� EN_CS_ADAPTATION_NACK

� These parameters apply to both GPRS and EGPRS TBFs

� GPRS

� 4 coding schemes: CS-1 to CS-4

� Only changes between consecutive CSs can occur

� Based on RXQUAL and I_LEVEL_TNi (interference level)

� EGPRS

� 9 modulation and coding schemes: MCS-1 to MCS-9

� Changes between any MCS can occur

� Based on MEAN_BEP and CV_BEP

CS-1: 20 useful bytes per RLC block.




MCS-1: 22 useful bytes per RLC block.






MCS-7: 2x56 useful bytes per RLC block.



Only changes between consecutive CSs can occur, except in case of defense mechanism, see “Defense

procedure” slide.






1 � 3 � 9


Principle

� The CS is adapted according to:

� QUALITY reporting

� I_LEVEL_TNi, only for DL TBF

� based on a comparison between the received level and the interference level

� BLER, only for DL TBF

� when I_LEVEL_TNi is not available

� for DL TBF:

� The MS makes measurements on all the received blocks on all the PDCHs

� The MS reports measurements in the “Packet DL Ack/Nack” messages

� Then, the MFS computes long term and short term averages

� for UL TBF:

� The BTS makes quality measurements on all TSs for each block sent by the MS

� Then the MFS computes long-term and short-term averages on all TSs






1 � 3 � 10


QUALITY Averaging in the MFS

� DL / UL TBF = 2 averages are computed:

� Short-term average AV_RXQUAL_ST

AV_RXQUAL_STn+1 = (1 - 1 / zn+1) * AV_RXQUAL_STn + (1 / zn+1) * RXQUALn,

with zn+1 = αST∆tn * zn + 1,

and αST = (1 - β)(1 / CS_AVG_PERIOD_ST)

� Long-term average AV_RXQUAL_LT

AV_RXQUAL_LTn+1 = (1 - 1 / yn+1) * AV_RXQUAL_LTn + (1 / yn+1) * RXQUALn,

with yn+1 = αLT∆tn * yn + 1,

and αLT = (1 - β)(1 / CS_AVG_PERIOD_LT)

In the formula above:

� RXQUALn is the RXQUAL value reported by the MS in the nth PACKET DL ACK/NACK message.

� ∆tn is the time difference in seconds between the (n-1)th and the nth PACKET DL ACK/NACK messages, therefore depending on DL_ACK_PERIOD parameter value, on the nb of PDCHs used by the MS and on the

traffic of the other MSs multiplexed on these PDCHs.

� AV_RXQUAL_STn (respectively AV_RXQUAL_LTn) is the value of AV_RXQUAL_ST (respectively

AV_RXQUAL_LT) after the nth PACKET DL ACK/NACK message.

� β is a hard coded end equal to 0.9.

Remark: the initial value of yn and zn is 0.

1/CS_AVG_PERIOD_LT and 1/CS_AVG_PERIOD_ST correspond to forgetting factors: number of seconds in the

past above which Quality measurements are considered as too old to be taken into account in the average.

Default values are:

� CS_AVG_PERIOD_ST = 0.32 s

� CS_AVG_PERIOD_LT = 2 s

both of them can be set at OMC-R level (cell level).






1 � 3 � 11


AV_SIR Computation in the MFS

� Interference level averaged by the MS

� γCH,n = (1-d) * γCH,n-1 + d * SSCH,n� d is the forgetting factor = d = 1/ MIN(n, N_AVG_I)

� SSCH,n is the interference measurement at iteration n

� I_LEVEL_TNi computed by the MS and sent to the MFS

� I_LEVEL_TNi = 0 when γCH > C

� I_LEVEL_TNi = 1 when C-2dB < γCH <= C

� I_LEVEL_TNi = 2 when C-4dB < γCH <= C-2dB

� Etc.

� I_LEVEL_TNi = 14 when C-28dB < γCH <= C-26dB

� I_LEVEL_TNi = 15 when γCH <= C-28dB

� AV_SIR = average value of I_LEVEL_TNi of all assigned DL TS i

C_VALUE and I_LEVEL_TNi measurements are already averaged with an exponential filter in the MS.

Therefore, additional averaging is not needed, which reduces the complexity, i.e., AV_SIRn = I_LEVEL_TNin

where n is the number of the packet downlink Ack/Nack message.

Default values is :

� N_AVG_I: 6

� It Can be set at the OMC-R level (cell level).

For more details about measurements and averages performed by the MS, see 3GPP 05.08.






1 � 3 � 12


DL CS Adaptation

O&M threshold

and hysteresis

new CS

current CS

- AV_RXQUAL_ST

- AV_RXQUAL_LT

- AV_SIR

MS MFS

(RXQUAL, I_Level_TNi)

Packet DL Ack/Nack

(RXQUAL, I_Level_TNi)

Packet DL Ack/Nack

Averaging

Link

adaptation

� Functional process






1 � 3 � 13


DL CS Adaptation [cont.]

� Threshold comparison when I_LEVEL_TNi are reported

� X = FH or NFH

� Y = ACK or NACK

AV_SIR < CS_SIR_DL_3_4_X_Y + CS_SIR_HST_DLNot possibleCS-4

AV_RXQUAL_LT > CS_QUAL_DL_2_3_X_Y + CS_HST_DL_LT

OR

AV_RXQUAL_ST > CS_QUAL_DL_2_3_X_Y + CS_HST_DL_ST

AV_RXQUAL_LT < CS_QUAL_DL_3_4_X_Y

AND

AV_SIR > CS_SIR_DL_3_4_X_Y

CS-3


OR


AV_RXQUAL_LT < CS_QUAL_DL_2_3_X_YCS-2

Not possibleAV_RXQUAL_LT < CS_QUAL_DL_1_2_X_YCS-1

Decreasing the coding scheme number

(CSi � CSi-1)

Increasing the coding scheme number

(CSi � CSi+1)

Current coding scheme

AV_RXQUAL_ST is a short term average whereas AV_RXQUAL_LT is a long term average. The short term

average is used to react quickly in case of fast degradation of the radio conditions.

X = FH or NFH: two thresholds are available for hopping and non-hopping TRXs.

Y = ACK or NACK: two thresholds are available for RLC acknowledged and unacknowledged modes.

The thresholds should be chosen so that: CS_HST_DL_ST > CS_HST_DL_LT > 0






1 � 3 � 14



� Threshold comparison when I_LEVEL_TNi are NOT reported

� X = FH or NFH

� Y = ACK or NACK

CS4_BLER > CS_BLER_DL_4_3Not possibleCS-4


OR


AV_RXQUAL_LT < CS_QUAL_DL_3_4_X_Y

AND

CS3_BLER < CS_BLER_DL_3_4

CS-3


OR


AV_RXQUAL_LT < CS_QUAL_DL_2_3_X_YCS-2

Not possibleAV_RXQUAL_LT < CS_QUAL_DL_1_2_X_YCS-1


(CSi � CSi-1)


(CSi � CSi+1)


As it has been observed (in the Alcatel labs during the B8 release validation) that some MSs do not report

any interference measurements when the BCCH carrier is included in the frequency hopping sequence of

the allocated PDCH, the algorithm described above is slightly modified in the MR2 version of the B8 release.

A new triggering condition is used for the CS change between CS3 and CS4. This new triggering condition

shall be applied only to the TBF that do not report any interference level measurements. Each time a

Packet DL Ack/Nack message is received:

� either it contains no interference measurement and the new algorithm is applied,

� or it contains interference measurements and the standard algorithm is applied.

With the new algorithm, the interference level is replaced by the BLER (RLC BLock Error Rate):

� the CS3 BLER is used for a CS change from CS3 to CS4,

� the CS4 BLER is used for a CS change from CS4 to CS3.

Remarks:

� Case of a DL TBF with PDCH allocated on the BCCH TRX and no frequency hopping on the BCCH TRX: the

MS does not report any interference level measurement in the Packet DL Ack/Nack message (no

interference measurement on the BCCH carrier).

� Case of a DL TBF with PDCH having the BCCH carrier belonging to the frequency hopping sequence:

depending on MS implementation, some MSs may not report any interference measurement (behavior

observed in the Alcatel labs during the B8 release validation).






1 � 3 � 15



� SummaryAV_RXQUAL_LT

AV_SIR

CS-1

CS-2

CS-3CS-4

CS_QUAL_DL_1_2_X_Y+ CS_HST_DL_LT

CS_QUAL_DL_1_2_X_Y

CS_QUAL_DL_2_3_X_Y

C S_QUAL_DL_3_4_X_Y

0

7

0 15CS_SIR_DL_3_4_X_Y CS_SIR_DL_3_4_X_Y +

CS_SIR_HST_DL

CS-1 or CS-2 (hysteresis)

CS-2 or CS-3 (hysteresis)

CS-3 or CS-4

(hysteresis)

CS_QUAL_DL_2_3_X_Y + CS_HST_DL_LT

BLER

100% 0%CS_BLER_DL_3_4 CS_BLER_DL_4_3

The change from CS-3 to CS-4 is not only based on AV_RXQUAL_LT for the two following reasons:

� RXQUAL range only goes down to 0.2%. However, the change of the coding scheme from CS-3 to CS-4 will

probably have to be done for even lower values. Indeed, when the coding scheme is CS-4, in static

(AWGN), a BLER of 0.1 (typical value of the BLER threshold to change from CS-3 to CS-4) is obtained for a

raw BER of 1-(1-0.1)1/456 = 2.10-4. This raw BER would be larger in multipath channels but is likely to

remain below 0.2%. This means that CS_QUAL_DL_3_4 should be close to 0 and that a condition based on

RXQUAL is not sufficient to change the coding scheme from CS-3 to CS-4.

� If the changes from CS-3 to CS-4 and from CS-4 to CS-3 are based on different metrics, a Ping-Pong effect

may occur. Indeed, it may happen that the conditions to change from CS-3 to CS-4 and CS-4 to CS-3 are

simultaneously true in some situations.






1 � 3 � 16


UL CS Adaptation [cont.]


O&M threshold

and hysteresis

new CS

current CS

- AV_RXQUAL_ST

- AV_RXQUAL_LT

UL RLC block

Averaging

Link

adaptation

MS MFSBTS

RXQUAL

measurement

UL RLC block (RXQUAL)

UL RLC block (RXQUAL)






1 � 3 � 17



� Threshold comparison

� X = FH or NFH

� Y = ACK or NACK

AV_RXQUAL_LT > CS_QUAL_UL_3_4_X_Y + CS_HST_UL_LT

OR

AV_RXQUAL_ST > CS_QUAL_UL_3_4_X_Y + CS_HST_UL_ST

Not possibleCS-4


OR


AV_RXQUAL_LT < CS_QUAL_UL_3_4_X_YCS-3


OR


AV_RXQUAL_LT < CS_QUAL_UL_2_3_X_YCS-2

Not possibleAV_RXQUAL_LT < CS_QUAL_UL_1_2_X_YCS-1


(CSi � CSi-1)


(CSi � CSi+1)


AV_RXQUAL_ST is a short term average whereas AV_RXQUAL_LT is a long term average. The short term

average is used to react quickly in case of fast degradation of the radio conditions.

X = FH or NFH: two thresholds are available for hopping and non-hopping TRXs.

Y = ACK or NACK: two thresholds are available for RLC acknowledged and unacknowledged modes.

The thresholds should be chosen so that: CS_HST_UL_ST > CS_HST_UL_LT > 0






1 � 3 � 18



� SummaryAV_RXQUAL_LT

AV_SIR

CS1

CS2

CS_QUAL_UL_1_2_X_Y + CS_HST_UL_LT

CS_QUAL_UL_1_2_X_Y

CS_QUAL_UL_2_3_X_Y

0

7

0 15

CS1 or CS2 (hysteresis)

CS2 or CS3 (hysteresis)

CS3

CS4

CS3 or CS4 (hysteresis)CS_QUAL_UL_3_4_X_Y



In the uplink, the RXQUAL is available in CS-4 and the SIR measurements are not reported by the BTS to the

MFS so far. Therefore, it is also possible to use RXQUAL measurements to change the coding scheme from

CS-3 to CS-4 or from CS-4 to CS-3, contrary to the downlink algorithm, where the SIR was used.






1 � 3 � 19


Execution

� UL TBF:

� the CS to be used is indicated to the MS during the establishment phase

� if a CS adaptation is decided by the MFS during the transfer phase, a PACKET UL ACK/NACK message is sent immediately to the MS

� DL TBF:

� if a CS adaptation is decided by the MFS during the transfer phase, the MFS modifies the CS






1 � 3 � 20


Defense Procedure

� In a DL TBF:

If the number of PACKET DL ACK/NACK messages consecutively lost from the MS on the radio interface goes over TBF_CS_DL, the coding scheme is changed to CS-1

� In a UL TBF:

If the number of radio blocks consecutively not decoded goes over the threshold Nb_allocated_TS x TBF_CS_UL, the coding scheme is changed to CS-1

� In both cases, the CS must not be changed again before TBF_CS_PERIOD RLC blocks are transmitted

B10

TBF_CS_DL = 8 (Alcatel recommended value) and it can be set at OMC-R level.

TBF_CS_UL = 32 (Alcatel recommended value) and it can be set at OMC-R level.

TBF_CS_PERIOD = 20 (Alcatel recommended value) and it cannot be set at OMC-R level.






1 � 3 � 21


Initial Coding Scheme

� The initial CS at TBF establishment is given by the cell parameters:

� TBF_DL_INIT_CS for a DL TBF & TBF_UL_INIT_CS for a UL TBF

� Range = CS-1, CS-2, CS-3, CS-4

� Default value = CS-2

� T_DL_GPRS_MeasReport: the time period to request for a “Packet Downlink Ack/Nack” with measurements

� Range: from 60 to 3000 ms

� Default value = 400 ms

� The initial CS and CS changes are limited by the cell parameter MAX_GPRS_CS

� Range = CS-2, CS-3, CS-4

� Default value = CS-2

Exercise

Rules:

� TBF_DL_INIT_CS < MAX_GPRS_CS

� TBF_UL_INIT_CS < MAX_GPRS_CS

When a new LLC PDU is received and the downlink transfer is resumed, the timer defined by

CS_MAX_IDLE_PERIOD shall be checked. If it has not expired, then the coding scheme or modulation and

coding scheme of the previous DL TBF shall be re-used.






1 � 3 � 22

2 EGPRS MCS Adaptation






1 � 3 � 23


Impact of the Output Power – DL Case

� The Average Power Decrease (APD)

� = max(0, modulation_delta_power - |BS_TXPWR_MAX|)

� Used in the MCS adaptation

TRE 1 TRE 2 TRE 3

353637383940414243444546474849

TRE 1 TRE 2 TRE 3

TR

E o

utp

ut p

ow

er (d

Bm

)

APD (dB)

8-PSK attenuation (dB)

8-PSK TRE power (dBm)

GMSK TRE power (dBm)

GMSK power in the cell (dBm)

Max GMSK power in the cell (dBm)

Max Power in the cell (dBm)


APD: Average Power Decrease

� The back-off between average GMSK and 8-PSK output power comes from physics since 8-PSK is a non-

constant envelope modulation unlike GMSK.

� As a consequence, power amplifiers can not be used at their maximum power. This results in a difference

between mean output powers for GMSK and 8-PSK modulations.

Output power handling

� The BTS sets all the TREs which transmit GMSK output powers at the same level which is the minimum

value among the maximum TRE output power in a sector and in a given band.

� On a TRE, the maximum GMSK output power is higher than the maximum 8-PSK output power.

� An O&M parameter (BS_TXPWR_MAX) allows a static power reduction of the maximum GMSK output power

of the sector.

� The TRE transmit power in 8-PSK shall not exceed the GMSK transmit power in the sector.

� The BTS determines for each TRE, the difference between the 8-PSK output power of the TRE and the

GMSK output power of the sector (8-PSK delta power).

� According to the 8-PSK delta power value, a TRE is called “High Power” or “Medium Power”.

� When a GCH channel is activated, the BTS sends the 8-PSK delta power to the MFS.

Together with BS_TXPWR_MAX (static power reduction), the 8-PSK delta power allows the MFS to

determine:

� a possible attenuation (BS_TX_PWR) for the 8-PSK DL RLC block emission, in order not to exceed the

GMSK power of the sector (for GMSK DL RLC block, the attenuation is BS_TXPWR_MAX).

� an Average Power Decrease which is the difference between the 8-PSK output power and the GMSK

output power after having taken into account BS_TXPWR_MAX. The Average Power Decrease is taken

into account in the link adaptation tables.






1 � 3 � 24


Impact of the Output Power – UL Case

� The APD of a mobile station is the difference between the maximum output power in GMSK and the maximum output power in 8-PSK

� The maximum output powers are known by "GMSK Power Class" and "8-PSK Power Class" fields of the MS Radio Access capability

� Examples of APD in case of GSM 900 and GSM 850:

APD = 6APD = 2APD = 0GMSK: Power Class 5

Max. output power = 29 dBm


Max. output power=33 dBm





8-PSK: Power Class E3





Max. output power=33 dBm






1 � 3 � 25


Measurement Reporting

� MEAN_BEP

� Average Bit Error Probability

� Range

� From 0 to 31

� MEAN_BEP = 0 means actual BEP > 25%

� MEAN_BEP = 31 means actual BEP < 0.025%

� CV_BEP

� Average coefficient of variation of the Bit Error Probability

� Range

� From 0 to 7

� CV_BEP = 0 means [1.75 < actual CV_BEP < 2.00]

� CV_BEP = 7 means [0.00 < actual CV_BEP < 0.25]






1 � 3 � 26


Measurement Reporting [cont.]

� The MCS is adapted according to MEAN_BEP and CV_BEP reporting (based on the Bit Error Probability)

� For a DL TBF:

� The MS makes MEAN_BEP and CV_BEP measurements on all the received blocks the header of which has been well decoded

� The MS computes MEAN_BEP and CV_BEP averages� Based on the forgetting factor principle

� The BEP_PERIOD cell parameter is used to compute the forgetting factor (default = 10)

� The MS reports MEAN_BEP and CV_BEP averages in the “EGPRS Packet DL Ack/Nack” messages

� For a UL TBF:

� The BTS makes MEAN_BEP and CV_BEP measurements on all the received blocks the header of which has been well decoded

For more details about MEAN_BEP and CV_BEP averages performed by the MS, refer to 3GPP 05.08.

Raw measurements on a radio block basis

� For EGPRS (that is during an EGPRS DL TBF), the MS shall calculate the following values, for each radio block (1 radio block = 4 bursts) addressed to it

(the DL TBF TFI contained in the radio block must be decoded):

� Mean Bit Error Probability (BEP) of a radio block:

� Coefficient of variation of the Bit Error Probability of a radio block:

� In the above equations, the BEP is measured on a burst basis by the MS before channel decoding.

Averaging of the raw measurements on a TS basis

� The raw measurements made by the MS on a radio block basis are averaged by the MS per TS (TN in the below equations) and per modulation type

(GMSK (MCS1 to MCS4), 8-PSK (MCS5 to MCS9)) as follows:

�

�

� with (Rn gives the reliability of the averaged quality parameters)

� In the above equations:

� n is the iteration index, incremented for each DL radio block,

� e is a forgetting factor and is calculated according to the BEP_PERIOD cell parameter,

SEE NEXT SLIDE

∑=

=4

14

1_

i

iburstblockBEPBEPMEAN

∑

∑ ∑

=

= =

−=

4

1

24

1

4

1

4

1

4

1

3

1

_

i

iburst

k i

iburstkburst

block

BEP

BEPBEP

BEPCV

nblock,n

n1n

n

nn MEAN_BEP

R

xeNMEAN_BEP_T)

R

xe(1NMEAN_BEP_T ⋅⋅+⋅⋅−= −

nblock,

n

n1n

n

nn CV_BEP

R

xeCV_BEP_TN)

R

xe(1CV_BEP_TN ⋅⋅+⋅⋅−= −

0R ,xeRe)(1R 1n1nn =⋅+⋅−= −−






1 � 3 � 27


DL MCS Adaptation


APD IR

link adapatation

tables

new MCS

current MCS

MS MFS

(Mean_BEP, CV_BEP)

EGPRS Packet DL Ack/Nack

Link

adaptation

Modulation

type

� xn denotes the existence of quality parameters for the nth block, i.e. if the radio block is intended for this MS. xn values 1 and

0 denote the existence and absence of quality parameters, respectively.

Measurements reporting

� An MS shall report the overall MEAN_BEP and CV_BEP (instead of reporting the RXQUAL and SIGN_VAR values) per modulation type

(that is GMSK_MEAN_BEP, GMSK_CV_BEP and/or 8-PSK_MEAN_BEP, 8-PSK_CV_BEP depending on the received blocks since the last

channel quality report sent to the network) averaged over all allocated channels (time slots) as follows:

� ,

� where n is the iteration index at reporting time and j the TS number.

� The MS reports the Mean_BEP and CV_BEP values to the MFS in the Channel Quality Report included in the EGPRS Packet DL Ack/Nack

and Packet Resource Request messages.

� The MS can report 32 different Mean_BEP values (MEAN_BEP_0 to MEAN_BEP_31). The mapping between the calculated Mean_BEP

value (linear scale) and the reported Mean_BEP value (logarithmic scale) depends on the used modulation (two mapping tables are

given in the 05.08 GSM recommendation: one for GMSK and one for 8-PSK).

� The MS can report 8 different CV_BEP values (CV_BEP_0 to CV_BEP_7). The mapping between the calculated and the reported values

is identical for the GMSK and 8-PSK modulations.

Measurements and reporting at BTS side

The BTS measures for each UL burst the BEP and calculates for each UL radio block (4 bursts) the Mean_BEP and the CV_BEP = Std_BEP

/ Mean_BEP. The Mean_BEP and the CV_BEP are reported on a radio block basis by the BTS to the MFS.

∑

∑ ⋅=

j

(j)

n

j

(j)

n

(j)

n

nR

NMEAN_BEP_TR

MEAN_BEP

∑

∑ ⋅=

j

(j)

n

j

(j)

n

(j)

n

nR

CV_BEP_TNR

CV_BEP






1 � 3 � 28


DL MCS Adaptation [cont.]

� In RLC acknowledged mode, the MFS applies a given MCS taking into account:

� Current MCS:

� MCS-1 to MCS-4 in GMSK

� MCS-5 to MCS-9 in 8-PSK

� Average Power Decrease: APD set = [0, 1, 3, 4, 5, 6, 8, 10]

� MS OUT OF MEMORY state

� = Off then LA tables with IR are used

� = On then LA tables without IR (Type I ARQ) are used

� In RLC unacknowledged mode, the MFS applies a given MCS taking into account:

� Current MCS:


� MCS-5 to MCS-9 In 8-PSK


� Each combination of these criteria corresponds to a specific LA table

LA = Link Adaptation. IR = Incremental Redundancy, also called Type II ARQ (Automatic ReQuest for

repetition).

Extract of an LA table when APD=0dB, Type 1 ARQ, 8-PSK table: if the current MCS belongs to {5,6,7,8,9}

0 1 2 3 4 5 6 7

0 5 5 5 5 1 1 1 1

1 5 5 5 5 1 1 2 2

2 5 5 5 5 1 2 2 2

3 5 5 5 5 2 2 2 3

4 5 5 5 5 5 2 3 3

5 5 5 5 5 5 3 3 3

6 5 5 6 5 5 5 3 3

7 5 5 6 5 5 5 3 3

8 5 5 6 6 5 5 5 4

9 5 6 6 6 5 5 5 5

10 5 6 6 6 6 5 5 5

11 6 6 6 6 6 6 5 5

12 6 6 6 6 6 6 5 5

13 6 6 6 6 6 6 5 5

14 7 6 6 6 6 6 6 6

15 7 6 6 6 6 6 6 6

16 7 7 6 7 6 6 6 6

17 7 7 7 7 7 6 6 6

18 7 7 7 7 7 7 7 7

19 7 7 7 7 7 7 7 7

20 7 7 7 7 7 7 7 7

21 7 7 7 7 7 7 7 7

22 7 8 8 8 8 8 8 8

23 8 8 8 8 8 8 8 8

24 8 8 8 8 8 8 8 8

CV_BEP

ME

AN

_BE

P

�If the effective APD

(=max(0,

modulation_delta_power -

|BS_TX_PWR_Max|) does not

belong to the APD set which

is described above, then the

APD value in the set which is

immediately higher than the

received APD.

�e.g., if effective APD = 0.7

dB then APD = 1 dB






1 � 3 � 29


RLC ACK Mode: New DL MCS Value Determination

� Evaluated every “EGPRS Packet DL Ack/Nack” message

� If MCS indicated by LA tables <= current MCS

� Then New MCS = MCS indicated by LA tables

� Else New MCS = MCS indicated by LA tables with the modified criteria

� MEAN_BEP = max[(current MEAN_BEP)-2, 0]

� CV_BEP = current CV_BEP

Example:

� The TBF mode is acknowledged: use of the RLC acknowledged mode algorithm and tables,

� MS OUT OF MEMORY = On: use of the group of tables for Type I ARQ (without IR),

� APD = 0 dB: use of the group of tables for APD = 0 dB (for each APD value, there is a GMSK table (for MCS

= MCS1, … , MCS4) and a 8_PSK table (for MCS = MCS5, …, MCS9)),

� current MCS = MCS6: use of the 8_PSK table for APD = 0 dB.

� If the MS reports the (MEAN_BEP = 3, CV_BEP = 2) values in the last Packet DL Ack/Nack message, the link

adaptation table indicates MCS5. As MCS5 < MCS6, the commanded MCS is MCS5.

� If the MS reports the (MEAN_BEP = 23, CV_BEP = 3) values in the last Packet DL Ack/Nack message, the

link adaptation table indicates MCS8. As MCS8 > MCS6, the commanded MCS is the MCS corresponding to

the (MEAN_BEP = 23 – 2 = 21, CV_BEP = 3) couple in the link adaptation table, that is MCS7.






1 � 3 � 30


RLC NACK Mode: New DL MCS Value Determination

� Evaluated every “EGPRS Packet DL Ack/Nack” message

� If MCS indicated by LA tables <= current MCS

� Then New MCS = MCS indicated by LA tables

� Else New MCS = MCS indicated by LA tables with the following criteria

� MEAN_BEP = max[(current MEAN_BEP)-8, 0]

� CV_BEP = current CV_BEP

� Exception: if � Current MEAN_BEP = 31

� AND Current CV_BEP = 7

� AND Current MCS belongs to {MCS-1, MCS-2, MCS-3, MCS-4)

� Then New MCS = MCS5

Exercise

The margin of 8 for the hysteresis has been chosen to have a long term average weighted BER close to

0.001. With this value, the MCS selected can never be higher than 7 in good radio conditions.






1 � 3 � 31


UL MCS Adaptation


UL RLC b lock (CV_BEP, Mean_BEP)

new MCS

current MCS

UL RLC b lock

Averaging

Link

adaptation

MS MFS BTS

CV_BEP, Mean_BEP

computation

UL RLC b lock (CV_BEP, Mean_BEP)

lin k adapatation

tables

APD

The MFS calculates average values (Mean_BEP and CV_BEP) each time a radio block is received from the

BTS. Then the MFS checks if an MCS change is needed using internal tables. However, the first decision shall

only be taken when TBF_MCS_Period radio blocks have been received.

The measurements performed by the BTS are averaged by the MFS as follows:

� yn+1 = e∆tn * yn +1 where e (forgetting factor) is equal to (1 - 0.9)(1 / MCS_AVG_PERIOD),� Mean_BEPn+1 = (1 - 1 / yn+1) * Mean_BEPn + (1 / yn+1) * Mean_BEPblock n,

� CV_BEPn+1 = (1 - 1 / yn+1) * CV_BEPn + (1 / yn+1) * CV_BEPblock n

� MCS_AVG_PERIOD = 0,1s (Alcatel recommended value) and it cannot be set at OMC-R level.






1 � 3 � 32


UL MCS Adaptation [cont.]

� In RLC acknowledged mode, the MFS applies a given MCS taking into account:

� Current MCS:


� MCS-5 to MCS-9 in 8-PSK

� APD of the MS: APD set = [0, 1, 3, 4, 5, 6, 8, 10]

� EN_IR_UL state

� = enabled then LA tables with IR are used

� = disabled then LA tables without IR (Type I ARQ) are used

� In RLC unacknowledged mode, the MFS applies a given MCS taking into account:

� Current MCS:


� MCS-5 to MCS-9 In 8-PSK


� Each combination of those criteria corresponds to a specific LA table

Modified B10

The same tables apply in the uplink as in the downlink.






1 � 3 � 33


RLC ACK Mode: New UL MCS Value Determination

� Evaluated every 12 radio blocks (decision window)

� For each radio block, the MFS computes an indicated MCS (MCSind)according to MEAN_BEP, CV_BEP and the appropriate LA table

� During the decision window:

� If MCSind > current MCS then N_sup = N_sup + 1

� If MCSind <= current MCS then N_inf = N_inf + 1

� At the end of the decision window:

� If N_inf > 6 and MCSindlast_block < current MCS

� Then New MCS = MCSindlast_block

� Else if N_sup > 6 and MCSindlast_block > current MCS

� Then New MCS = Max[current MCS, MCSindlast_block modified]� Where MCSindlast_block modified is computed with the following criteria

� MEAN_BEP = Max[MEAN_BEPlast_block-2, 0]

� CV_BEP = CV_BEPlast_block

� Else, New MCS = current MCS

The following complex algorithm is used by the MFS to determine the MCS to be used in RLC acknowledged

mode:

� the MFS determines the MCS to be used every 12 radio blocks (decision window),

� during the decision window, for each received measurement (that is for each received radio block), the

averaged (Mean_BEP, CV_BEP) couple indicates a best MCS (called MCSind) according to the appropriate link

adaptation table. This best MCS is compared to the current MCS, and 2 counters (N_sup, N_inf) are

maintained (N_sup (respectively N_inf), is incremented by one when MCSind is higher (respectively lower)

than the current MCS) that gives, for the current decision window, the number of MCSind that are higher or

equal to the current MCS,

� at the end of the decision window, the decision process is as follows:

� the new MCS is determined according to the trend observed during the decision window (that is

trend towards upper MCS or trend towards lower MCS). It is considered that a trend towards upper

(respectively lower) MCS is observed if strictly more than half (that is 6) of the MCSind are higher

(respectively lower) than the current MCS,

� moreover, the new MCS is applied only if the last MCSind of the decision window corresponds to the

trend observed during the decision window (that is MCSindlast block > current MCS for a trend

towards upper MCS or MCSindlast block < current MCS for a trend towards lower MCS),

� finally, the new MCS is the MCSind of the last block (MCSindlast block) in case of trend towards a

lower MCS. In case of trend towards upper MCS, an hysteresis is applied on the measurements as the

new MCS in that case is equal to max(current MCS, MCSindlast block (max(Mean_BEP - 2, 0),

CV_BEP)).






1 � 3 � 34


RLC NACK Mode: New UL MCS Value Determination

� Evaluated every 12 radio blocks (decision window)

� If MCSindlast_block <= current MCS

� Then New MCS = MCSindlast_block

� Else

� Then New MCS = Max[current MCS, MCSindlast_block modified]� Where MCSindlast_block modified is computed with the following criteria

� MEAN_BEP = Max[MEAN_BEPlast_block-8, 0]

� CV_BEP = CV_BEPlast_block

Exercise






1 � 3 � 35


Defense Procedure

� In a DL TBF:

If the number of EGPRS PACKET DL ACK/NACK messages consecutively lost from the MS on the radio interface goes over TBF_MCS_DL, the modulation and coding scheme are changed to MCS-1

� In a UL TBF:

If the number of radio blocks consecutively not decoded goes over the threshold Nb_allocated_TS x TBF_MCS_UL, the modulation and coding scheme are changed to MCS-1

� In both cases, the MCS must not be changed again before TBF_MCS_PERIOD RLC blocks are transmitted

TBF_MCS_DL = 12 (Alcatel recommended value) and it can be set at OMC-R level.

TBF_MCS_UL = 32 (Alcatel recommended value) and it can be set at OMC-R level.

TBF_MCS_PERIOD = 12 (Alcatel recommended value) and it cannot be set at OMC-R level.






1 � 3 � 36


Initial MCS

� The initial MCS at TBF establishment is given by the cell parameters:

� TBF_DL_INIT_MCS for a DL TBF & TBF_UL_INIT_MCS for a UL TBF

� Range for a DL TBF = from MCS-1 to MCS-9

� Range for a UL TBF = from MCS-1 to MCS-9

� Default value = MCS-3

� T_DL_EGPRS_MeasReport: the time period to request for a “EGPRS Packet Downlink Ack/Nack” with measurements

� Values: from 60 to 3000 ms

� Default value = 200 ms

� The initial MCS and MCS changes are limited by the cell parameter MAX_EGPRS_MCS

� Range = from MCS-2 to MCS-9

� Default value = MCS-9

Rules:

� TBF_DL_INIT_MCS < MAX_EGPRS_MCS

� TBF_UL_INIT_MCS < MAX_EGPRS_MCS

When a new LLC PDU is received and the downlink transfer is resumed, the timer defined by

CS_MAX_IDLE_PERIOD shall be checked. If it has not expired, then the coding scheme or modulation and

coding scheme of the previous DL TBF shall be re-used.






1 � 3 � 37

3 RLC Blocks Retransmission






1 � 3 � 38


New Modulation and Coding Schemes

� Data rate per radio TS (RLC payload)

Scheme Modulation Maximum rate (kbps)

GPRS

CS-4 GMSK 20

CS-3 GMSK 14.4

CS-2 GMSK 12

CS-1 GMSK 8

EGPRS

MCS-9 8PSK 59.2

MCS-8 8PSK 54.4

MCS-7 8PSK 44.8

MCS-6 8PSK 29.6 / 27.2 *

MCS-5 8PSK 22.4

MCS-4 GMSK 17.6

MCS-3 GMSK 14.8 / 13.6 *

MCS-2 GMSK 11.2

MCS-1 GMSK 8.8

* case of padding

8-PSK modulation

to provide

higher data rates

GMSK modulation to

ensure a certain level of

performance in case of

poor radio conditions

MCSs are defined only for the EGPRS packet data traffic channels (PDTCH). For all the EGPRS packet control

channels, the corresponding GPRS control channel coding is used (that is CS1 for the PACCH, PBCCH,

PAGCH, PPCH and downlink PTCCH).






1 � 3 � 39


New Modulation and Coding Schemes [cont.]

� MCSs are divided into 4 families: A, A padding, B and C

MCS-5 MCS-6 MCS-7 MCS-8 MCS-9MCS-1 MCS-2 MCS-3 MCS-4

FamilyC

FamilyB

FamilyA

padding

FamilyA

28

22

34+3

22 22

28 28

34+3 34+3

28 28

28 28

34 34

34 34

37 37 37 37 37

37 37

GMSK 8-PSK

28RLC data block Unit of payload (in

bytes)

Radio data

block

The main GPRS imperfections are linked to:

� the design of the GPRS coding schemes which were designed independently from the others with their

own data unit.

� the fact that once the information contained in a radio block has been transmitted with a certain CS, it is

not possible via the Automatic ReQuest for repetition (ARQ) mechanism to retransmit with another CS.

� This could lead to the release of the TBF and to the establishment of a new one in order to transmit

the LLC frame.

EGPRS coding schemes have been designed to offset this problem. Four MCS families have been created

with for each of them a basic unit of payload.

� This allows the re-segmentation of the RLC data blocks when changing of modulation and coding schemes

(within the same family).

� Example: if one MCS-6 radio block has not been received correctly by the receiver and if radio

conditions have degraded in the meantime, it is possible to re-send the same information in two

radio blocks with MCS-3 (more protection).

� The level of protection applied (MCS usage) in case of retransmissions is in line with the radio conditions.

The different code rates within a family are achieved by transmitting a different number of payload units

within one radio block.

When 4 payload units are transmitted, these are split into 2 separate RLC blocks (i.e., with separate

sequence numbers).






1 � 3 � 40


Automatic ReQuest for Repetition (ARQ)

� In RLC ACK mode, the retransmission can be performed using:

� Selective type I ARQ mechanism

� The blocks which are not decoded are simply retransmitted

� Available since B6 release

� Another MCS from the same family can be used

� Type II hybrid ARQ mechanism (also called Incremental Redundancy (IR))

� The blocks which are not decoded are retransmitted� Using or not another MCS of the same family

� Using a different Puncturing Scheme (PS)

� The non decoded block and the retransmitted one(s) are “soft combined” to retrieve the right information

� This applies only when the MS memory for IR is not full

� This can apply for both UL and DL EGPRS TBF

Appendix

The type 2 ARQ mechanism or incremental redundancy (IR) is an ETSI function, mandatory for the EGPRS MS

receiver (downlink path) and optional for the BTS receiver (uplink path).

The incremental redundancy is not used for the signaling blocks, the GPRS data blocks and the data blocks

in RLC unacknowledged mode. It is only used for the EGPRS data blocks in RLC acknowledged mode.

In the type II ARQ mechanism (IR):

� the first emission of an RLC data block is done using a first puncturing scheme (PS1),

� in case of re-transmission of this RLC block, the transmitter uses the same MCS or an MCS of the same

family as the one used for the initial block. The re-segmentation of the RLC block may be performed or not,

� at the output of the demodulator, the receiver combines the information of soft bits corresponding to the

first transmission of the block and its different re-transmissions, thus increasing the decoding probability of

the RLC block.

� Remark: according to the 04.60 (RLC/MAC layers) GSM recommendation, the soft combining inside the MS

receiver is only performed between:

� an MCSx block and an MCSx block (that is the same MCS is used for the re-transmission),

� an MCS9 block and an MCS6 block (in that case the RLC data blocks carry the same number of

payload units),

� an MCS7 block and an MCS5 block (in that case the RLC data blocks carry the same number of

payload units).






1 � 3 � 41


Type I ARQ Mechanism

� MCS selection for retransmission of a DL or UL RLC data block

Commanded MCS (given by the link adaptation algorithm)

MCS9 MCS8 MCS7 MCS6 MCS5 MCS4 MCS3 MCS2 MCS1

MCS9 MCS9 MCS6 MCS6 MCS6 MCS3 MCS3 MCS3 MCS3 MCS3

MCS8 MCS8 MCS8 MCS6 (pad.)

MCS6 (pad.)

MCS3 (pad.)

MCS3 (pad.)

MCS3 (pad.)

MCS3 (pad.)

MCS3 (pad.)







Initial MCS


With the type 1 ARQ mechanism, the decoding of a re-transmitted RLC block does not use the previously

transmitted versions (not correctly received) of this RLC block. The decoding of an RLC data block is only

based on the current transmission.

The type 1 ARQ mechanism is always used for the GPRS.






1 � 3 � 42


Type I ARQ Mechanism [cont.]

� Example of a UL EGPRS TBFMS BSS

UL RLC data block B1, MCS4, PS1

Packet UL Ack/Nack (B1 not decoded)

UL RLC data block first part B1, MCS1, PS1

UL RLC data block second part B1, MCS1, PS1

B1 block not decoded by the BTS

Resegment bit set

Second part of B1 block not decoded by the BTS

Packet UL Ack/Nack (B1 not decoded)Resegment bit set

UL RLC data block first part B1, MCS1

UL RLC data block second part B1, MCS1

MS BSS

UL RLC data block B1, MCS4, PS1

Packet UL Ack/Nack (B1 not decoded)

UL RLC data block first part B1, MCS1, PS1

UL RLC data block second part B1, MCS1, PS1

B1 block not decoded by the BTS

Resegment bit set

Second part of B1 block not decoded by the BTS

Packet UL Ack/Nack (B1 not decoded)Resegment bit set

UL RLC data block first part B1, MCS1

UL RLC data block second part B1, MCS1

, PS1

, PS1

The picture above shows the case of a UL EGPRS TBF where one block is not decoded by the BTS and is then

re-transmitted by the MS with a lower MCS in the same MCS family. In this example, the second part of the

re-transmitted block is not correctly decoded by the BTS. As it is not possible to indicate separately in the

Packet Uplink Ack/Nack message whether the first part of the block or the second part has been decoded,

if one part is not received the MS will retransmit again the two parts of the block.






1 � 3 � 43


Type II ARQ Mechanism

� The MCS used to re-transmit a DL RLC data block depends on:

� The initial MCS used to send this RLC data block

� The resegmentation allowed or not

� The DL resegmentation is allowed If EN_FULL_IR_DL = disabled

� The UL resegmentation is allowed If EN_RESEGMENTATION_UL= enabled

� The possible memory shortage in the MS (case of a DL EGPRS TBF)

� MS OUT OF MEMORY = On, in the EGPRS packet DL Ack/Nack message

� The MCS commanded by the link adaptation algorithm (refer to session 2 EGPRS MCS Adaptation)

� As IR is optional in UL, the feature can be enabled/disabled using the Cell parameter EN_IR_UL

Modified B10

Modified B10

B10

Modified B10

The TRX manages the IR UL. Indeed, the TRX decodes the RLC/MAC header of all the UL RLC/MAC data

blocks received on each PDCH to know which TBF the RLC data block(s) pertain.

For each TBF, the maximum number of different RLC data blocks stored is equal to the window size which

depends on the maximum number of RTSs used in uplink (512 for 4 TS).

The TRX is able to store 4,000 RLC data blocks which have not been correctly decoded. If an RLC data block

is received with the same PS as an already received RLC data block belonging to the same TBF, only the last

instance is taken into account.

EN_FULL_IR_DL, parameter changed from BSS level in B9 to Cell Level in B10.

EN_RESEGMENTATION_UL, parameter changed from BSS level in B9 to Cell Level in B10.






1 � 3 � 44


Type II ARQ Mechanism [cont.]

� The following table is used to select the MCS if

� In DL, EN_FULL_IR DL = enabled and MS OUT OF MEMORY = off

� In UL, EN_RESEGMENTATION_UL = disabled

� In all the other cases, the table used for Type I ARQ is applied

Commanded MCS (given by the link adaptation algorithm)

MCS9 MCS8 MCS7 MCS6 MCS5 MCS4 MCS3 MCS2 MCS1


MCS8 MCS8 MCS8 MCS6 (pad.)

MCS6 (pad.)

MCS6 (pad.)

MCS6 (pad.)

MCS6 (pad.)

MCS6 (pad.)

MCS6 (pad.)







Initial MCS


WITHOUT

RESEGMENTATION

B10

“In all the other cases” means:

� In DL, EN_FULL_IR_DL=disabled or MS OUT OF MEMORY=on.

� In UL, EN_RESEGMENTATION_UL=disabled.

EN_FULL_IR_DL, parameter changed from BSS level in B9 to Cell Level in B10.

EN_RESEGMENTATION_UL, parameter changed from BSS level in B9 to Cell Level in B10.






1 � 3 � 45



� The PS used to re-transmit an RLC data block depends on:

� If the selected MCS has not changed then

� The PS is changed in a cyclic way: PS1, PS2, PS3, PS1, etc.

� Else, the PS to be used is indicated in the table below:

� PS1 is used in case of the first transmission of an RLC data block

Exercise

Previous MCS New MCS Previous PS New PS

PS1 or PS3 PS1 MCS9 MCS6

PS2 PS2

PS1 PS3 MCS6 MCS9

PS2 PS2

MCS7 MCS5 Any PS1

MCS5 MCS7 Any PS2

All other combinations Any PS1

If the selected MCS has not changed: if all the different punctured versions of the data block have been

sent, the procedure shall start over and PS1 shall be used, followed by PS2, then by PS3 (if available for the

considered MCS), so that the PS selection is cyclic.






1 � 3 � 46

4 UL Power Control






1 � 3 � 47

4 UL Power Control

Measurements

� The MS makes level measurements defined by the 05.08 GSM recommendation:

� in Packet Idle Mode:

� BCCH of the serving cell (paging blocks monitored by the MS);

� if MPDCH established, measurement on PCCCH = received signal on each paging block monitored, according to its DRX mode and paging group

� in Packet Transfer Mode:

� behavior defined by the parameter PC_MEAS_CHAN broadcast on the PBCCH (PSI1)� PBCCH of the serving cell (or BCCH if no MPDCH)

� on all the blocks of the PDCH carrying the PACCH

The MS uses DL level measurements to determine the power: open loop PC.






1 � 3 � 48

4 UL Power Control

Averaging

� Cn = a * (SSn + Pb) + (1-a) * Cn-1� a is the forgetting factor:

� Packet Idle Mode: 1 / min(n, max(5, T_AVG_W / TDRX))� TDRX = BS_PA_MFRMS (number of 51 multi-frame between 2 paging)

� Packet Transfer Mode: 1/ (6 * T_AVG_T) (BCCH)or 1/ (12 * T_AVG_T) (PDCH)

� SSn is the measurement at iteration n:

� average level of block n in Packet Idle Mode and Packet Transfer Mode (PDCH)

� level of the sample in Packet Transfer Mode (BCCH)

� Pb is a correcting factor relating to the power reduction value applied by the BTS on a PCCCH and/or PDCH, to be compared with the output power used on the BCCH

Use of a recursive filtering to obtain an average level.

Average levels calculated in Packet Idle Mode used in Packet Transfer Mode and vice versa: a proper

average level is available at the beginning of the transfer

The respective values of the T_AVG_T and T_AVG_W averaging windows are broadcast on PSI1.






1 � 3 � 49

4 UL Power Control

MS Power

� The MS uses the same power during a radio block (4 bursts)

� MS power = min(Γ0 - Γch - α * (C + 48), Pmax)

� Γ0 = 39 dBm in GSM 900, 36 dBm in GSM 1800� α and Γch are sent to the MS (α: SI 13, α and Γch: Packet UL and DL assignment) and are tuned in order to obtain a given behavior

� Pmax is the maximum transmitted power, and is equal to:

� GPRS_MS_TXPWR_MAX_CCH if there is a PBCCH

� MS_TXPWR_MAX_CCH otherwise

� C is the average DL level calculated by the MS

The MS power access on an RACH can be MS_TXPWR_MAX_CCH. In fact, the MS will use the first of the 2

values listened on the cell broadcast information.

The 05.08 GSM recommendation suggests to:

� use α = 1

� tune Γch in order to reach a given UL level (LEVUL) at the BTS side: Γch = Γ0 - 48 - LEVUL - PBTS (PBTS: BTS power)

� explanation:

� Pm = Γ0 - Γch - α * (C + 48)

� Pm = LEVUL - LEVDL + PBTS

� When you fix α=1, you get a specific value for Γch, which is not usable for any value of α.� Proceed by dichotomy to find the proper value of Γch

Another possibility:

� if path balance: PBTS - Pm = Sm - SBTS (S: sensitivity)

� therefore: LEVDL - LEVUL = Sm - SBTS

� and Pm = Γ0 - Γch - α * (LEVUL + Sm - SBTS + 48)

� example with G3 BTS: Pm = Γ0 - Γch - α * (LEVUL + 57)

� possibility of tuning:

� power reduction when the UL level is higher than U_RXLEV_UL_P

� MS power not lower than 13/4 dBm in GSM 900/1800






1 � 3 � 50

5 NC0 Cell Selection and Reselection






1 � 3 � 51


Introduction

� 2 kinds of selection – reselection are implemented in the Alcatel-LucentBSS:

� NC0

� The MS performs autonomous cell reselection

� All the algorithms (criteria computation, triggering, target cell choice) are implemented in the MS

� No measurement reporting

� NC2

� The network (MFS) controls the cell reselection

� All the algorithms (criteria computation, triggering, target cell choice) are implemented in the MFS

� The MS sends periodically measurement reports

� The main important parameters involved in the cell selection andreselection are broadcast in PSI3 & PSI3bis (if PBCCH) or in SI3 (if BCCH).

� The GPRS neighboring cells list is identical to the GSM one

Further details concerning Cell selection and Cell reselection in case of PBCCH in the appendix.






1 � 3 � 52


Principles

� Procedures defined in the 05.08 GSM recommendation

� Cell selection:

� made using the C1 criterion as for GSM

� Cell reselection:

� made using the C1 and C2 criteria as for GSM in the serving cell

In GSM

C1 = A - Max (0,B) with:

� A = RLA_C - RXLEV_ACCESS_MIN

� B = MS_TXPWR_MAX_CCH - MS_TXPWR_MAX + POWER_OFFSET(1800)

C2 = C1 + CELL_RESELECT_OFFSET - TEMPORARY_OFFSET(T) when Penalty_time<31

C2 = C1 - CELL_RESELECT_OFFSET when Penalty_Time=31

In GPRS ready and standby states, cell reselection is performed by the MS except for a class A MS while in

dedicated mode of a circuit-switched connection, in which case the cell is determined by the network

according to the handover procedures.

For a class B MS which can combine GSM and GPRS states, C1 criterion is used when the MS simultaneously

attached to both the network and the MS is in Packet Idle Mode (refer to GSM 05.08).






1 � 3 � 53


Selection

� Criteria computation:

� Without PBCCH

� C1 = (RLA_C – RXLEV_ACCESS_MIN) – max (0, MS_TXPWR_MAX_CCCH – P)

� RLA_C: average DL level received

� Cell choice: the best cell is the cell with the highest C1

C1 is the same as in GSM except that:

� A = RLA_P – GPRS_RXLEV_ACCESS_MIN: “listening capacity of MS in the cell”

� B = GPRS_MS_TXPWR_MAX_CCH – P: “talking capacity of MS in the cell”

� C1 shall be positive and as high as possible

In Packet Idle Mode, the MS shall make one measurement for each BCCH carrier monitored every 4 seconds,

as well as more than one sample per second for each BCCH carrier.

A list of 6 strongest cells shall be kept updated at a rate of at least one update per running average period.

In Packet Transfer Mode, the MS shall monitor a list of 6 strongest non-serving cell BCCH carriers. It shall

attempt to check the BSIC for each of these 6 strongest cells at least once every 10 seconds.






1 � 3 � 54


Reselection Criteria Computation Without PBCCH

� If CELL_RESELECT_PARAM_IND=not present then C2=C1 else:

� C2 = C1 + CELL_RESELECT_OFFSET - TEMPORARY_OFFSET(T)(if PENALTY_TIME <> 31)

� if T > PENALTY_TIME, TEMPORARY_OFFSET(T)=0

� used to avoid locating on “transient cell”

� CELL_RESELECT_OFFSET used to favor a cell among others (e.g., micro-cell vs. umbrella, once T > PENALTY_TIME)

� C2 = C1 - CELL_RESELECT_OFFSET(if PENALTY_TIME = 31)

� CELL_RESELECT_OFFSET used to handicap some cells among others

The same algorithm is used in case the MS is in GSM Idle Mode.






1 � 3 � 55


Target Cell Choice Without PBCCH

� The MS triggers a cell reselection if:

� C1(serving) <0

and/or

� In Standby Mode

� C2(neighbor) > C2(serving) if cells belong to a same RA

� C2(neighbor) > C2(serving)+CELL_RESELECT_HYSTERESIS if cells from different RAs

� In Ready Mode

� C2(neighbor) > C2(serving)+CELL_RESELECT_HYSTERESIS even if cells belong to a same RA

� Cell choice: the best cell is the cell with the highest C2

The normal procedures apply in case of Cell reselection for a DTM capable MS in PTM.






1 � 3 � 56


Reselection During a UL TBF

� UL TBF:

� MFS: after a cell reselection, the MFS receives no more data in the UL blocks allocated to the MS => TBF release

� MS: in the new cell, after the SI messages acquisition, a new UL TBF is established

� SGSN: the SGSN is informed of the cell change when receiving an LLC unit from the MS in the new cell. Then the SGSN notifies the BSS about the cell change (FLUSH PDU)

After a TBF release, it is up to the originator to reinitiate the transfer: the MS in the UL, the SGSN in the

DL.






1 � 3 � 57


Reselection During a UL TBF [cont.]

SGSNCell Reselection

?

FLUSH LLMFS







1 � 3 � 58


Reselection During a DL TBF

� DL TBF:

� MFS: after a cell reselection, the MFS receives no more acknowledgements from the MS => abnormal TBF release

� MS: in the new cell, after the SI messages acquisition, a UL TBF is established to send a cell update to the SGSN (MS in Ready state)

� SGSN: when the SGSN is informed of a cell change it sends a message to the MFS to discard LLC units stored for the MS in the old cell (FLUSH PDU)

� The SGSN resumes the DL transfer by sending a DL LLC unit => DL TBF establishment in the new cell

After a TBF release, it is up to the originator to reinitiate the transfer: the MS in the UL, the SGSN in the

DL.






1 � 3 � 59


Reselection During a DL Transfer: Example

Cell update (new BVCI)SGSN aware of the cell reselection

MFS « aware » of a radio problem

MFS aware of the cell reselection

Exercise

MFS: after a cell reselection, the MFS receives no more acknowledgement from the MS ⇒ TBF release.

MS: in the new cell, after the SI messages acquisition, a UL TBF is established to send a cell update to the

SGSN (MS in Ready state).

SGSN: when the SGSN is informed of a cell change, it sends a message to the MFS to discard LLC units stored

for the MS in the old cell. The SGSN resumes the DL transfer by sending a DL LLC unit ⇒ DL TBF estab in the

new cell.

The MFS is always aware of a successful cell change afterwards, upon reception of the flush LL message

from the SGSN.

If the cell change is unsuccessful, the TBF release is counted as abnormal.

DL_UDT = DL user data

RAD_STATUS = radio status message sent by the MFS to the SGSN (BSSGP signaling).

FLUSH_LL = BSSGP message sent by the SGSN to the MFS to notify a successful change of cell by the MS.






1 � 3 � 60


NACC and (P)SI Status

� 2 features are available to reduce the duration of the reselection

� NACC: Network Assisted Cell Change

� If EN_NACC = enabled then� before the cell reselection,

� in the serving cell,

� the network sends to the MS a part of the SI messages of the new cell

� (P)SI Status: (Packet) System Info Status

� If EN_PSI_STATUS = enabled then

� after the transfer resumption,

� in the target cell,

� the MS can ask the network to send it:

� the remaining PSI messages if PBCCH is present

� the remaining SI messages otherwise

B10

The NACC procedure is a new feature standardized in Release 4, mandatory for Release 4 onwards mobile

stations supporting GERAN Feature Package 1.

The Packet PSI Status procedure is a feature standardized from Release 97 onwards, optional for Release

97, Release 98 and Release 99 MS, and mandatory for Release 4 onwards MS supporting GERAN Feature

Package 1.

The Packet SI Status procedure is a new feature standardized in Release 4, mandatory for Release 4

onwards mobile stations supporting GERAN Feature Package 1.

NACC and (P)SI Status features are supported only if:

� The MS is neither in dedicated mode nor Dual Transfer Mode

� The MS is in NCO or NC1 mode

� The MS is in Packet Transfer Mode.






1 � 3 � 61


NACC and (P)SI Status [cont.]

� NACC

� The MS informs the system that it wants to select a neighboring cell

� The BSS transmits the partial SI of the neighboring cell

MSMFSCell A

UL or DL TBF

Packet Cell Change Notification

Retrieval of SI

instances

Packet Neighbor Cell Data (SI1)



Packet Cell Change Continue

MFSCell B

When the MS detects a need of cell reselection in transfer mode, it sends a "Packet Cell Change

Notification" message to the MFS (on PACCH):

� If the MFS knows the (P)SI of the target cell:

� If there is no PBCCH in the target cell, it sends SI1, SI3, SI13 in (several) "Packet Neighbor Cell Data"

messages, followed by a "Packet Cell Change Continue" message.

� If there is a PBCCH in the target cell, it sends PSI1, PSI2, PSI14 in (several) "Packet Neighbor Cell

Data" messages, followed by a "Packet Cell Change Continue" message.

� If the MFS does not know the (P)SI of the target cell, it sends only a "Packet Cell Change Continue"

message.

If no PBCCH is present, the BSC sends SI1, SI3, SI13 messages to the MFS in a "System Information Update"

message.






1 � 3 � 62


NACC and (P)SI Status [cont.]

� SI Status

� The MS has resumed the data transfer in the neighboring cell

� Then, it asks the system to retrieve the missing Sys-info

� This mechanism is applied in both NC0 and NC2

MSMFSCell A

MFSCell B

Packet SI Status (SI2, SI2bis, SI2ter msg type missing)

Packet Serving Cell Data (SI2)

Packet Serving Cell Data (SI2bis)

Packet Serving Cell Data (SI2ter)

UL or DL TBF

UL or DL TBF

When accessing a new cell, the MS must get SI13, SI3, SI1, or PSI1 and PSI2 (if not already known through

NACC).

If the Packet (P)SI Status is offered in the cell, it can start PTM and send:

� a "Packet SI Status" message (when there is no PBCCH), with the list of missing SI messages.

� a "Packet PSI Status" message (when there is a PBCCH), with the list of missing PSI messages.

The MFS shall then send:

� the SI instances required by the MS (the MS is the only addressee) in one or several "Packet Serving Cell

Data" messages in case there is no PBCCH in the target cell.

� the PSI instances required by the MS (all MSs listening to this PDCH will get the information) directly on a

PACCH in case a PBCCH is present in the target cell.

The MS can request updated (P)SI instances whenever it wants, provided it is in PTM.






1 � 3 � 63

6 NC2 Cell Reselection






1 � 3 � 64


Aim

� Impact of reselection on data transfer

� NC2 aims at reducing the number of cell reselections triggered whenthe MS is in Packet Transfer Mode

� The lower the number of cell reselections, the better the end-user QoS

Data

Transfer

: TBF establishment

: TBF release due to cell reselection

: Reselection

1 2 3

1

23

4

4

Each time the MS performs a cell reselection, the data transfer is interrupted and a retransmission of some

LLC PDUs may be required:

� The on-going TBF is released in the old cell.

� The MS performs the PSI or SI acquisition in the new cell.

� Then, the MS establishes a new UL TBF in this cell to send a Cell Update message to the SGSN.

� The MFS deletes or reroutes towards the new cell the LLC PDUs stored in the old cell.

� if they are deleted, a retransmission is needed.

� Finally, the data transfer is re-started (after a DL TBF establishment, in case of DL transfer).

All these steps degrade the data throughput or the page access time perceived by the end user.






1 � 3 � 65

�The NC2 process for PS is similar to the HO process for CS

� HO

� NC2


Functional Entities

RadioLink Measurements

ActiveChannelPre-processing

HO DetectionHO CandidateCell Evaluation

HO management

HO Preparation

MS - BTS BSC

MS - BTS MFS

NC cell Reselectionactivation

NC measurementReporting andprocessing

NC cellReselection Detection

NC cell ReselectionEvaluation

NC cell Reselectionmanagement

NC cell reselection Preparation

The RRM layer is in charge of the measurement processing. It is also in charge of the selection of the target

cell, as it is the layer having the knowledge of the network topology and parameters. The RRM layer is

actually in charge of managing the overall NC cell reselection procedure.

The RLC layer is in charge of forwarding the packet measurements to the RRM layer. Finally, the RLC layer

is in charge of the RXLEV and RXQUAL measurements processing (per TBF) and of the corresponding NC cell

reselection detection.






1 � 3 � 66


NC Cell Reselection Activation / De-activation

� Activation

� NETWORK_CONTROL_ORDER has 3 possible values:

� NC0 mode of operation for all MSs

� NC2 mode of operation for R99 onwards MSs

� NC2 mode of operation for all MSs

� NC2 cell reselection can be used only when the MS is in READY state, otherwise NC0 is used

� De-activation

� NC2_DEACTIVATION_MODE has 2 possible values:

� NC2 deactivation at the end of the packet transfer

� NC2 deactivation at T_READY (GMM Ready timer) expiry

� In case of Dual Transfer Mode, Cell Reselection will be ignored by the MS

B10

NETWORK_CONTROL_ORDER is a cell parameter tunable at OMC-R level.

The R97 and R98 MSs are differentiated from the other MSs. Indeed, all the MSs shall support the NC2 mode,

however since no network manufacturer has implemented the NC2 mode, the R97 and R98 MSs may not

have been sufficiently tested and therefore there is a risk of interoperability with these MSs.

The “Packet Measurement Order” message is used to activate and de-activate the NC2 mode of operation

for a given MS.

� Activation

� The “Packet Measurement Order (NC2)” message is sent when:

� establishing the first Downlink TBF of the Packet Transfer Mode or when re-establishing the DL

TBF while T3192 is running and there is not any on-going UL TBF.

� no measurement report has already been received for that MS during its on-going packet

transfer(s) (UL and/or DL).

� the MS has not been forced to operate in NC2 mode by a Packet Cell Change Order message

(during an intra-RA cell reselection).

� De-activation

� The “Packet Measurement Order (RESET)” message is sent at the end of the data transfer, in case of

NC2_DEACTIVATION_MODE = “NC2 deactivation at the end of the packet transfer”.

� When the MS goes back to the STANDBY state, in case of NC2_DEACTIVATION_MODE = “NC2

deactivation at GMM Ready timer expiry”.






1 � 3 � 67


NC Cell Reselection Activation / De-activation [cont.]

� Activation

� There is NC2 activation only at the beginning of a DL packet transfer

T_WAIT_PMR

NC_REPORTING_PERIOD_T

MS BSS

Packet Measurement Order [NC2] / PACCH (4)

Packet Measurement Report / PACCH (5)

On-going UL TBF (1)


Packet Downlink Assignment / PACCH (2)

Packet Control Acknowledgement (3)

DL LLC PDU

READY MS in NC0

READY MS in NC2

(1) It is assumed the MS has just initiated the establishment of a UL TBF, but no DL TBF is on-going. If there

is no UL TBF on-going, the NC2 activation is also done on receipt of the acknowledgement of the DL TBF

establishment performed on the (P)CCCH.

(2) The receipt of a DL LLC PDU triggers the establishment of the DL TBF on the PACCH of the UL TBF.

(3) The MS acknowledges the Packet Downlink Assignment message by a Packet Control Acknowledgement

message.

(4) Upon receipt of the Packet Control Acknowledgement message, the BSS sends to the MS on the PACCH of

the on-going DL TBF a Packet Measurement Order message forcing the MS to operate in NC2 mode and

starts the timer T_WAIT_PMR. The Packet Measurement Order message is sent without a polling

indication. The Packet Measurement Order message provides the MS with the following NC measurement

parameters NETWORK_CONTROL_ORDER, NC_NON_DRX_PERIOD, NC_REPORTING_PERIOD_I,

NC_REPORTING_PERIOD_T. The timer T_WAIT_PMR monitors the reception of the Packet Measurement

Report messages.

(5)-(6) On the allocated UL RLC blocks, the MS sends a Packet Measurement Report message every

NC_REPORTING_PERIOD_T seconds. The timer T_WAIT_PMR is stopped at the receipt of the first Packet

Measurement Report message. At T_WAIT_PMR expiry, if MAX_RETRANS_SIG ≠ 0 a new Packet

Measurement Order is sent to the MS and the timer T_WAIT_PMR is started. Such mechanism is applied

MAX_RETRANS_SIG attempts.






1 � 3 � 68


NC Cell Reselection Activation / De-activation [cont.]

� De-activation at the end of the packet transfer

� There is NC2 de-activation at the normal end of a DL Packet Transfer Mode

Packet Downlink Ack/Nack / PACCH (2)

Last DL RLC data block with a polling indication (1)

Packet Measurement Order [Reset] / PACCH (3)

MS BSS

On-going DL TBF

READY MS in NC2

STANDBY MS in NC0

(1) It is assumed that a DL TBF is on-going. The BSS sends to the MS the last useful data block (case of

normal TBF release) or the RLC block containing the last dummy UI command (case of a delayed TBF

release).

(2) The MS acknowledges the received block by sending the final Packet Downlink Ack/Nack message to the

BSS.

� Note: When an RLC mode change is detected, the BSS waits for the final Packet Downlink Ack/Nack

message before re-establishing the DL TBF with the new RLC mode. As the fast DL TBF establishment

occurs on receipt of the final Packet Downlink Ack/Nack message, a Packet Measurement Order

[Reset] message would be immediately followed by a Packet Measurement Order [NC2] message. In

order to avoid that useless message exchange, the NC2 mode is not deactivated in this case.

(3) If there is no on-going UL TBF, upon receipt of the final Packet Downlink Ack/Nack message, the BSS

sends to the MS on the PACCH of the DL TBF a Packet Measurement Order message with a Reset command.

The Reset command forces the MS to realign its behavior on the parameters broadcast in the (packet)

system information messages on (P)BCCH (i.e., return to NC0). To ensure a high probability of correct

reception by the MS, the RRM orders MAC to repeat the Packet Measurement Order (Reset) message several

times. The number of repetitions is defined by the O&M parameter N_SIG_REPEAT. In case the Packet

Measurement Order (Reset) message is not received by the MS although repeated, the MS will remain in NC2

mode for the whole duration of the Ready timer, while the operator requested the network to deactivate

NC2 at the end of the Packet Transfer Mode. Because repetitions should ensure that this happens very

scarcely, the Alcatel BSS will not handle those rare events. Then, if a Packet Measurement Report is

received in Packet Idle Mode, it will be discarded.

There is no NC2 deactivation at the end (normal or abnormal) of the UL Packet Transfer Mode, and at the

abnormal end of the DL Packet Transfer Mode.






1 � 3 � 69


NC Measurement Reporting and Processing

� DL RXLEV

� The MS sends a “Packet Measurement Report” message containing the RXLEV measured on the serving cell and the 6 best neighbor cells

� The “Packet Measurement Report” message is sent every:

� NC_REPORTING_PERIOD_T ms, in case of Packet Transfer Mode

� NC_REPORTING_PERIOD_I ms, in case of Packet Idle Mode

On the one hand the NC_REPORTING_PERIOD_T parameter is defined by O&M on a per cell basis. On the

other hand, the NC_REPORTING_PERIOD_I is set on a BSS cell basis without OMC-R access (default value =

max value = 61.44 s i.e., 256 52-multiframes).

These parameters are provided to the MS either in a Packet Measurement Order message or in a Packet Cell

Change Order message. They are never broadcast on (packet) system information messages.

Packet Measurement Report message contents

� TLLI of the MS

� NC_MODE (Set to NC2)

� RXLEV_SERVING_CELL (RXLEV measured on the serving cell)

� NUMBER_OF_NC_MEASUREMENTS (Number of measurements reported for the neighboring cells)

� FREQUENCY_N (Refer to the ARFCN or ARFCN and BSIC of a neighboring cell)

� BSIC (BSIC of the indexed neighboring cells)

� RXLEV_N (RXLEV of the indexed neighboring cells)






1 � 3 � 70


NC Measurement Reporting and Processing [cont.]

� DL RXLEV averaging for serving cell and neighbor cell

� AV_DL_RXLEV_NC2p = (1-f)* AV_DL_RXLEV_NC2p-1 + f*RXLEV_Np

� RXLEV_Np is the RXLEV sample reported by the MS in the incoming Packet Measurement Report message

� f is the averaging forgetting factor and is derived from the parameter NC_RXLEV_FORGETTING_FACTOR

� p is the iteration index

NC_RXLEV_FORGETTING_FACTOR = 0.13 ((Alcatel recommended value) and it can be set at OMC-R level).

If the neighboring cell n was not reported in the precedent Packet Measurement Report but just in the last

one:

� AV_DL_RXLEV_NC2p(n) = (1-f) *AV_DL_RXLEV_NC2p-1(n) + (1-(1-f) ) *RXLEV_Np(n)

� = (p –1) – i

� The index i represents the last time an NC measurement for that neighboring cell has been reported

in a Packet Measurement Report message.






1 � 3 � 71



� RXQUAL for a GPRS TBF

� In the DL, the MS sends a “Packet DL Ack/Nack” message containing the RXQUAL measured on the serving cell every T_DL_GPRS_MeasReport ms

� In the UL, the MFS assesses the RXQUAL for each RLC block received

� MeanBEP for an EGPRS TBF

� In the DL, the MS sends an “EGPRS Packet DL Ack/Nack” message containing the MeanBEP measured on the serving cell every T_DL_EGPRS_MeasReportms

� In the UL, the MFS assesses the MeanBEP for each RLC block received

T_DL_GPRS_MeasReport and T_DL_EGPRS_MeasReport are defined by O&M on a per cell basis.






1 � 3 � 72



� DL RXQUAL averaging

�

�

� forgetting factor:�

� is the time between 2 DL RXQUAL samples

� TNC2 set to the parameter NC_RXQUAL_AVG_PERIOD

� UL RXQUAL and MeanBEP (UL & DL) are averaged using the same formula and the same parameter

DL_RXQUALU

12AL_NCAV_DL_RXQU*

U

112AL_NCAV_DL_RXQU

p

1p

p

p +

−= −

1UαU 1p-∆tp

2NCp +=

( ) 2NC/T12NC β-1α =

0.9β=

pt∆

U0 = 0, consequently U1 = 1.

In CS4, if RxQual = 7 is reported by the MS, then this measurement is tagged as invalid and so, not taken

into account in the averaging.






1 � 3 � 73


NC Cell Reselection Detection

� Trigger conditions

Detection of a better neighboring cell

Too low downlink received signal level

Too bad downlink radio quality

Too bad uplink radio quality

Name of the cause

Cause 12PT2Lowest

Cause 5PT1…

Cause 4PT3…

Cause 2PT4Highest

Similar HO cause

Cause reference

Priority






1 � 3 � 74


NC Cell Reselection Detection [cont.]

� Cause PT3 for a GPRS TBF

� AV_DL_RXQUAL_NC2 > NC_DL_RXQUAL_THR

� If NC_DL_RXQUAL_THR = 7 (Never), the cause PT3 is disabled

� Cause PT3 for an EGPRS TBF

� AV_DL_MeanBEP_NC2 < NC_DL_MeanBEP_THR_xxSK_yyyyyy

� xxSK: GMSK or 8-PSK

� yyyyy:� type1: type 1 ARQ (no Incremental Redundancy)

� type2: type 2 ARQ (with Incremental Redundancy)

� If NC_DL_MeanBEP_THR_xxSK_yyyyyy = 0, the cause PT3 is disabled

Cause PT3 is checked only for the serving cell each time an (EGPRS) Packet Downlink Ack/Nack message is

received provided that the DL TBF is not in delayed release state and provided that the

T_NC_RXQUAL_VALID seconds have elapsed since the receipt of the first Packet Downlink Ack/Nack message

of the DL TBF.

T_NC_RXQUAL_VALID aims at not triggering false alarms at the beginning of the TBF and not triggering an

NC cell reselection for a very short TBF.

In CS4, if RxQual = 7 is reported by the MS, then this measurement is tagged as invalid and so, it will be not

taken into account in the averaging and will not disturb the PT3 triggering.






1 � 3 � 75



� Only one NC_DL_MeanBEP threshold is applied for cause PT3 during an EGPRS TBF, and derived from NC_DL_RXQUAL_THR

� NC_DL_MeanBEP_THR_xxSK_yyyyyy = NC_DL_MeanBEP

� NC_DL_MeanBEP = (23-3* NC_DL_RXQUAL_THR)

� If NC_DL_RXQUAL_THR = 7 then NC_DL_MeanBEP = 0

� i.e., the cause PT3 is disabled

� Same behavior for cause PT4 in the UL

� NC_UL_MeanBEP = (23-3* NC_UL_RXQUAL_THR)






1 � 3 � 76



� Cause PT4 for a GPRS TBF

� AV_UL_RXQUAL_NC2 > NC_UL_RXQUAL_THR

� If NC_UL_RXQUAL_THR = 7 (Never), the cause PT4 is disabled

� Cause PT4 for an EGPRS TBF

� AV_UL_MeanBEP_NC2 < NC_UL_MeanBEP_THR_xxSK_yyyyyy

� xxSK: GMSK or 8-PSK

� yyyyy:� type1: type 1 ARQ (no Incremental Redundancy)

� type2: type 2 ARQ (with Incremental Redundancy)

� NC_UL_MeanBEP_THR_xxSK_yyyyyy = NC_UL_MeanBEP

� NC_UL_MeanBEP = (23-3* NC_UL_RXQUAL_THR)

� If NC_UL_RXQUAL_THR = 7 then NC_UL_MeanBEP = 0

� i.e., the cause PT4 is disabled

Cause PT4 is checked only for the serving cell whenever one UL RLC data block is correctly received for the

on-going UL TBF provided that T_NC_RXQUAL_VALID seconds have elapsed since the computation of the

first UL samples of the UL TBF.

T_NC_RXQUAL_VALID aims at not triggering false alarms at the beginning of the TBF and not triggering an

NC cell reselection for a very short TBF.






1 � 3 � 77



� Cause PT1

� AV_DL_RXLEV_NC2 < NC_DL_RXLEV_THR + Max(BNC2,0)

� without PBCCH, BNC2 = MS_TXPWR_MAX_CCH – P

� with PBCCH BNC2 = GPRS_MS_TXPWR_MAX_CCH – P

� If NC_DL_RXLEV_THR = -110dBm (Never), the cause PT1 is disabled

� Cause PT2

� C2NC2(n) - C2NC2(s) > NC_RESELECT_HYSTERESIS(s,n)

� If PBCCH is present in the serving cell, C2NC2 is replaced with C32NC2AND

� AV_DL_RXLEV_NC2 <= NC_DL_RXLEV_LIMIT_THR

� If NC_RESELECT_HYSTERESIS(s,n) = 128dB (Never), the cause PT2 from s to n is disabled

The cause PT1 is equivalent to check the condition C1NC2 < 0 assuming that the (GPRS_)RXLEV_ACCESS_MIN

threshold is replaced with NC_DL_RXLEV_THR threshold.

Cause PT2 is checked among the neighboring cells n upon receipt of a Packet Measurement Report message.

It is triggered if the value C2NC2 or C32NC2 of one neighboring cell n exceeds the value C2NC2 or C32NC2

of the serving cell s by at least the O&M hysteresis NC_RESELECT_HYSTERESIS(s,n) defined per cell

adjacency link (respectively whether or not there is a PBCCH in the serving cell).






1 � 3 � 78


NC Cell Reselection Evaluation: Functional Entities

� Cell filtering process

� This process builds a Filtering Cell List depending on:

� The content of the Rejected Cell list

� EN_OUTGOING_GPRS_REDIR

� C1NC2(n)

� GPRS operational state of the neighbor cells

� Cell ranking process

� This process builds a Filtering Cell List depending on:

� C31NC2� Load situation

� C1NC2 & C2NC2

Cell filtering process

Cell rankingprocess

- cause reference

- neighboring cell that checked the cause

Filtering cell list

Reference of the target cell

The Cell Filtering process is computed on receipt of an NC Cell Reselection Evaluation Request message.

Before processing the cell ranking, all the candidate neighboring cells are gathered in the Raw Cell List.

The serving cell is always included in this latter list.

The contents of the Raw Cell List depend on the cause reference that triggered the NC cell reselection

evaluation:

� If Cause PT1, or PT3, or PT4 is checked, then the Raw Cell List regroups the cells reported in the Packet

Measurement Report message provided that the cells are configured by O&M. The serving cell is always

included in the list.

� If only Cause PT2 is checked, then the Raw Cell List regroups all the neighboring cells that verify Cause

PT2 and that are reported in the Packet Measurement Report message. The serving cell is always included

in the list.

The Raw Cell List is then filtered according to the contents of a Rejected Cell List, according to the flag

EN_OUTGOING_GPRS_REDIR of the serving cell and of the neighboring cells, and according to the C1NC2

parameter of the neighboring cells. The output cell list is here called the Filtering Cell List.






1 � 3 � 79


NC Cell Reselection Evaluation: Criteria Computation

� Path loss criterion parameter C1NC2

� C1NC2(n) = AV_RXLEV_NC2(n) – RXLEV_ACCESS_MIN(n) - max(MS_TXPWR_MAX_CCH(n)

– P(n), 0)

� Cell ranking criterion parameter C2NC2� If PENALTY_TIME <> 31:

� C2NC2(n) = C1NC2(n) + CELL_RESELECT_OFFSET(n)

� Else� C2NC2(n) = C1NC2(n) - CELL_RESELECT_OFFSET(n)

� If the T_NC_PING_PONG timer is running, the anti-ping-pong offset NC_PING_PONG_OFFSET is subtracted from the C2NC2 of the neighboring cells

The cell n denotes either the serving cell or a neighboring cell.

In the above equations, the following notations mean:

� AV_RXLEV_NC2(n) is the average received signal level measured by the MS on the BCCH of the cell n.

� RXLEV_ACCESS_MIN(n) is the minimum received signal level required to perform an access to the cell n.

� MS_TXPWR_MAX_CCH(n) is the maximum transmit power of the MS when accessing the cell n.

P(n) is the maximum output RF power of the MS in the BCCH frequency band of the cell n. P(n) gives the MS

Radio Access Capability Information Element provided in the Packet Resource Request message or in the DL

LLC PDU. In the NC cell reselection procedure, the parameter P(n) shall always refer to the RF power

capability of the GMSK modulation.

Note that all values are expressed in dBm.

The cell ranking criterion parameter C2NC2 is used to order the candidate cells on a radio criterion. This

criterion applies only in serving cells where there is no PBCCH established.

� CELL_RESELECT_OFFSET(n) is a positive offset which favors or disfavors the cell n.

� PENALTY_TIME(n) indicates whether the cell reselection offset shall be positive or negative.






1 � 3 � 80


NC Cell Reselection Evaluation: Cell Filtering Process

� A candidate neighboring cell n is filtered out when:

� A previous NC cell reselection failed toward this cell and T_NC_REJ_CELL[n] is running

� The timer T_NC_REJ_CELL[n] is started or restarted each time the new cell n is inserted in the Rejected Cell List

� At expiry of T_NC_REJ_CELL[n], the rejected cell is removed from the Rejected Cell List

� One T_NC_REJ_CELL by cell (and not by MS)

� EN_OUTGOING_GPRS_REDIR(n) = enabled

� C1NC2(n) < 0

� GPRS not activated

� RA_COLOUR = -1

The notation T_NC_REJ_CELL[n] refers to the timer associated to the cell n.

For the purpose of filtering cells towards which a previous NC cell reselection failed, the RRM manages a

Rejected Cell List. Each neighboring cell n of the list is guarded by the timer T_NC_REJ_CELL[n]. While

T_NC_REJ_CELL[n] is running, the neighboring cell n shall not be selected for any NC cell reselection.

The Rejected Cell List shall be able to contain up to 32 neighboring cells. If the Rejected cell List is full,

the oldest cell is discarded and the new one is stored.

In addition, if the flag EN_OUT_GOING_GPRS_REDIR(s) of the serving cell is set to “Enabled”, the serving

cell is removed from the Raw Cell List. Indeed, in such cells, the neighboring cells do not need to be better

than the serving cell as a GPRS redirection is not triggered due to a bad radio link, but is triggered in order

to redirect the MS towards a more appropriate neighboring cell to carry PS traffic.






1 � 3 � 81


NC Cell Reselection Evaluation: Cell Ranking Process

� Without PBCCH

� The best candidate cell is the cell for which the following ordered criteria are fulfilled:

1. C31NC2 >= 0

2. Load situation = low

3. Best C2NC2

� If all the candidate cells have their criterion C31NC2 < 0, then

� The best candidate cell is the cell which has the best C2NC2

Once the best candidate cell has been found, the MFS checks whether or not the best cell is the serving

cell:

� If the best cell is not the serving one, the NC cell reselection evaluation function sends an NC Cell

Reselection Alarm Indication message to the NC cell reselection execution function in order to trigger the

execution of the NC cell reselection.

� If the best cell is the serving cell, the NC cell reselection procedure is stopped.






1 � 3 � 82


NC Cell Reselection Evaluation: Cell Ranking Process [cont.]

� Serving cell:� C31NC2(n) = AV_RXLEV_NC2(n) – HCS_THR(n)

� HCS_THR(n), signal threshold for applying the load cell situation criterion.

� C31NC2 is used in serving cell, to differentiate the low loaded target cells from the high loaded target cells.

Once the best candidate cell has been found, the MFS checks whether or not the best cell is the serving

cell:

� If the best cell is not the serving one, the NC cell reselection evaluation function sends an NC Cell

Reselection Alarm Indication message to the NC cell reselection execution function in order to trigger the

execution of the NC cell reselection.

� If the best cell is the serving cell, the NC cell reselection procedure is stopped.






1 � 3 � 83


NC Cell Reselection Evaluation: Load Evaluation

� Every 5 seconds, the MFS computes for each cell

� Where:

� UL_PS_used_Bandwidth = Nb of UL TBFs / MAX_UL_TBF_SPDCH

� DL_PS_used_Bandwidth = Nb of DL TBFs / MAX_DL_TBF_SPDCH

� Total_PS_Bandwidth = MAX_PDCH – NB_TS_MPDCH

� CS_Used_Bandwidth = Total_PS_Bandwidth – N_PDCH_ALLOCATED

� N_PDCH_ALLOCATED = Number of SPDCHs currently allocated to the MFS

[ ]100

andwidthTotal_PS_B

ndwidthCS_Used_Ba_BandwidthDL_PS_Used;_BandwidthUL_PS_UsedMAX%)(in NC2_Load ×+=

∑=

ALLOCATEDPDCHN

i

__

1

∑=

ALLOCATEDPDCHN

i

__

1

UL_PS_Used_Bandwidth is the bandwidth used by PS traffic in the UL direction.

DL_PS_Used_Bandwidth is the bandwidth used by PS traffic in the DL direction.

CS_Used_Bandwidth is the bandwidth used by CS traffic.






1 � 3 � 84


NC Cell Reselection Evaluation: Load Evaluation [cont.]

� NC2_Load is averaged using the sliding window NC2_LOAD_EV_PERIOD(=3)

� This load average is then compared to the threshold THR_NC2_LOAD_RANKING as followed:

� If Load average <= THR_NC2_LOAD_RANKING then

� Load situation = low

� Else (Load average > THR_NC2_LOAD_RANKING)

� Load situation = high

� Case of the external cells (inter BSC)

� If THR_NC2_LOAD_RANKING < 100% then Load situation = low

� Else (THR_NC2_LOAD_RANKING = 100%) then Load situation = high

Exercise

The MFS shares the NC2 load situation information among the different cells of the BSS (or at least between

the cells having a cell reselection link with the serving cell).

In case of an external cell, the load evaluation is different since the load situation of such cells is unknown

in the serving cell.






1 � 3 � 85


NC Cell Reselection Execution with NACC

T_Wait_Flush

T_Ack_Wait

On going UL TBF (8)

On going UL or DL TBF (1)

MS BSSServing cell

Packet Cell Change Order / PACCH (3)


BSSTarget cell

SGSN

Packet Control Acknowledgement / PACCH (5)(4)

Packet Channel Request / PRACH (6)

Packet Uplink Assignment / PCCCH (7)

UL LLC PDU [TLLI] (9)

FLUSH-LL PDU [TLLI, old BVCI] (10)

FLUSH-LL-ACK PDU [TLLI, “deleted”] (11)

Packet Neighbour Cell Data (SI3) / PDCH

Packet Neighbour Cell Data (S1) / PDCH

Packet Neighbour Cell Data (SI13) / PDCH

(1) A UL or DL TBF is assumed to be on-going.

(2) The MS sends a Packet Measurement Report message on one of the allocated UL blocks on the PACCH.

(3) Upon receipt of the Packet Measurement Report message, the BSS detects that an NC cell reselection

must be triggered and therefore orders the MS to reselect a new cell by sending a Packet Cell Change

Order message on the PACCH of the DL or UL TBF. If both a UL and a DL TBF are on-going, the message is

preferentially addressed by a DL TFI. The Packet Cell Change Order message is sent in acknowledged

mode and contains the ARFCN and the BSIC of the target cell plus the NC parameters of the target cell

(if the MS can operate in NC2 mode in the target cell). When sending the Packet Cell Change Order

message, the BSS starts the timer T_ACK_WAIT to monitor the receipt of the Packet Control

Acknowledgement message.

(4)-(5) Upon receipt of the Packet Cell Change Order message, the MS aborts its on-going TBF in the serving

cell and sends the Packet Control Acknowledgement message. Once the MS has sent the Packet Control

Acknowledgement message, the MS switches to the new cell. Upon receipt of the Packet Control

Acknowledgement message, the BSS starts the timer T_WAIT_FLUSH (which monitors the reception of

the FLUSH-LL PDU) and requests the release of the on-going TBF(s) (if any). The radio resources are

immediately released, i.e., without freezing them. In case the radio resources are already frozen, the

freezing timer is stopped and the radio resources are immediately released.

(6)-(8) After acquiring the full PSI cycle and successfully decoding the PSI1 and PSI2 messages of the target

cell, the MS initiates a UL TBF establishment in the new cell.

(9) The target BSS sends to the SGSN the first UL LLC PDU containing the TLLI of the MS.

(10) By comparing the previous couple and the new one (BVCI; NSEI), the SGSN detects that the MS has

changed of cell and sends a FLUSH-LL PDU to the old cell.

(11) Upon receipt of the FLUSH-LL PDU, the BSS stops the timer T_WAIT_FLUSH, and either transfers the

pending DL LLC PDUs to the new cell (if the old and new cells belong to the same routing area and the

same NE) or deletes them. The BSS acknowledges the FLUSH-LL PDU by sending a FLUSH-LL-ACK PDU to

the SGSN. When the feature “full intra-RA LLC PDU rerouting” is implemented, a rerouting will be

possible between 2 different NSEs.






1 � 3 � 86


NC Cell Reselection Execution with LLC PDU Rerouting

T_Wait_Flush

T_Ack_Wait

On going UL TBF

On going DL TBF

MS BSSServing cell

Packet Cell Change Order / PACCH

Packet Measurement Report / PACCH

BSSTarget cell

SGSN

Packet Control Acknowledgement / PACCH

Packet Channel Request / PRACH

Packet Uplink Assignment / PCCCH

UL LLC PDU [TLLI]

FLUSH-LL [TLLI, old BVCI, new BVCI, (new NSEI)]

FLUSH-LL-ACK [TLLI, “transferred”]

LLC PDU(s) rerouting

In Packet Transfer Mode, it happens that Downlink LLC PDU frames, which have been transmitted by the

SGSN to the BSS, are not received by the MS because the MS performs a cell reselection. Indeed, these PDUs

are discarded by the BSS. The BSS informs the SGSN that it has discarded these PDUs, and the SGSN has to

send them again. With the feature Downlink LLC PDU rerouting, the BSS keeps in memory these PDUs, and

transmits them to the MS in the target cell, after the cell reselection.

EN_DL_LLC_PDU_REROUTING is the OMC-R parameter that activates the DL rerouting on a per BSS basis.

If the SGSN supports the INR option (Inter-NSE Rerouting), a rerouting is requested by providing the BVCI

and the NSEI of the new cell in the FLUSH-LL message in case of a cell change between two different NSEs.

Otherwise (same NSE) only the BVCI of the new cell is provided.






1 � 3 � 87


Full Intra-RA LLC PDU Rerouting

� Available whatever the NETWORK_CONTROL_ORDER: NC0 or NC2

� If EN_DL_LLC_PDU_REROUTING = enabled

� If the SGSN requests a rerouting (new BVCI included in Flush LL)� The MFS can perform an intra-RA intra-NSE rerouting

� If the SGSN does not support Inter-NSE rerouting� The MFS performs an autonomous intra-RA inter-NSE rerouting

� Summary

DL LLC PDU reroutingNoonly old BVCI

DL LLC PDU deletiononly old BVCI

DL LLC PDU reroutingYes

old BVCI + new BVCI

same RA

different NSEs

DL LLC PDU deletiononly old BVCI

DL LLC PDU reroutingold BVCI + new BVCIsame RAsame NSE

MFS behaviorSGSN Inter-NSEcapability

Flush LLinformation

Old and new ell

In case of MS cell change, the control of the rerouting of DL LLC PDUs from one cell to another is left to the

SGSN

In case of MS cell change, the SGSN sends a Flush-LL PDU to the BSS, in order:

� either to delete the outstanding PDUs in the old cell buffer,

� or to reroute them to the new cell.

When the MS operates in NC2 mode, the “old” and the “new” cells are known by the BSS.

Consequently an autonomous rerouting can be performed at Flush-LL receipt.

When the MS operates in NC0 mode, the BSS does not know the link between the “old” and the “new” cells.

To find this link, the BSS uses the TLLI of the cell Update.

� As this message cannot be identified as such, it is checked that the TLLI of a UL TBF:

� does not exist in any of the MS contexts stored in the cell.

� is not a foreign or a random TLLI.

� A “TLLI retrieval” process is started to try to find a cell of the same RA, on any GPU, in which this TLLI

exists.

� When the search is successful, the rerouting can be performed at Flush-LL receipt.






1 � 3 � 88

� Useful in multilayer and multiband networks, in order to reduce the number of reselections

� If EN_OUTGOING_GPRS_REDIR(s) = enabled

� As soon as the MS is in Packet Transfer Mode, it is redirected from the cell

� Then, the cell ranking process is started to find the best candidate cell

� According to the operator strategy

� See session 4 for a strategy example


Outgoing GPRS Redirection

Exercise

An outgoing GPRS redirection is an NC cell reselection which is triggered when the MS enters the packet

transfer mode in the serving cell even if the radio link is good.

The intention of GPRS redirections is to redirect the MS towards a target cell more appropriate to carry PS

traffic (for instance a macro cell).






1 � 3 � 89

7 Flow Control at the Gb Interface






1 � 3 � 90


BSSGP

� Only DL flow control is performed between the BSS and the SGSN

� Principle of the DL flow control mechanism:

� the BSS sends to the SGSN the flow control parameters in the FLOW-CONTROL-MS/BVC messages

� the flow control parameters allow the SGSN to locally control its transmission towards the BSS

PayloadTLLIBVCI

LLC frame

BSSGP frame

Used to perform MS flow controlUsed to perform BVC flow control

Caution: LLC frames are encapsulated 1:1 into BSSGP frames. This is the reason why we can say that there

is an LLC frame flow control mechanism at BSSGP level.






1 � 3 � 91


Flow Control Performed on SGSN Side

� The SGSN shall perform flow control on each BVC and on each MS

� The flow control is performed on each LLC PDU first by the MS flow control mechanism and then by the BVC flow control mechanism:

� if an LLC PDU is passed by the MS flow control then the SGSN applies the BVC flow control to the LLC PDU

� if an LLC PDU is passed by both flow control mechanisms, the entire LLC PDU is delivered to the BSS

MS flow control MS flow control MS flow control

BVC flow control

BSS






1 � 3 � 92


Flow Control Performed on SGSN Side [cont.]

� Leaky bucket algorithm:

� An LLC PDU is passed as long as the bucket counter (B) plus the length of the LLC PDU does not exceed the bucket size (Bmax)

� When the LLC PDU is passed, its length is added to B

� Any LLC PDU not passed is delayed until B plus the LLC PDU length is less than Bmax

� The algorithm takes into account the leak rate of the bucket (R)

Leaky bucket principle

bucket size (B)

Max bucket size (Bmax)

leaking rate (R)

new LLC PDU?






1 � 3 � 93


Flow Control Performed on BSS Side

� The BSS controls the DL transmission of the SGSN by sending the parameters Bmax and R in the flow control PDU:

� after the sending of a FLOW_CONTROL_BVC PDU, the BSS cannot send a new FLOW_CONTROL_BVC PDU before T_Flow_Ctrl_Cell seconds

� T_Flow_Ctrl_Cell is a BSS parameter

� Default value = 0

� By default, the BVC flow control is disabled

� after the sending of a FLOW_CONTROL_MS PDU, the BSS cannot send a new FLOW_CONTROL_MS PDU before T_Flow_Ctrl_MS seconds

� T_Flow_Ctrl_MS is a BSS parameter

� Default value = 10 s

NB: the cell flow control is performed more frequently than the MS flow control because:

� The radio resource availability for a TBF is always shorter than the guarding time of a PDCH, therefore

the MS individual traffic is less of an influence on the leaking rate.

� The radio resource available for one MS may change from one TBF to another.

� The combined traffic of all the GPRS MSs in the cell exchanging data with the SGSN has to be mapped

onto a BVC, which may become the blocking factor as the BVC is mapped on an NSVC, which is mapped on

a PVC, carried by a BC which has a fixed maximum capacity.






1 � 3 � 94


Flow Control Performed on BSS Side [cont.]

� FLOW_CONTROL_BVC PDU:

� BVC_Bucket_Size: the maximum size of the cell buffer in the MFS

� BVC_Bucket_Leak_Rate: the measured throughput in the cell from the RRM to the RLC

� Bmax_default_MS: the default value of the maximum size of the MS buffer in the MFS

� R_default_MS: the default value of the measured throughput for the MS from the RRM to the RLC

Formulas:

�The BVC_Bucket_Size (value expressed in octet) is calculated as follows:

� Case T_Flow_Ctrl_Cell ≠ 0:

� BVC_Bucket_Size = Flow_Dim_safety_MS * Max_PDCH * Max_Rate_PDCH * (1/8) * Max_Rate_Safety * T_Flow_Ctrl_Cell

� Case T_Flow_Ctrl_Cell = 0:

� BVC_Bucket_Size = Flow_Dim_safety_MS * Max_PDCH * Max_Rate_PDCH * (1/8) * Max_Rate_Safety *

Def_value_T_Flow_Ctrl_Cell

�The BVC_Bucket_Leak_Rate (value expressed in 100 bits/sec) is calculated as follows:

� Case T_Flow_Ctrl_Cell ≠ 0:

� IF B_BVC < BVC_Bucket_Size

� BVC_Bucket_Leak_Rate = [(BVC_Bucket_Size - B_BVC )*8] / [T_Flow_Ctrl_Cell * Flow_Dim_safety_BVC * 100]

� ELSE

� BVC_Bucket_Leak_Rate = 0

� Case T_Flow_Ctrl_Cell = 0

� BVC_Bucket_Leak_Rate = (Max_PDCH * Max_Rate_PDCH * Max_Rate_Safety)/100

�The Bmax_default_MS (value expressed in octet) is calculated as follows:

� Case T_Flow_Ctrl_MS ≠ 0:

� Bmax_default_MS = Flow_Dim_safety_MS * Max_Rate_PDCH * (1/8) * Max_Rate_Safety * T_Flow_Ctrl_MS

� Case T_Flow_Ctrl_MS = 0:

� Bmax_default_MS = Flow_Dim_safety_MS * Max_Rate_PDCH * (1/8) * Max_Rate_Safety * Def_value_T_Flow_Ctrl_MS

�The R_default_MS (value expressed in 100 bits/sec) is calculated as follows:

� R_default_MS = (Flow_Dim_safety_MS * Max_Rate_PDCH * Max_Rate_Safety)/100






1 � 3 � 95


Flow Control Performed on BSS Side [cont.]

� FLOW_CONTROL_MS PDU:

� MS_Bucket_Size: the maximum size of the MS buffer in the MFS

� MS_Bucket_Leak_Rate: the measured throughput for the MS from the RRM to the RLC

Formula:

� The MS_Bucket_Size (value expressed in octet) is calculated as follows:

� MS_Bucket_Size = n * Max_Rate_PDCH * (1/8) * Max_Rate_Safety * T_Flow_Ctrl_MS

� The MS_Bucket_Leak_Rate (value expressed in 100 bit/sec) is calculated as follows:

� MS_Bucket_Leak_Rate = B_MS * 100 / MS_Bucket_Size

Explanation:

� Max_PDCH

� O&M parameter indicating the maximum number of PDCHs that can be established in the cell.

� Max_Rate_PDCH

� maximum rate of one PDCH in the considered cell (value in bits/s)

� Max_Rate_Safety

� Safety factor to compensate the Max_Rate_PDCH in the calculation of BVC_Bucket_Size and MS_Bucket_Size

� Flow_Dim_safety_BVC

� O&M safety factor, used to tune the BVC bucket value

� Flow_Dim_safety_MS

� O&M safety factor, used to tune the MS bucket value

� MAX_LLC_PDU

� maximum length of a DL LLC PDU (the SGSN has to be able to send at least one DL LLC PDU)

� B_BVC

� value in octet of the current bucket size at MFS side for the cell. It corresponds to the amount of LLC waiting frames for this BVC (cell)

� T_Flow_Ctrl_Cell

� sending period of NM-FLOW-CONTROL-CELL-req

� T_Flow_Ctrl_MS

� sending period of NM-FLOW-CONTROL-MS-req

� Def_value_T_Flow_Ctrl_Cell

� default value forT_Flow_Ctrl_Cell (set to 5 sec)

� Def_value_T_Flow_Ctrl_MS

� default value for T_Flow_Ctrl_MS (set to 10 sec)

� B_MS

� Value in octets of the current bucket size at MFS side for the MS. It corresponds to the amount of LLC PDUs waiting to be transmitted for this MS.

� n

� Maximum number of PDCHs that can be allocated to the MS according to its multislot class.






1 � 3 � 96

8 Radio Link Supervision






1 � 3 � 97


Principles

� During a UL or DL packet transmission, the corresponding TBF can be released due to an abnormal situation:

� no acknowledgement or data received

� the transmission is stalled

� too low transmission efficiency

� The abnormal release is always followed by the re-establishment of the TBF in case of an uplink transfer (initiative of the MS)

� In case of a downlink transfer, most of the SGSNs do not take the initiative to re-establish the TBF

The RLS mechanisms processes in the MFS are based on the following assumption:

« in a specific transfer situation, the MFS is expecting the MS to behave in a specific way »:

� In a UL TBF, the MFS schedules a USF for UL blocks and expects the MS to understand the MFS’s

acknowledgements.

� In a DL TBF, the MFS sends blocks to the MS and expects them to be acknowledged when scheduled by the

MFS (use of RRBP).






1 � 3 � 98


DL TBF

DL ACK/NACK PERIOD blocks

RRBP

≠≠≠≠ false

Ø N3105 = N3105+1

PDTCH PDTCH

PACCH

Scheduling of “Packet DL Ack/Nack”

PACCH block

N3105>N3105_LIMIT

PDTCH

Stop sending DLPDTCH blocks

The MFS counts the number of consecutive PACKET DL ACK/NACK not received due to loss on the radio

interface:

� For a GPRS TBF, if the counter is above the threshold TBF_CS_DL and CS-4, CS-3 or CS-2 is used, the MFS

switches to CS-1.

� For an EGPRS TBF, if the counter is above the threshold TBF_MCS_DL and MCS-9, MCS-8…, MCS-3 or MCS-2

is used, the MFS switches to MCS-1.

� If the counter is above the threshold N3105_LIMIT, the DL TBF is abnormally released:

� the MFS stops sending packets to the MS and sends a message to the SGSN (Radio Status).

� it is up to the SGSN to re-establish the DL TBF.

� the MS releases the TBF on its side.

If N3105_LIMIT < TBF_CS_DL then the loss of consecutive packet DL ACK/NACK will not trigger a link

adaptation but a TBF release.

For an EGPRS TBF, the MFS considers EGPRS_N3105_LIMIT.

N3105_LIMIT = 20 (Alcatel recommended value) and it can be set at OMC-R level (cell level).

EGPRS_N3105_LIMIT = 20(Alcatel recommended value) and it can be set at OMC-R level (cell level).






1 � 3 � 99


DL TBF [cont.]

DL ACK/NACK PERIOD blocks

RRBP

≠≠≠≠ false

PDTCH PDTCH

PACCH

Packet DL Ack/Nack

Scheduling of “Packet DL Ack/Nack”

PACCH block

PDTCH

Stop sending DLPDTCH blocks

N_StagnatingWindowDL =N_StagnatingWindowDL

+1

N_StagnatingWindowDL > NstagnatingWindowDL_LIMIT

Same oldest RLC block

Nack in the RBB

Other Abnormal DL TBF release: DL window stalled

� In GPRS acknowledged mode, NstagnatingWindowDL counter shall be incremented when the same oldest

RLC data block in the transmit window is not acknowledged by the last received bitmap.

� If N_StagnatingWindowDL exceeds its limit, then the network shall terminate the TBF.

� NStagnatingWindowDL_LIMIT = 32 (Alcatel recommended value) and it canbe set at OMC-R level (cell

level).






1 � 3 � 100


UL TBF - Abnormal Release with Random Access

N3101 = N3101+NØPDTCH

N3101>N3101_LIMIT

PACCH

Stop sending “Packet ULACK/NACK” PACCH blocks

ØPDTCH

N consecutive

USF

USF

Packet Random Access


The MFS manages several counters:

� N3101: number of RLC PDUs consecutively lost since the last reception of a UL RLC PDU:

� N3101 is incremented each time a UL radio block is allocated to the MS and no data is received.

� if N3101 is above N3101_LIMIT, the UL TBF is abnormally released: the MFS stops sending PACKET UL

ACK/NACK to the MS.

� The MS waits for PACKET UL ACK/NACK and then releases the TBF on its side.

� Then the MS sends a random access to re-establish the UL TBF.






1 � 3 � 101


UL TBF - Abnormal Release with Random Access [cont.]

SI=1

PDTCH

N_StagnatingWindowUL =N_StagnatingWindowUL +1

PACCH


SI=1

PDTCH

USF

USF

Packet Random Access

PACCH

N_StagnatingWindowUL > NstagnatingWindowUL_LIMIT

Other Abnormal UL TBF release: UL window stalled

� SI=1 in a UL RLC DATA BLOCK indicates that the MS transmit window is stalled.

� Upon detection of a stall condition, the network sends a Packet Uplink Ack/Nack message and after a

round trip delay has elapsed, it increments N_ULStagnatingWindow.

� If N_StagnatingWindowUL exceeds its limit, then the network shall terminate the TBF.

� NstagnatingWindowUL_LIMIT = 10 (Alcatel recommended value) and it can be set at OMC-R level (cell

level).






1 � 3 � 102


UL TBF - Abnormal Release with Cell Reselection

RESELECTION

N3102

PAN_MAX

0

UL TBF

Abnormal

release

Random

Access

PAN_DEC

PAN_INC

Packet UL Ack/Nack

received OK

Abnormal release with cell reselection:

� procedure linked to the counter N3102 internal to the MS and initialized to PAN_MAX after each

reselection:

� each time the MS performs an abnormal release with random access, it decreases N3102 by PAN_DEC.

� each time the MS receives a PACKET UL ACK/NACK, it increases N3102 by PAN_INC.

� if N3102 reaches 0, the MS performs an abnormal release with cell reselection.

� the MS triggers a cell reselection procedure but nothing allows it to change its serving cell (need of a

Master PDCH to be able to re-select a new cell).

� after the cell reselection, the MS sends a random access to re-establish the UL TBF.






1 � 3 � 103


UL TBF - Abnormal Release with Cell Reselection [cont.]

� If a Master PDCH is available in the serving cell

AND

� If RANDOM_ACC_RETRY = Allowed

� Then the following reselection algorithm is applied:

� The MS re-selects the cell with the highest RLA among the 6 best levels

� In this cell, if the MS cannot decode the PBCCH data block, it reselects the next highest Received Level Average

� If the cells with the 6 strongest RLAs have been tried but cannot be used, the MS performs a normal reselection (see 5 NC0 Cell Selection and Reselection)

� Else the normal reselection algorithm is applied (see 4)

� After T_RESEL, the MS is allowed to reselect the serving cell

RLA = Received Level Average.

T_RESEL = 5s (default value).

Extract of the 05.08 GSM standard:

In the event of an abnormal release with cell reselection (see 3GPP TS 04.60) when PBCCH exists, an

abnormal cell reselection based on BA(GPRS) shall be attempted. The MS shall perform the following

algorithm to determine which cell to be used for this cell reselection attempt.

If access to another cell is not allowed, i.e., RANDOM_ACCESS_RETRY bit is not set on the serving cell:

� i) The abnormal cell reselection attempt shall be abandoned.

If access to another cell is allowed, i.e., RANDOM_ACCESS_RETRY bit is set on the serving cell:

� i) The received level measurement samples taken on the carriers indicated in the BA (GPRS) received on

the serving cell in the last 5 seconds shall be averaged, and the carrier with the highest Received Level

Average (RLA) with permitted BSIC, i.e., the same as broadcast together with BA (GPRS), shall be taken.

� ii) On this carrier, the MS shall attempt to decode the PBCCH data block containing the parameters

affecting cell selection.

� iii) If the cell is suitable (see 3GPP TS 022), abnormal cell reselection shall be attempted on this cell.

� iv) If the MS is unable to decode the PBCCH data block or if the conditions in iii) are not met, the carrier

with the next highest Received Level Average (RLA) with permitted BSIC shall be taken, and the MS shall

repeat steps ii) and iii) above.

� v) If the cells with the 6 strongest Received Level Average (RLA) values with permitted BSICs have been

tried but cannot be used, the abnormal cell reselection attempt shall be abandoned.

The MS is under no circumstances allowed to access a cell to attempt abnormal cell reselection later than

20 seconds after the detection within the MS of the abnormal release causing the abnormal cell reselection

attempt. In the case where the 20 seconds elapse without a successful abnormal cell reselection, the

attempt shall be abandoned.






1 � 3 � 104


UL TBF in Ending Phase

N3103 = N3103+1ØPACCH

N3103>N3103_LIMIT

PACCH


Final

Block

PDTCH

USF

U

SF

PACCH

Final

Ack

Scheduling of “Packet Control Ack”

The MFS can also trigger an abnormal release at the end of a UL TBF:

� the MFS counts the number of PACKET CONTROL ACK not received in response to the PACKET UL

ACK/NACK which indicates the end of the TBF.

� if the counter is above N3103_LIMIT, the UL TBF is abnormally released: the MFS stops sending PACKET UL

ACK/NACK to the MS.







1 � 3 � 105


UL and DL TBF

� A (E)TX_Efficiency is computed every

� TX_EFFICIENCY_PERIOD transmitted RLC data blocks for a GPRS TBF

� E_TX_EFFICIENCY_PERIOD transmitted RLC data blocks for an EGPRS TBF

� and compared to the following thresholds:

� TX_EFFICIENCY_ACK_THR in Acknowledged mode

� TX_EFFICIENCY_NACK_THR in Non-Acknowledged mode

� If the TX_Efficiency is below these thresholds, the TBF must be released

� It is done as an abnormal release by the MFS:

� the MFS stops sending DL RLC PDUs in case of a DL TBF

� the MFS stops sending PACKET UL ACK/NACK in case of a UL TBF

Exercise

Radio Link Supervision based on TX_Efficiency monitoring

� It was proposed since B7 to use the transmission efficiency, i.e., the ratio of the average net bit rate over the gross

bit rate.

� This transmission efficiency can be computed approximately as:

� Where:

� NB_SENT is the number of transmitted RLC data blocks,

� NB_RECEIVED is the number of correctly received RLC data blocks (i.e., blocks such that a positive

acknowledgment is reported),

� ρi is equal to the number of information bits in the i-th correctly received RLC data block divided by the number of bits per RLC data block with GMSK modulation (456 in GPRS). This ratio only depends on the coding scheme

used for the i-th correctly received RLC data block and is between 0 and 1 in GPRS and between 0 and 3 in

EGPRS (3 because there are 3 information bits per 8-PSK symbol).

� ni is the number of RLC data blocks in the i-th radio block. Therefore, this number is always equal to 1 for GPRS

and EGPRS for MCS-1 to MCS-6, and is equal to 2 in EGPRS for MCS 7 to MCS 9.

� ρi = 0,40 for CS-1, 0,59 for CS-2, 0.68 for CS-3 and 0.94 for CS-4.

� TX_EFFICIENCY is computed during a fixed window of TX_EFFICIENCY_PERIOD data blocks and then compared to

threshold (TX_EFFICIENCY_ACK_THR if Ack mode and TX_EFFICIENCY_NACK_THR if Nack).

Then if TX_EFFICIENCY < Tx_efficiency_threshold then the TBF is release (abnormally).

� TX_EFFICIENCY_ACK_THR = 10%, TX_EFFICIENCY_NACK_THR = 15%, TX_EFFICIENCY_PERIOD = 200

all can be set at OMC-R level.

∑

∑

=

==SENTNB

i i

RECEIVEDNB

i i

n

nEFFICIENCYTX

_

1

_

1

i

1100_

ρ

2.602.452.021.300.980.770.650.490.39ρi

MCS-9MCS-8MCS-7MCS-6MCS-5MCS-4MCS-3MCS-2MCS-1






1 � 3 � 106

9 Exercises






1 � 3 � 107

9 Exercises

GPRS CS Adaptation

� CS adaptation / DL measurements

� Network parameters:

� MAX_GPRS_CS = CS-2

� TBF_DL_INIT_CS = CS-1

� CS_QUAL_DL_1_2_X_Y = 2

� CS_HST_DL_LT = 2

� CS_HST_DL_ST = 4

� Objective: Find CS used in the DL

Time allowed:

10 minutes






1 � 3 � 108

9 Exercises

GPRS CS Adaptation [cont.]

� Find which CS is used at each measurement

Measurement 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

RXQUAL_DL 0 0 2 3 4 5 5 5 6 0 0 0 0 6 7 7

AV_RXQUAL_DL_LT 0,0 0,0 1,1 1,9 2,6 3,3 3,8 4,1 4,5 3,5 2,7 2,1 1,7 2,6 3,5 4,2

AV_RXQUAL_DL_ST 0,0 0,0 1,7 2,7 3,8 4,8 5,0 5,0 5,8 1,2 0,2 0,0 0,0 4,8 6,6 6,9

CS ?

0

1

2

3

4

5

6

7

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

RXQUAL_DL AV_RXQUAL_DL_LT AV_RXQUAL_DL_ST

Back

Short term average is calculated with AlphaST = 0.2

Short term average is calculated with AlphaLT = 0.8




9 Exercises

RLC ACK Mode: New DL MCS Value Determination (1/3)

APD=0dB, Type 2 ARQ, GMSK table: APD=0dB, Type 2 ARQ, 8PSK table:

if MCScurrent belongs to {1,2,3,4} if MCScurrent belongs to {5,6,7,8,9}

0 1 2 3 4 5 6 70 1 1 1 1 1 1 1 14 1 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 13 1 1 1 1 1 1 1 14 1 1 1 1 1 1 1 15 2 2 1 1 1 1 1 16 2 2 2 2 1 1 1 17 5 2 2 2 2 1 1 18 5 2 2 2 2 2 2 29 5 2 2 2 2 2 2 2

10 5 3 3 2 2 2 2 211 5 3 3 3 3 3 3 312 5 3 3 3 3 3 3 313 5 3 3 3 3 3 3 314 5 5 3 3 3 3 3 315 6 5 3 3 3 3 3 316 6 5 3 3 3 3 3 317 6 5 3 3 3 3 3 318 6 5 3 3 3 3 3 319 6 6 5 3 3 3 3 420 6 6 5 5 3 3 3 421 7 6 5 5 3 3 4 422 7 6 6 5 5 4 4 423 7 6 6 6 5 4 4 424 7 7 6 6 5 5 5 425 7 7 7 6 6 5 5 526 7 7 7 6 6 5 5 527 7 7 7 7 6 6 5 528 7 7 7 7 6 6 6 629 7 7 7 7 7 7 6 630 7 7 7 7 7 7 7 731 7 7 7 7 7 7 7 9

CV_BEP

ME

AN

_BE

P

0 1 2 3 4 5 6 70 5 5 5 5 1 1 1 11 5 5 5 5 1 1 2 2

2 5 5 5 5 1 2 2 23 5 5 5 5 2 2 3 34 5 5 5 5 2 2 3 35 5 5 5 5 5 3 3 36 5 5 6 5 5 3 3 37 5 5 6 5 5 5 3 38 5 5 6 5 5 3 3 49 7 6 6 6 5 5 5 4

10 7 6 6 6 5 5 5 511 7 6 6 6 6 5 5 512 7 6 6 6 6 6 5 513 7 6 6 6 6 6 5 514 7 6 6 6 6 6 5 615 7 7 6 6 6 6 6 616 7 7 7 7 6 6 6 617 7 7 7 7 7 7 7 618 7 7 7 7 7 7 7 719 7 7 7 7 7 7 7 720 7 7 8 7 7 7 7 721 7 8 8 8 8 7 7 722 8 8 8 8 8 8 8 823 8 8 8 8 8 8 8 824 8 8 8 8 8 8 8 825 8 8 8 8 8 8 8 826 8 8 8 8 9 9 9 927 8 8 8 9 9 9 9 928 9 9 8 9 9 9 9 929 9 9 8 9 9 9 9 930 9 9 8 9 9 9 9 931 9 9 9 9 9 9 9 9

CV_BEP

ME

AN

_BE

P






1 � 3 � 110

9 Exercises

RLC ACK Mode: New DL MCS Value Determination [cont.]

� Using the previous LA tables and the following information, fill in the next table:

� APD = 0 dB

� DL RLC mode = ACK

� MS OUT OF MEMORY = Off

Time allowed:

10 minutes






1 � 3 � 111

9 Exercises

RLC ACK Mode: New DL MCS Value Determination [cont.]

� Find which DL MCS is used at each measurement

Back

1 2 3 4 5 6 7 8CV_BEP 2 2 2 3 7 5 3 0MEAN_BEP 20 24 24 18 9 8 5 8MCSindNew MCS 4




9 Exercises

RLC ACK Mode: New UL MCS Value Determination (1/3)

APD=0dB, Type 1 ARQ, GMSK table: if MCScurrent belongs to {1,2,3,4}

0 1 2 3 4 5 6 70 1 1 1 1 1 1 1 11 1 1 1 1 1 1 1 1

2 1 1 1 1 1 1 1 13 1 1 1 1 1 1 1 14 1 1 1 1 1 1 1 15 2 2 1 1 1 1 1 16 2 2 2 2 1 1 1 17 5 2 2 2 2 2 2 28 5 2 2 2 2 2 2 29 5 2 2 2 2 2 2 210 5 3 2 2 2 2 2 311 5 3 3 3 3 3 3 312 5 5 3 3 3 3 3 313 5 5 3 3 3 3 3 314 6 5 3 3 3 3 3 315 6 5 3 3 3 3 3 316 6 5 5 3 3 3 3 317 6 5 5 3 3 3 3 318 6 5 5 3 3 3 3 319 6 6 5 5 3 3 3 420 6 6 5 5 5 3 3 421 7 6 6 5 5 4 4 422 7 6 6 5 5 5 4 423 7 6 6 6 5 5 5 424 7 6 6 6 6 5 5 425 7 7 7 6 6 5 5 526 7 7 7 6 6 5 5 527 7 7 7 7 6 6 6 528 7 7 7 7 6 6 6 629 7 7 7 7 7 7 6 630 7 7 7 7 7 7 6 631 7 7 7 7 7 7 7 9

CV_BEP

ME

AN

_BE

P






1 � 3 � 113

9 Exercises

RLC ACK Mode: New UL MCS Value Determination [cont.]

� Using the previous LA table and the following information, fill in the next table:

� APD = 0 dB

� UL RLC mode = ACK

� EN_RESEGMENTATION_UL = enabled

Time allowed:

20 minutes






1 � 3 � 114

9 Exercises

RLC ACK Mode: New UL MCS Value Determination [cont.]

� Find which UL MCS is used every 12 radio blocks

Back

1 2 3 4 5 6 7 8 9 10 11 12CV_BEP 2 1 2 1 3 7 5 3 0 2 5 7MEAN_BEP 20 20 24 24 18 9 8 5 8 10 18 8MCSindN_inf 0N_sup 0New MCS 2

13 14 15 16 17 18 19 20 21 22 23 24CV_BEP 5 2 1 3 2 1 2 5 2 3 7 3MEAN_BEP 8 7 8 12 16 20 24 10 12 7 20 27MCSindN_inf 0N_sup 0New MCS






1 � 3 � 115

9 Exercises

Type II ARQ Mechanism

� Replace the ‘?’ by the right values in the 6 next cases:

� Cases 1a, 1b and 1c:


� EN_FULL_IR_DL = disabled

� Case 2:


� EN_FULL_IR_DL = enabled

� Cases 3a and 3b:

� MS OUT OF MEMORY = On

� EN_FULL_IR_DL = disabled

Time allowed:

20 minutes

Back






1 � 3 � 116

� Case 1a: MS OUT OF MEMORY = Off, EN_FULL_IR_DL = disabled

MS BSS

DL RLC data block first part B2, MCS?, PS?

EGPRS Packet DL Ack/Nack (B2 not received)

DL RLC data block second part B2, MCS?, PS?

DL RLC data blocks B3+B4, MCS7, PS1

DL RLC data blocks B5+B6, MCS7, PS1, + polling request

MCS4 commanded by

the Link Adaptation

algorithm


DL RLC data block B7, MCS?, PS?

9 Exercises







1 � 3 � 117

� Case 1b: MS OUT OF MEMORY = Off, EN_FULL_IR_DL = disabled

MS BSS






MCS5 commanded by

the Link Adaptation

algorithm


9 Exercises







1 � 3 � 118

� Case 1c: MS OUT OF MEMORY = Off, EN_FULL_IR_DL = disabled

MS BSS

DL RLC data blocks B2+B4, MCS?, PS?

EGPRS Packet DL Ack/Nack (B2 and B4 not received)





9 Exercises







1 � 3 � 119

� Case 2: MS OUT OF MEMORY = Off, EN_FULL_IR_DL = enabled

MS BSS






MCS4 commanded by

the Link Adaptation

algorithm


9 Exercises







1 � 3 � 120

� Case 3a: MS OUT OF MEMORY = On, EN_FULL_IR_DL = disabled

MS BSS


EGPRS Packet DL Ack/Nack (B2 and B4 not received,

MS OUT OF MEMORY = On)





9 Exercises







1 � 3 � 121

� Case 3b: MS OUT OF MEMORY = On, EN_FULL_IR_DL = disabled

MS BSS

DL RLC data block first part B2, MCS?, PS?

EGPRS Packet DL Ack/Nack (B2 not received,

MS OUT OF MEMORY = On)

DL RLC data block second part B2, MCS?, PS?



MCS4 commanded by

the Link Adaptation

algorithm



9 Exercises


Back






1 � 3 � 122

9 Exercises

NC0 Cell Selection and Reselection

� Master Channel is NOT used

� Network configuration is explained hereafter

� The MS (2W, class B) is selecting a first cell and immediately starts a transfer

� Objective: Find cells selected by the MS

Time allowed:

10 minutesCI=6169

GSM900

CI=1823

GSM900

CI=1964GSM900

CI=6270

GSM900

CI=6271

GSM900 Cell 3 (8557, 1823)

Cell 2 (8564,6169)

Cell 1 (8564, 1964)






1 � 3 � 123

9 Exercises

NC0 Cell Selection and Reselection [cont.]

� Parameters settings

� For all cells:

� RX_LEV_ACCESS_MIN = -103 dBm

� MS_TXPWR_MAX_CCH_= 33 dBm

� PENALTY_TIME = 0 (20s)

� TEMPORARY_OFFSET = 0 dB

� CELL_RESELECT_OFFSET = 0 dB

� CELL_RESELECT_HYSTERESIS

� Cell 1: 4 dB

� Cell 2: 6 dB

� Cell 3: 6 dB

CI=6169

GSM900

CI=1823

GSM900

CI=1964GSM900

CI=6270

GSM900

CI=6271

GSM900 Cell 3 (8557, 1823)

Cell 2 (8564,6169)

Cell 1 (8564, 1964)






1 � 3 � 124

9 Exercises

NC0 Cell Selection and Reselection [cont.]

� Find the cell selected by the MS

CI=6169

GSM900

CI=1823

GSM900

CI=1964GSM900

CI=6270

GSM900

CI=6271

GSM900 Cell 3 (8557, 1823)

Cell 2 (8564,6169)

Cell 1 (8564, 1964)

5

4

3

2

1

Measurements

-77-85-89

-82-87-88

-87-90-88

-100-90-84

-104-96-80

RxLev (3)RxLev

(2)

RxLev

(1)

Back






1 � 3 � 125

9 Exercises

NC2 Cell Reselection

� Parameters settings

� For all cells:

� RX_LEV_ACCESS_MIN = -103 dBm

� MS_TXPWR_MAX_CCH_= P = 33 dBm

� PENALTY_TIME = 0 (20s)

� T_NC_PING_PONG = 0s

� NC_PING_PONG_OFFSET = 0 dB

� THR_NC2_LOAD_RANKING = 70 %

� Objectives:

� Fill in the following table

� Find the best candidate cell

Time allowed:

20 minutes

Back






1 � 3 � 126

9 Exercises

NC2 Cell Reselection [cont.]

� All the cells belong to the same BSS

C2NC2

Load situation

C31NC2

C1NC2

disableddisableddisabledenableddisabledEN_OUTGOING_GPRS_REDIR

0 dB0 dB0 dB+20 dB+20 dBCRO

-90 dBm-90 dBm-90 dBm-75 dBm-75 dBmHCS_THR

-88 dBm-85 dBm-82 dBm-70 dBm-78 dBmAV_Rxlev_NC2

30%20%80%0%10%Load Average

Cell5

umbrella

Cell4

umbrella

Cell3

umbrella

Cell2

micro

Cell1

MicroCell Type






1 � 3 � 127

9 Exercises

NC2 Cell Reselection [cont.]

� GPRS redirection

� Find a parameter setting ensuring that when the MS enters the Packet Transfer Mode, it is redirected towards a macro cell

Time allowed:

10 minutes

Back

Macro cell

Micro cell Micro cell

GPRSRedirection






1 � 3 � 128

9 Exercises

Radio Link Supervision

� List in the DL and UL the different cases of abnormal release.

Time allowed:

10 minutes






1 � 3 � 129









1 � 3 � 130

End of ModuleRadio Link Control






Module 4Algorithms Dynamic Behaviors









EVOLIUM � E-GPRS Radio Algorithms and Parameters Description B10Radio Algorithms � Algorithms Dynamic Behaviors

1 � 4 � 2

Blank Page




Document History






1 � 4 � 3

Module Objectives


� Estimate qualitatively the impact of a parameter change in order to solve the typical problems or enhance the GPRS performance






1 � 4 � 4








1 � 4 � 5

Table of Contents


1 Optimization of CS Adaptation 72 Optimization of Cell Reselection 103 Enhance the (E)GPRS Performance 17






1 � 4 � 6









1 � 4 � 7

1 Optimization of CS Adaptation






1 � 4 � 8


Quiz

� Why CS_QUAL_XX_I_J_X_NACK < CS_QUAL_XX_I_J_X_ACK, by default?

� What is the meaning of CS_SIR_DL_3_4_FH_NACK = 15?

� What is the meaning of CS_QUAL_UL_3_4_FH_NACK = 0?

Time allowed:

15 minutes






1 � 4 � 9


Qualitative Impact

� Fill in the table

CS_HST_UL_LT

CS_SIR_DL_3_4_X_Y

CS_QUAL_DL_1_2_X_Y

N_AVG_I

Qualitative ImpactChange






1 � 4 � 10

2 Optimization of Cell Reselection






1 � 4 � 11


Cases

� Case 1: Multilayer network with PBCCH

� Use NC0 with C31 and C32

� Case 2: Multilayer network without PBCCH

� Use Outgoing GPRS redirection

� Tuning of NC2 parameters

� Case 3: NC2 parameters versus HO parameters






1 � 4 � 12


Case 1: Multilayer Network with PBCCH

� Cell reselection for CS traffic

� Aim: favor cell reselection on micro cells for slow mobiles

Macro cell


Slowmobiles

CRO=0dB

PENALTY_TIME=20s

TEMPORARY_OFFSET=0dB

CRO=10dB

PENALTY_TIME=60s

TEMPORARY_OFFSET=40dB






1 � 4 � 13


Case 1: Multilayer Network with PBCCH [cont.]

� Cell reselection for PS traffic

� Aim: favor cell reselection on macro cells for all mobiles

� Use C31 and C32

Macro cell


PStraffic

PRIORITY_CLASS=?

HCS_THR=?

PRIORITY_CLASS=?

HCS_THR=?






1 � 4 � 14


Case 2: Multilayer Network Without PBCCH

� Cell reselection for PS traffic

� Aim: favor cell reselection on macro cells for all mobiles

� Use Outgoing GPRS redirection

Macro cell


GPRSredirection

NC_DL_RXLEV_LIMIT_THR=?

Macro cellLOADED

Micro cell

: NC_RESELECT_HYSTERESIS(micro, Macro)

: NC_RESELECT_HYSTERESIS(Macro, micro)

: NC_RESELECT_HYSTERESIS(micro, micro)

NC_DL_RXLEV_LIMIT_THR=?

? ? ? ? ?

? ?






1 � 4 � 15


Case 3: NC2 Parameters Versus HO Parameters

� Link the NC2 parameters to their equivalent HO parameters

NC2

NC_RXQUAL_AVG_PERIOD

NC_RXLEV_FORGETTING_FACTOR

NC_UL_RXQUAL_THR

NC_DL_RXQUAL_THR

NC_DL_RXLEV_THR

NC_RESELECT_HYSTERESIS

HO

L_RXLEV_DL_H

A_QUAL_HO

L_RXQUAL_DL_H

HO_MARGIN

A_LEV_HO

L_RXQUAL_UL_H

L_RXLEV_DL_H: Downlink level threshold for handover.

A_QUAL_HO: Window size for quality averages for handover.

L_RXQUAL_DL_H: Downlink quality threshold for handover for non AMR calls.

HO_MARGIN: Difference in power budget (PBGT) between cell(0) and cell(n) which is required for a power

budget HO.

A_LEV_HO: Window size for level averages for handover.

L_RXQUAL_UL_H: Uplink quality threshold for handover for non-AMR calls.






1 � 4 � 16


Case 3: NC2 Parameters Versus HO Parameters [cont.]

� Fill in the table

NC_RESELECT_HYSTERESIS

NC_DL_RXLEV_THR

NC_DL_RXQUAL_THR

NC_UL_RXQUAL_THR

NC_RXLEV_FORGETTING_FACTOR

NC_RXQUAL_AVG_PERIOD

Qualitative ImpactChange






1 � 4 � 17

3 Enhance the (E)GPRS Performance






1 � 4 � 18


GPRS Services

Location services• Traffic Conditions• Itineraries• Nearest Restaurant, Cinema, Chemist, Parking, ATM, etc.

Fun• Games (Hangman, Poker, etc.)• Screen Saver• Ring Tone• Horoscope• Biorhythm

MediaAlways-on

M-commerce

Mobile Office• Voice (!)• E-mail• Agenda• IntraNet/InterNet• Corporate Applications• Database Access

Transportation• Flight/train

Schedule

• reservation

Vertical application

•Traffic

Management

•Automation

•Mobile branches

•Health

Music• Downloading ofmusic files orvideo clips

News(general/specific)• International/National News• Local News• Sport News• Weather• Lottery Results• Finance News, etc.

Physical• On-line shopping• On-line food

Non physical• on-line Banking• Ticketing• Auction• Gambling, etc.

Directories• Yellow/White Pages• International Directories• Operator Services






1 � 4 � 19


GPRS Services vs QoS Criteria

BER E-mail (smtp)

BERFile transfer (ftp)

ThroughputVideo streaming

- Access delay

- Throughput

WEB Browsing (http)

Access delayWAP

Key QoS criteriaGPRS and EGPRS Services

� According to the service type, the QoS criteria to take into account are different






1 � 4 � 20


Examples

� Data applications use TCP/IP protocol layers which have a great impact on the end user QoS

(*)

First IP router Last IP router

GTP

relay

SNDCPP

BSSGPRLC BSSGP

relay

relay

SNDCPP

PPP

FTP

TCP

IP

PPP

LLC

RLC

MAC

RF

MAC

RF

NS

L1

NS

L1

LLC UDP

IP-Gn

L2

L1’

UDP

IP-Gn

L2

L1’

IP

relay

IP

GTP

IP

relay

IP

relay

IP

UmRGb

Gn GiTE (PC,PDA …)

MSBSS

SGSNGGSN

Possible repartition on the end to end path of the TCP flight size

TCP data segment

TCP acknowledgement

(*) this graphical representation is used toexpress the fact that many data segments arecurrently waiting to be transmitted on therepresented link and are stored in buffers of thedevice handling the link . It doesn’t mean thatsimultaneous segments are being transmitted.

FTP above TCP/IP layers






1 � 4 � 21


PDU Lifetime and TCP Performance

Time

Throughput

Congestion Window increase CW decrease

CW increase CW decrease

� Impact on TCP performance






1 � 4 � 22


PDU Lifetime and TCP Performance [cont.]

TCP

server

SGSN GGSN

TCP

server

SGSN GGSN

MFS

MFS

PDU Life Time

TCP/IP frames

TCP/IP frames

LLC frames

LLC frames

TCP Window = 16 KB

TCP Window = 64 KB

We observe a break within each FTP transfer.

We observe within the Gb traces several "LLC Discarded" messages, just before the TCP starts

retransmissions. Those "LLC discarded" messages show that several kilobytes of data are discarded by the

BSS.

This LLC frame discarding is caused by a "PDU lifetime" timer expiry: indeed this parameter is set by the

SGSN to ** 8 seconds **.

Clearly this value is not enough as the RTT (TCP Round Trip Time) with a TCP window of 64 KB is roughly

12.3 seconds.

As most of the RTTs are composed of queuing in the BSS buffers, this inevitably causes PDU lifetime expiry.

This is a normal behavior as at the beginning of a transfer, the FTP server increases continuously its

congestion window. The BSS has to send more and more data with the same radio bandwidth.






1 � 4 � 23


PDU Lifetime and TCP Performance [cont.]

MFS

FTP

server

SGSN GGSN

PDU Life Time

TCP/IP framesLLC frames

TCP Window = 64 KB

� Solution:

� Increase the PDU Life Time (SGSN parameter)

� PDU Life Time = 63s






1 � 4 � 24


TCP Window Size and FTP Performance

� TCP/IP packet:

� MSS: The maximum number of user data bytes that can be included in the packet without fragmentation.

� MTU: The maximum number of bytes that can be sent in a single packet.

� TCP window size: the period used for acknowledgment. Its value is a multiple of the MSS (x4, 8, 16, 32). The maximum value is 65.535 (64 KB).

MSS

MTU

TCP/IP Header

40 bytes

MSS: Maximum Segment Size.

MTU: Maximum Transfer Unit.

A too large MTU size may mean retransmissions if the packet encounters a router that cannot handle that

size of packet. A too small MTU size means relatively more header overhead and more acknowledgements

that have to be sent and handled.






1 � 4 � 25


TCP Window Size and FTP Performance [cont.]

� Maximum throughput obtained with MTU = 1500 bytes and

TCP window size = 8 (x1460 bytes).

The throughput has been calculated at the application layer.

N* is the number which permits to reach, by multiplying the MSS, a Window Size close to the maximal

allowable value of 65.535 Bytes.

The downloaded file has a size of 1 MB.

The table below illustrates the results:

1500 Bytes is the best MTU size because it permits to reach the maximum throughput value. But it is

important to note that even if the MTU size is set to 500 Bytes, the throughput can reach a high value close

to the maximum.

The asymptote characterizing the graph can be explained by the fact that the GPRS network limits the

throughput. Even if the client can receive many TCP packets without acknowledging them, the file

downloading can not be faster.

The recommended value of MTU size should be 1500 Bytes. This value is the best because the TCP window

size, which permits to reach the maximum throughput, is the smallest. In fact, with a small TCP window

size, retransmission can be avoided.

MSS Multiplying

Integer

MTU of 500 Bytes MTU of 576 Bytes MTU of1000 Bytes MTU of 1500 Bytes

4 1.391 1.628 2.95 2.91

8 2.681 2.952 4.016 4.111

16 3.812 3.877 4.045 4.113

32 3.813 3.878 4.043 4.125

N* 3.808 3.88 4.051 4.092






1 � 4 � 26









1 � 4 � 27

End of ModuleAlgorithms Dynamic Behaviors






Module 5Appendix









EVOLIUM � E-GPRS Radio Algorithms and Parameters Description B10Radio Algorithms � Appendix

1 � 5 � 2

Blank Page




Document History






1 � 5 � 3

Module Objectives


� Detect the main GPRS QoS problems of the Alcatel-Lucent BSS






1 � 5 � 4








1 � 5 � 5

Table of Contents


1 OMC-R B10 Screens Display 71.1 BSS 81.2 Cell 201.3 Adjacency 29

2 System Information Broadcasting 302.1 On BCCH 312.2 On PBCCH 332.3 On PACCH 36

3 MPDCH 374 Handovers for DTM MS 455 Detailed Available Throughput Computation 496 Examples of Puncturing Schemes 537 Extended Cell Overview 598 B10 Features 75






1 � 5 � 6








1 � 5 � 7

1 OMC-R B10 Screens Display






1 � 5 � 8


1.1 BSS B10






1 � 5 � 9


1.1 BSS [cont.] B10






1 � 5 � 10


1.1 BSS [cont.] B10






1 � 5 � 11


1.1 BSS [cont.] B10






1 � 5 � 12


1.1 BSS [cont.] B10






1 � 5 � 13


1.1 BSS [cont.] B10






1 � 5 � 14


1.1 BSS [cont.] B10






1 � 5 � 15


1.1 BSS [cont.] B10






1 � 5 � 16


1.1 BSS [cont.] B10






1 � 5 � 17


1.1 BSS [cont.] B10






1 � 5 � 18


1.1 BSS [cont.] B10






1 � 5 � 19


1.1 BSS [cont.] B10






1 � 5 � 20


1.2 Cell B10






1 � 5 � 21


1.2 Cell [cont.] B10






1 � 5 � 22








1 � 5 � 23








1 � 5 � 24








1 � 5 � 25








1 � 5 � 26








1 � 5 � 27








1 � 5 � 28








1 � 5 � 29


1.3 Adjacency B10






1 � 5 � 30

2 System Information Broadcasting






1 � 5 � 31


2.1 On BCCH

� The BCCH indicates if GPRS is supported in the cell:

� SI 3: RA_COLOUR field present if GPRS supported

� If GPRS is supported:

� SI13 is broadcast on the BCCH

� SI13 broadcast instead of retransmission of SI 1

� SI 4 content:

� SI13_PBCCH_LOCATION: gives SI 13 schedule or PBCCH location

B10

Note: do not confuse RA_COLOUR and RA Code. The former is used as a flag which has two uses for the MS

entering a new cell:

� To know if the GPRS service is supported in the cell (RA_COLOUR has a value different from -1).

� To trigger an RA update when the value of the RA_COLOUR changes. It is easy to monitor because it is

broadcast often.

The Routing Area Code is necessary for the RA update procedure (message content).

The SI13 takes the place of a few SI1 occurrences.






1 � 5 � 32


2.1 On BCCH [cont.]

� SI 13 content (non-exhaustive list):

� RA_CODE: routing area code

� NMO: Network Mode of Operation

� PAN_DEC, PAN_INC, PAN_MAX: radio link supervision

� ALPHA: uplink power control

� T_AVG_T, T_AVG_W: MS calculation of average levels

� PC_MEAS_CHAN: level measurements on BCCH / PDCH

� NETWORK_CONTROL_ORDER:

� if set to NC0 = MS controlled cell reselection, no measurement reporting

� If set to NC2 = Network controlled cell reselection, thanks to measurement reporting from the MS

� GPRS MA: for hopping SPDCH group

� ACCESS_BURST_TYPE

� DTM_SUPPORT

B10

The MS has to get SI13 information on a regular basis:

� each time the SI13 content is updated (PSI field = SI13_CHANGE_MARK set to 1).

� every 30 seconds max (even if the TBF has to be interrupted).

Through 2 different ways: SI13 on the BCCH or PSI13 in a PACCH block.

The MS has always the time to switch on PSI13 in NMOIII and/or NMOI with a Master PDCH because PBCCH

blocks are always after an I or X TS within the 52 multiframe.

Access Burst Type: it defines the access burst (8 bits or 11 bits) to be used on the PRACH, PTCCH and the

“Packet Control Ack” on a PACCH.

When the Master Channel is present in the cell, the System Information Type 13 message has different

contents from those described above. It mainly consists of:

� The radio description of the Primary Master Channel (in terms of time slot number, training sequence

code and frequency parameters).

� One GPRS Mobile Allocation (MA), if frequency hopping is used for GPRS. This is the GPRS MA of the

Primary Master Channel, if hopping. If the Primary Master Channel is not hopping, the MA corresponds to

the hopping TRX(s) used for GPRS, if any.

Three modes of cell reselection have been defined by the 3GPP Standard for GPRS MSs. These Network

Control (NC) modes, known as the NC0, NC1 and NC2, are shortly described below:

� NC0: the GPRS MS performs autonomous cell reselection without sending measurement reports to the

network.

� NC1: the GPRS MS performs autonomous cell reselection. Additionally it sends measurement reports to the

network.

� NC2: the GPRS MS shall not perform autonomous cell reselection. It sends measurement reports to the

network. The network controls the cell reselection.






1 � 5 � 33


2.2 On PBCCH

� If a primary MPDCH is available, a GPRS MS monitors the PBCCH

� PSI blocks available: PSI1, PSI2, PSI3, PSI3bis, PSI8, PSI13

� PSI1, PSI2 and PSI13 (=SI13 on BCCH) can be sent in a PACCH block for an MS in Packet Transfer Mode

� PSI1 content:

� Cell and BSS parameters

� PRACH access control parameters

� PCCCH organization parameters

� Power Control parameters

� CN features (MSC Release, SGSN Release)

Cell Parameters = NMO, MS Timers, DRX info, RLS parameters, etc.

PRACH access control parameters = access burst type, access control class, etc.

PCCCH organization parameters = BS_PBCCH_BKLS, BS_PAGCH_BLKS_RES, BS_PRACH_BLKS

The GPRS cell adjacencies are the same for an MS in Packet Idle Mode as for an MS in Packet Transfer Mode.

The GPRS cell adjacencies are equal to CS cell adjacencies.






1 � 5 � 34


2.2 On PBCCH [cont.]

� PSI2 content:

� Cell identification (PLMN Id, LAC, RAC, Cell Id)

� Non-GPRS O&M parameters (BS_PA_MFRMS, BS_AG_BLKS_RES)

� PCCCH information (TS and frequencies)

� Cell allocation information (HSN, BCCH band, frequency channels)

� PSI3 /PSI3bis content:

� BCCH allocation in neighboring cells

� Serving cell parameters (GPRS_RXLEV_ACCESS_MIN, GPRS_MS_TXPWR_MAX_CCH)

� General reselection parameters of serving and neighboring cell

� Neighbor cell parameters

PSI3, PSI3bis:

� One PSI3 instance shall be sent and, as a minimum, one PSI3bis instance shall be sent as well

� There may be up to 16 PSI3bis instances.

� Reselection parameters: C31_HYST, C32_HYST, GPRS_CELL_RESELECT_HYST, PRIORITY_CLASS, HCS_THR,

RA_RESELECT_HYSTERESIS

� Neighboring cell parameters: BSIC, BCCH frequency, SI13 PBCCH location, GPRS_RXLEV_ACCESS_MIN,

GPRS_MS_TXPWR_MAX_CCH, GPRS_TEMPORARY_OFFSET, GPRS_PENALTY_TIME,

GPRS_RESELECTION_OFFSET.

� Up to 32 neighboring cells may be defined. The field Same_RA_As_Serving_Cell provides complementary

information for the reselection process.






1 � 5 � 35


2.2 On PBCCH [cont.]

� PSI8 content:

� Optionally sent on the PBCCH

� Cell broadcast information

� CBCH channel description (TS number)

� Frequency parameters for hopping CBCH

� Frequency parameters for non-hopping CBCH (TSC, ARFCN)

TSC: the Training Sequence Code used for CBCH is the BCC value.

MAIO: Mobile Allocation Index Offset.

HSN: Hopping Sequence Number (law for frequency hopping).

MA_bitmap: MA_Bitmap is related to the BCCH band location.

MA_length: the length of the MA_bitmap, giving the number of frequency to hop on.






1 � 5 � 36


2.3 On PACCH

� PSI broadcast on PACCHs is controlled at MAC layer by the O&M parameters T_PSI_PACCH

� PSI13 content:

� GPRS cell access information (RA_CODE, NCO, ACCESS_BURST_TYPE, DTM_SUPPORT , etc.)

� Radio Link Supervision parameters, Power control information

� MS timers for TBF establishment (T3168, T3182, etc.)

B10

When there is no Master Channel, the GPRS mobiles have to read the System Information Type 13 message

at least once every 30 seconds. Because of this, GPRS mobiles in data transfer may lose data. In order to

avoid this, the Packet System Information Type 13 message is sent to each MS doing data transfer, via its

assigned downlink Packet Associated Control Channel (PACCH). This message provides the same information

as the System Information Type 13 message on BCCH. The PSI Type 13 message is not used when there is a

Master Channel.

The GPRS cell information as well as the Radio Link Supervision and the Power Control information are

similar to the one included in the SI13 on the BCCH.

T3168, T3164, T3182, T3190, T3180: refer to their use for TBF establishment and Radio Link Control (causes

of TBF releases at the MS side). We can note that the MS timer names use even numbers when the BSS one

uses odd numbers.

T_PSI_PACCH = 14s (Alcatel recommended value) but it can be set at OMC-R level.






1 � 5 � 37

3 MPDCH






1 � 5 � 38

3 MPDCH

Details on the MPDCH

� Logical channels dynamically multiplexed:

� PBCCH, PCCCH

� Identified by the PCCCH group (used for paging purposes)

� Primary Master Channel

� PBCCH carrier, indicated in SI13

� Static allocation

� Secondary Master Channels: additional MPDCH

� PRACH, PPCH and PAGCH carrier

� NB_TS_MPDCH = Number of TSs reserved for MPDCHs (Primary + Secondary)






1 � 5 � 39

3 MPDCH

Details on the MPDCH [cont.]

� Master channels configuration

� On the BCCH TRX

� Starting from the left

� BCCH and SDCCH are not overwritten

� Example

� NB_TS_MPDCH = 4






1 � 5 � 40

3 MPDCH


� MPDCH configuration

� Downlink multi-frame

� primary MPDCH: the first BS_PBCCH_BLKS blocks of the ordered list (B0, B6, B3, B9, B1, B7, B4, B10, B2, B8, B5, B11) are reserved for PBCCH

� Secondary MPDCH: the first BS_PBCCH_BLKS blocks of the ordered list above are reserved for PAGCH

� BS_PAG_BLKS_RES: number of blocks reserved for PAGCH, after reservation of PBCCH blocks

� remaining blocks are used for PPCH, PAGCH






1 � 5 � 41

3 MPDCH


� Example of DL Primary Master Channel:

� BS_PBCCH_BLKS=2

� BS_PAG_BLKS_RES=6






1 � 5 � 42

3 MPDCH


� MPDCH configuration

� Uplink multi-frame

� PRACHs occur on any PDCH carrying MPDCH

� PRACH blocks are statically allocated on the BS_PRACH_BLKS first blocks of the ordered list:

B0, B6, B3, B9, B1, B7, B4, B10, B2, B8, B5, B11

� blocks are marked by USF=free






1 � 5 � 43

3 MPDCH

Exercise

� Fill in the blocks of a DL secondary MPDCH multiframe having thefollowing criteria:

� BS_PBCCH_BLKS=2


� Fill in the blocks of a UL PDCH multi-frame having the following criterion:

� BS_PRACH_BLKS=4

Time allowed:

10 minutes






1 � 5 � 44

3 MPDCH

Exercise [cont.]

� DL multiframe: secondary MPDCH

� BS_PBCCH_BLKS=2


� UL multiframe:

� BS_PRACH_BLKS=4

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

?????

?????

?????

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

PAGCH

PAGCH

PPCH, PAGCH

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

?????

?????

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

PRACH

B0 B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11

CAUTION: animated slide






1 � 5 � 45

4 Handovers for DTM MS






1 � 5 � 46


Characteristics

� Handover should be avoided as much as possible since they interrupt the PS connection for a while.If the MS is in DTM MODE( with both CS and PS service), see next slide.

� Low priority Inter-cell HO are forbidden for MS operating in DTM

� Scenario for CS handover

� MS in DTM in the serving cell

� Handover CMD sent to the MS

� MS abnormally releases the TBF

� MS enters target cell in dedicated mode

� TBFs are re-established in the new cell in DTM

B10






1 � 5 � 47

Family Cause Inter-cell HO causeHO cause

number

Behavior when the

MS is in DTM

Emergency Inter-cell HO Too low quality on the UL cause 2 Enable

Emergency Inter-cell HO Too low level on the UL cause 3 Enable

Emergency Inter-cell HO Too low quality on the DL cause 4 Enable

Emergency Inter-cell HO Too low level on the DL cause 5 Enable

Emergency Inter-cell HO Too long MS-BS distance cause 6 Enable

Emergency Inter-cell HO Consecutive bad SACCH frames received in a microcell cause 7 Enable

Emergency Inter-cell HO Too low level on the UL, inner zone cause 10 Disable

Emergency Inter-cell HO Too low level in the DL, inner zone cause 11 Disable

Better conditions Inter-cell HO Power budget cause 12 Enable

Better conditions Inter-cell HO Too high level on the UL & DL, outer zone cause 13 Disable

Better conditions Inter-cell HO

High level in neighbor lower or indoor cell for slow

mobile cause 14 Disable

Emergency Inter-cell HO Too high interference level on the UL cause 15 Enable

Emergency Inter-cell HO Too high interference level on the DL cause 16 Enable

Emergency Inter-cell HO

Too low level on the UL in a microcell compared to a

high threshold cause 17 Enable

Emergency Inter-cell HO

Too low level on the DL in a microcell compared to a

high threshold cause 18 Enable

Forced directed retry HO Forced directed retry cause 20 Enable

Better conditions Inter-cell HO High level in neighbor cell in preferred band cause 21 Disable

Emergency Inter-cell HO Too short MS-BS distance cause 22 Enable

Better conditions Inter-cell HO Traffic HO cause 23 Disable

Better conditions Inter-cell HO General capture HO cause 24 Disable

Channel adaptation intra-cell HO HR-to-FR channel adaptation due to bad radio quality cause 26 Disable

Channel adaptation intra-cell HO FR-to-HR channel adaptation due to good radio quality cause 27 Disable

Better conditions Inter-cell HO Fast traffic HO cause 28 Disable

Resource management intra-cell HO TFO HO cause 29 Disable

Resource management intra-cell HO Move from PS to CS zone cause 30 Disable

InterRat HO 2G-3G HO cause 31 Enable


Characteristics [cont.] B10






1 � 5 � 48

MS Old BTS BSC MFS MSC

[old FACCH]

[old SACCH]

New BTS

Dedicated mode

Meas Rep. Measurement Result

HO decisionPhysical Context Req.

Physical Context Conf.

Chan. Act (new TCH)

Chan. Act ACK

DR[HO CMD]HO CMD

[new FACCH]Handover Access HO Detect

SABMEst. Ind.

UA

HO Complete DI[HO CMP]Handover Performed

BSC Shared DTM Info Ind (new cell)

BSC Shared DTM Info Ind (old cell)


Handover Scenario B10






1 � 5 � 49

5 Detailed Available Throughput Computation






1 � 5 � 50


Formula

� available_throughput_candidate_XL is � the overall throughput provided by its PDCHs

� It depends on:

� the potential throughput of its PDCHs� potential_throughput_PDCH

� the available capacity on each of its PDCHs

� available_capacity_PDCH_XL

� For a GPRS TBF� Potential_throughput_PDCH = R_AVERAGE_GPRS

� Available_capacity_PDCH_XL = (1 – USED_CAPACITY_GBR_XL * (1 + QOS_SAFETY_MARGIN/100)) / (Nb_BE_TBF_HIGHER_PRIOR_XL*SCHEDULING_PRIORITY_FACTOR + Nb_BE_TBF_SAME_PRIOR_XL + 1)

� Same formulas are used for an EGPRS TBF with R_AVERAGE_EGPRS and only the EGPRS TBFs are considered






1 � 5 � 51


Formula [cont.]

� USED_CAPACITY_GBR_XL� total PDCH capacity that has already been allocated to RT PFCs (both GPRS and EGPRS) on the PDCH in the XL direction

� RT PFCs: Real Time Packet Flow Context

� Nb_BE_TBF_HIGHER_PRIOR_XL (respectively Nb_BE_TBF_SAME_PRIOR_XL) � total number of Best Effort TBFs (GPRS or EGPRS)

� which have some radio resources allocated on the considered PDCH in the XL direction

� and whose priority (combination of THP and Precedence) is strictly higher than (respectively equal to) the priority of the TBF to establish / reallocate

� THP: Traffic Handling Priority� QoS parameter used for the interactive traffic class






1 � 5 � 52


Formula [cont.]

� For a given candidate time slot allocation with n PDCHs

� available_capacity_candidate_XL = ∑i=1 to n available_capacity_PDCHi_XL

� available_throughput_candidate_XL = potential_throughput_PDCH * available_capacity_candidate_XL


� available_throughput_candidate_XL = R_AVERAGE_GPRS *


� available_ throughput _candidate_XL = R_AVERAGE_EGPRS *

∑= +

n

1i PDCHi 1NB_TBF

1

∑= +

n

1i PDCHi 1NB_TBF

1

Back

NB_TBFPDCHi represents the number of already allocated GPRS and EGPRS TBFs on the PDCH i, in case of

GPRS allocation.

NB_TBFPDCHi represents the number of already allocated EGPRS TBFs on the PDCH i, in case of EGPRS

allocation.






1 � 5 � 53

6 Examples of Puncturing Schemes






1 � 5 � 54


Puncturing Schemes and MCS

� The 05.03 GSM recommendation (Channel coding) indicates for eachMCS the available puncturing schemes

PS1, PS2, PS3MCS-9

PS1, PS2, PS3MCS-8

PS1, PS2, PS3MCS-7

PS1, PS2MCS-6

PS1, PS2MCS-5

PS1, PS2, PS3MCS-4

PS1, PS2, PS3MCS-3

PS1, PS2MCS-2

PS1, PS2MCS-1

PS of last transmission before MCS switch

PS of first transmission after MCS switch

MCS switched from

MCS switched to

The puncturing process consists of transmitting only some of the coded bits obtained after the rate 1/3

convolutional coding. Depending on the considered puncturing scheme, different coded bits are

transmitted. Therefore, when the receiver receives two versions of the same RLC block sent with two

different puncturing schemes, it obtains additional information leading to an increased decoding

probability.






1 � 5 � 55

P2 P3 P1 P2

puncturing puncturing

1404 bits

USF RLC/MAC

Hdr.

36 bits

Rate 1/3 convolutional coding

135 bits

468 bits

612 bits 124 bits 36 bits SB = 8

1392 bits

45 bits

Data = 448 bits BCS TB

468 bits

612 bits 612 bits

1404 bits


FBI E Data = 448 bits BCS TB FBI E

612 bits 612 bits 612 bits

P3 P1

3 bits

HCS

puncturing


Coding and Puncturing for MCS-7

� Coding and puncturing for MCS-7; rate 0.76 8PSK, two RLC blocks per 20ms

For MCS-7, the data part of the RLC block contains 468 bits. The rate 1/3 convolutional coding gives 1404

bits: C(0), C(1), ..., C(1403). The code is then punctured depending on the value of the CPS field as defined

in the 04.60 GSM recommendation. Three puncturing schemes named P1, P2 and P3 are applied in such a

way that the following coded bits are transmitted:

The result is a block of 612 (8*78 - 12) coded bits.

P1 {C(18j), C(1+18j), C(4+18j), C(8+18j), C(11+18j), C(12+18j), C(13+18j), C(15+18j)

for j = 0,1,...,77} are transmitted

except {C(k) for k = 1,19,37,235,415,595,775,955,1135,1351,1369,1387} which are not transmitted

P2 {C(2+18j), C(3+18j), C(5+18j), C(6+18j), C(10+18j), C(14+18j), C(16+18j), C(17+18j)


except {C(k) for k = 16,34,52,196,376,556,736,916,1096,1366,1384,1402} which are not transmitted

P3 {C(2+18j), C(5+18j), C(6+18j), C(7+18j), C(9+18j), C(12+18j), C(13+18j), C(16+18j)


except {C(k) for k = 13,31,49,301,481,661,841,1021,1201,1363,1381,1399} which are not

transmitted






1 � 5 � 56



� Coding and puncturing for MCS-5; rate 0.37 8PSK, one RLC block per 20 ms

P2 P1

puncturing

1404 bits

USF RLC/MAC

Hdr. Data = 56 octets = 448 bits BCS

36 bits


99 bits

468 bits


1392 bits

33 bits

TB E FBI HCS

3 bits

1248 bits

+1 bit






1 � 5 � 57

P1 P3 P2

puncturing

1116 bits

USF RLC/MAC

Hdr. Data = 44 octets = 352 bits BCS

12 bits


108 bits

372 bits


464 bits

36 bits

TB E FBI HCS

3 bits

372 bits 372 bits

puncturing



� Coding and puncturing for MCS-4; uncoded GMSK, one RLC block per 20 ms






1 � 5 � 58

P1 P2

puncturing

588 bits

USF RLC/MAC

Hdr. Data = 22 octets = 176 bits TB

12 bits


108 bits

196 bits


464 bits

36 bits

BCS

puncturing

HCS E FBI

3 bits

372 bits



� Coding and puncturing for MCS-1; rate 0.53 GMSK, one RLC block per 20 ms

Back

For MCS-1, the data part of the RLC block contains 196 bits. The rate 1/3 convolutional coding gives 588

bits: C(0), C(1), ..., C(587). The code is then punctured depending on the value of the CPS (Coding and

Puncturing Scheme indicator) field (EGPRS RLC/MAC header) as defined in the 04.60 GSM recommendation

(RLC/MAC). Two puncturing schemes named P1 and P2 are applied in such a way that the following coded

bits are not transmitted:

The result is a block of 372 (588 - 8*28 + 8) coded bits.

P1 {C(2+21j), C(5+21j), C(8+21j), C(10+21j), C(11+21j), C(14+21j), C(17+21j), C(20+21j) for j =

0,1,...,27} are not transmitted except {C(k) for k = 73,136,199,262,325,388,451,514} which are

transmitted

P2 {C(1+21j), C(4+21j), C(7+21j), C(9+21j), C(13+21j), C(15+21j), C(16+21j), C(19+21j) for j =

0,1,...,27} are not transmitted except {C(k) for k = 78,141,204,267,330,393,456,519} which are

transmitted






1 � 5 � 59

7 Extended Cell Overview






1 � 5 � 60

� One BTS (G3 or G4): 2 cells

� INNER cell: range from 0 to 35 km

� OUTER cell: range from 33 to 70 km


Presentation - General

The extended cell has up to 4 TRX in the inner cell and up to 4 TRX in the outer cell.






1 � 5 � 61


Presentation - Synchronization

OUTER cellINNER cell

� Freq BCCH OUTER <> Freq BCCH INNER

� MS reports measurements on both cells for the handover algorithms

� BSICINNER = BSICOUTER

� INNER cell can decode the RACH received on OUTER BCCH frequency

� INNER cell always BARRED

� MS always camps on OUTER cell

At the border of the two cells, an overlapping area allows to provide a continuous coverage. When the MS

moves from one cell to the other, a handover is triggered in the overlap zone. Two BCCH channels are

needed (one for the inner cell, one for the outer cell), so that the MS reports measurements on both cells

for the handover algorithms.

The TRXs of the inner cell and of the outer cell are synchronised, but the reception of the outer cell is

delayed by 60bits period to account for the propagation delay.

In the inner cell, the MS can receive the BCCH inner frequency as wells as the outer BCCH frequency. To

avoid to manage RACH reception on two different frequencies in the inner cell, the MS is forced to access

the inner cell on the outer BCCH frequency. For this purpose, the RACH reception (BCCH TRX) of the inner

cell is tuned to the outer BCCH frequency, and the inner cell is barred1. So on time slot 0 of the inner cell,

transmission is done on the inner cell BCCH frequency, and reception is done on outer BCCH frequency.

The chosen implementation allows to make use of all timeslots2 of the TDMA frame and to use the

combined configuration for the CCCH channel.






1 � 5 � 62

� UL interference on TS0 of the INNER cell if

� Access burst received in the INNER cell (on frequency BCCH OUTER)

AND

� Call on TS7 of the OUTER cell

� Then, TS7 of the OUTER cell is always set to IDLE (never used)


Presentation - RF Interference






1 � 5 � 63


Radio Link Establishment - MS Located in the Outer Cell Area

The inner cell is always barred, so the MS cannot camp on the inner cell, even if located in the inner cell range. In the

whole extended cell coverage, the MS has a good reception of the outer cell BCCH, so the MS will always be camping in

the outer cell, whether in the inner cell or outer cell range.

For this reason, a special radio and link establishment procedure is used to cope with this behaviour .

It consists of receiving the CHANNEL REQUEST messages on outer BCCH frequency, and allocating the SDCCH channel

according to the MS estimated position. The IMMEDIATE ASSIGNMENT COMMAND for an SDCCH is sent on the outer cell

BCCH frequencies, but the SDCCH may be allocated in either inner or outer cell, depending on the MS position.

(1) The MS camping on the outer cell sends an access burst on the RACH on outer cell BCCH frequency. These bursts will

be received successfully in the inner cell by the BCCH TRE. In the outer cell, the access burst arrives too early and

cannot be decoded.

(2) The inner cell BCCH TRE sends a CHANNEL REQUIRED message to the BSC containing the random reference sent by

the mobile, the TDMA frame number when the message was sent over the air and the measured TOA.

(3) The TCU controlling this TRE allocates an SDCCH subchannel to the transaction in the inner cell and asks the BTS to

activate this subchannel.

(4) The BTS activates the requested channel and sends back and acknowledgement, once this is done.

(5) The TCU sends the IMMEDIATE ASSIGNMENT COMMAND (which provides the description of the allocated SDCCH) to

the BCCH TRE of the inner cell.

The TCU controlling the inner cell BCCH sends a copy of the message to the TCU handling the BCCH of the outer cell.

This is done if and only if the timing advance IE included in the CHANNEL REQUIRED is smaller than 60, thus indicating

that the MS is strictly in the inner cell (in order to avoid that the MS receives two Immediate Assignment messages when

located in the overlap zone).

The TCU controlling the outer cell BCCH forwards the IMMEDIATE ASSIGNMENT COMMAND to the outer cell BCCH TRE.

(6) The IMMEDIATE ASSIGNMENT message is sent over the air to the MS on the AGCH of the outer cell.

(6') The IMMEDIATE ASSIGNMENT message sent by the inner cell is lost, because the MS listens to the outer cell

frequency.

(7) The mobile switches its transceiver to the SDCCH allocated in the inner cell and sends repeatedly an SABM frame to

establish the layer 2 connection with the BTS.

(8) The BTS acknowledges the establishment of the LapDm link to the MS with a UA frame sent on the SDCCH allocated

to the MS.






1 � 5 � 64

If TA < 60


Radio Link Establishment - MS Located in the Inner Cell Area

The TCU sends the IMMEDIATE ASSIGNMENT COMMAND (which provides the description of the allocated

SDCCH ) to the BCCH TRE of the inner cell.

The TCU controlling the inner cell BCCH sends a copy of the message to the TCU handling the BCCH of the

outer cell. This is done if and only if the timing advance IE included in the CHANNEL REQUIRED is smaller

than 60, thus indicating that the MS is strictly in the inner cell (in order to avoid that the MS receives two

Immediate Assignment messages when located in the overlap zone).

The TCU controlling the outer cell BCCH forwards the IMMEDIATE ASSIGNMENT COMMAND to the outer cell

BCCH TRE.

(1) The MS in the outer cell sends an access burst on the RACH of the outer cell. This burst is successfully

received by the outer cell BCCH TRE. In the inner cell, the access burst arrives too late to be successfully

decoded.

(2) The outer cell BCCH TRE sends a CHANNEL REQUIRED message to the BSC containing the random

reference sent by the mobile, the TDMA frame number when the message was sent over the air and the

measured TOA.

(3) The TCU controlling this TRE allocates an SDCCH subchannel in the outer cell to the transaction and asks

the BTS to activate this subchannel.

(4) The BTS activates the requested channel and sends back an acknowledgement, once this is done.

(5) The TCU then sends the description of the channel in the IMMEDIATE ASSIGNMENT COMMAND to the

outer cell BCCH TRE.

(6) The IMMEDIATE ASSIGNMENT message is sent over the air to the MS on the AGCH of the outer cell.

(7) The mobile switches its transceiver to the required channel and sends repeatedly an SABM frame to

establish the layer 2 connection with the BTS.

(8) The BTS acknowledges the establishment of the LAPDm link to the MS with a UA frame sent on the

SDCCH allocated to the MS.






1 � 5 � 65


Radio Link Establishment - MS Located in the Overlap Zone






1 � 5 � 66

�


Radio Link Establishment - MS Located in the Overlap Zone [cont.]

(1a&b) The MS camping on the outer cell sends an access burst on the RACH. This burst is correctly received by the inner cell BCCH TRE

and outer cell BCCH TRE.

(2a&b) The inner cell and outer cell BCCH TRE send a CHANNEL REQUIRED message to the BSC containing the random reference sent by

the mobile, the TDMA frame number when the message was sent over the air and the measured TOA.

(3a&b) Both TCUs controlling the TREs having BCCH allocate an SDCCH subchannel to the transaction and ask the BTS to activate this

subchannel.

(4a&b) The BTS activates the requested channels and sends back an acknowledgement for each, once this is done.

(5b) The TCU controlling the outer cell, sends the IMMEDIATE ASSIGNMENT COMMAND with SDCCH description in the outer cell to the

outer cell BCCH TRE.

(5a&c)The TCU controlling the inner cell sends in the IMMEDIATE ASSIGNMENT COMMAND with SDCCH description in the inner cell. Two

cases are possible:

� Access Delay IE > 59 the inner cell TCU will not send a copy of the IMMEDIATE ASSIGNMENT command to the outer cell TCU. This is

the desired behaviour.

� Access Delay in [58,59] range, the inner cell TCU sends a copy of the IMMEDIATE ASSIGNMENT command to the outer cell TCU. This is

not the desired behaviour (corresponds to inner cell scenario). This is due to the fact that the BSC definition of the overlap zone does

not match the exact BTS overlap area (negative values of TOA in the outer cell up to –2, are clipped to 0).

(6b) The IMMEDIATE ASSIGNMENT message describing the SDCCH allocation in outer cell, is sent to the MS on the outer cell BCCH

frequency. In most cases this message should be received by the MS (except if 6c is received first)

(6a) The IMMEDIATE ASSIGNMENT message describing the SDCCH allocation in inner cell is lost on the inner cell air interface, because

the MS does not listen to that frequency. The unused SDCCH will be released by the BSC when the supervising timer expires6.

(6c) Access Delay in [58,59] range: The IMMEDIATE ASSIGNMENT message describing the SDCCH allocation in inner cell is sent on the

BCCH frequency of the outer cell. In most cases, the MS should have received message (6b) before and has already switched to the

SDCCH in the outer cell, and so this message is lost. It is however possible, in case the message (6b) is delayed in the inner cell, that

the message (6c) is received earlier by the MS. In this case establishment will occur on the SDCCH allocated in the inner cell (not

drawn).

(7b) The mobile receives the IMMEDIATE ASSIGNEMENT describing the SDCCH allocation in outer cell on the BCCH outer cell frequency.

It then switches to the designated channel and sends repeatedly an SABM frame to establish the layer 2 connection with the BTS in the

outer cell. If the message (6c) is received before (6b), then the establishment will occur in the inner cell.

(8b) The BTS acknowledges the establishment of the LapDm link to the MS with a UA frame sent on the SDCCH allocated to the MS.

(9) The unused SDCCH is released in the inner cell (double SDCCH allocation). If message 6c arrives first, then the unused SDCCH

release will occur in the outer cell.






1 � 5 � 67

� CAUSE 6: Too long distance

AV_RANGE_HO > U_TIME_ADVANCE

and EN_DIST_HO = ENABLE

� U_TIME_ADVANCE = 62

� EN_PBGT_FILTERING = Disable


Handover - from the INNER cell to the OUTER Cell

In the extended cell , the handover procedure is purely controlled by settings of the handover detection

parameters. Two special causes allow handover from the inner cell to the outer cell and handover from the

outer cell to the inner cell. There is no change in the BSC handover algorithm either for handover

preparation or execution.

From the inner cell to the outer cell , the handover alarm is only triggered by the handover cause “too long

MS-BS distance”. When this cause is triggered the extended outer cell is always a candidate cell.

However the operator setting of the handover parameters must insure that this cause is only triggered when

the distance from the serving inner cell BTS is greater than the limit of the overlap zone (TA > 62) by

setting U_TIME_ADVANCE to 62.

In order to avoid the extended outer cell to be filtered by the filtering process the flag

EN_PBGT_FILTERING must be set to DISABLE.

The candidate cell evaluation process is recommended to be the GRADE mode.






1 � 5 � 68

� CAUSE 22: Too short distance

AV_RANGE_HO < L_TIME_ADVANCE

� L_TIME_ADVANCE = 0

� EN_PBGT_FILTERING = Disable

� Cause 22 is only checked if

� Cell_range(serving) = extended_outer


Handover - from the OUTER cell to the INNER Cell

In the same way, from the outer cell to the inner cell , the handover alarm is only triggered by the

handover cause “too short MS-BS distance”. When this cause is triggered the extended inner cell is always a

candidate cell.

However the operator setting of the handover parameters must insure that this cause is only triggered when

the timing advance applied by the mobile reaches 0, this is achieved by setting L_TIME_ADVANCE to 0.

In order to avoid the extended inner cell to be filtered by the filtering process the flag

EN_PBGT_FILTERING must be set to DISABLE.

The candidate cell evaluation process is recommended to be the GRADE mode.






1 � 5 � 69

� All the standard HO causes can be used

� Emergency HO causes 2, 3, 4, 5

� Better condition HO causes 12, 23, 24

� The OUTER or INNER cell is always present in the Candidate CellEvaluation


Handover - from the OUTER or INNER Cell towards Another Cell

The setting of the handover parameter does not prevent any handover cause to trigger an alarm for a

handover towards a third cell.

It is possible to use exactly the same rules and parameters for handover towards a third cell as in the macro

cellular normal cases.

The synchronous handover does not work between the inner and the outer cell.

In order to avoid call terminations due to directed retry into the inner or outer cell with an incorrect

distance range it is recommended to disable the forced directed retry towards the inner and the outer cell.

For this purpose, the parameter FREELEVEL_DR(n) is set to the maximum value (255) for the inner and the

outer cell.

But the Normal DR can be activated.






1 � 5 � 70

� The Inner Cell shall always be BARRED

� If combined CCCH/SDCCH is used in the inner extended cell, then the same configuration is required in outer extended cell, and vice-versa (ie same in both cells)

� BSICINNER = BSICOUTER

� The TS 7 of BCCH TRX of outer cell must be set to IDLE

� The INNER cell and OUTER cell must belong to the same location area

� Synchronous handover must be disabled.

� U_TIME_ADVANCE = 62

� L_TIME_ADVANCE = 0

� EN_PBGT_FILTERING = DISABLE.

� CELL_EV = “Grade”

� FREELEVEL_DR(n) = 255 (this is done automatically, at configuration time)

� INNER cell and OUTER cell must be neighbour, handover relationship must exist in both directions


CS Parameters Setting






1 � 5 � 71


Packet Service

� Activation of the PS service in an Extended cell

� No specific parameter is foreseen

� Same procedure as the one used for standard cell is applied� TRX_PREF_MARK = 0

� If used, PS must be activated in both INNER and OUTER cell

� Reselection

� Because the INNER cell is barred

� this cell should must not be declared in the neighbor cells reselection adjacencies

� NC2 is not allowed

� NACC and (P)SI STATUS are not allowed

� The Master channel is not allowed in both INNER and OUTER cell






1 � 5 � 72


Packet Service [cont.]

� Packet access procedure (1/2)

� Same principle as in CS, since it’s performed on CCCH only

� The MS always performs its access on the RACH of the outer BCCH frequency

� The BTS provides the BSC with the initial TA

� Depending on the TA value, the BSC chooses the suitable cell (INNER or OUTER)

� In UL, whatever the multislot class of the MS, only one PDCH is allocated

� Its right or left TS can not be allocated neither for PS nor for CS (see comment)

� This TS is considered as a restricted TS by the MSF

� The same constraint is applied in DL for the TS carrying the PACCH

UL

Restricted

Allocated

Restricted

Allocated

INNEROUTER

When a MS passes from inner/outer cell to outer/inner cell, the TA estimated by the BTS stalls

progressively. So the MS is not able to apply the suitable correction of its TA for its uplink transfer (data

and/or signaling). This leads progressively to the impossibility for the BTS to decode the uplink radio blocks

because they shift out of their allocated RTS.

For a given MS, its uplink radio blocks progressively come out of its allocated RTS and jams the neighbor

RTS.

� It jams the right RTS when the MS moves from inner to outer cell. This right RTS can also be the RTS0 of

the next TDMA frame if the RTS7 is allocated to a TBF.

� It jams the left RTS when the MS moves from outer to inner cell. This left RTS can also be the TS7 of the

previous TDMA frame if the RTS0 is allocated to a TBF.

If the neighboring RTS is dedicated to other MS for PS or CS call, this jam causes interferences on these RTS

and the BTS can not decode the radio blocks of those MS leading to the drop of these calls.

This drawback only occurs for the uplink direction. The downlink direction does not raise any problem.

To overcome this drawback, some radio resource allocation constraints are to be applied:

� An UL TBF is only allocated on one RTS.

� On BCCH or non BCCH inner TRX,

� A RTS is allocable to a UL TBF if its right RTS is allocated for PS traffic to the MFS, and is not used by

a UL TBF.

� When a RTS is allocated, its right RTS cannot be allocated to PS call.

� On BCCH or non BCCH outer TRX,

� A RTS is allocable to a UL TBF if its left RTS is allocated for PS traffic to the MFS, and is not used by

a UL TBF.

� When a RTS is allocated, its left RTS cannot be allocated to PS call.






1 � 5 � 73


Packet Service [cont.]

� Miscellaneous

� In the OUTER cell, the maximum MCS is limited to MCS-4

� The Streaming TBFs (i.e. RT PFC) are not supported

� The INNER and OUTER cells must be mapped on the same GPU

� The INNER and OUTER cell must belong to the same routing area






1 � 5 � 74


PS Parameters Setting

� NETWORK_CONTROL_ORDER = NC0

� EN_NACC = Disable

� EN_PSI_STATUS = Disable

� NB_TS_MPDCH= Disable

� MAX_PDCH, MAX_PDCH_HIGH_LOAD and MIN_PDCH must be set to even values (see comments)

� EN_STREAMING = Disable

As in UL TBF allocation, the MFS uses at least 2 TS (a “restricted” one and the one allocated in UL) the

number of PDCH allocable in the extended cells (MAX_PDCH, MIN_PDCH, MAX_PDCH_HIGH_LOAD ) must be

even.






1 � 5 � 75

8 B10 Features






1 � 5 � 76

8 B10 Features

Summary

� B10 MR1

� All B9 MR4 Features

� Dual Transfer Mode (simultaneous CS and PS traffic)

� GTTP usage

� Improved delayed UL TBF Release

� B10 MR2

� Extended Dynamic Allocation (more than 2 TS in UL)

B10






1 � 5 � 77









1 � 5 � 78

End of ModuleAppendix

@@SECTIONTITLE - @@MODULETITLE

All Rights Reserved © Alcatel-Lucent 2007@@PRODUCT

@@COURSENAME

@@SECTION - @@MODULE - 1

D R A F T



Copyright © 2007 by Alcatel-Lucent - All rights reservedPassing on and copying of this document, use and communication of its contents not permitted without written authorization from Alcatel

EGPRS Radio Algorithms & Parameters Description B10.pdf

Documents

b10 screens display

rights reserved passing

parameters description b10

rights reserved alcatel

t round trip delay t fast dl margin

gprs radio algorithms

sampling instant tk

left blank intentionally