edge multicarrier

Nokia Research Center Radio Communications Laboratory

Chalmers University of Technology Department of Signals and Systems

Higher Data Rates in GSM/EDGE with

Multicarrier

Written by Frédéric Noël

Master Thesis from September 2000 – April 2001

Supervised by Eero Nikula

Technical Report EX024/2001 Dept. Signals and Systems, Chalmers University of Technology

Higher Data Rates in GSM/EDGE with Multicarrier

17-Sep-01 2/61


17-Sep-01 3/61

Master Thesis supervisors: Eero Nikula Research Manager Radio Communications Laboratory NOKIA Research Center P.O.Box 407 FIN-00045 Nokia Group Finland Email: [email protected] Tony Ottosson Associate Professor Communication Systems Group Department of Signals and Systems Chalmers University of Technology S-412 96 Göteborg Sweden Email: [email protected] Author: Frédéric Noël Assistant Research Engineer Radio Communications Laboratory NOKIA Research Center P.O.Box 407 FIN-00045 Nokia Group Finland Email: [email protected] Master of Science in Digital Communication Systems and Technology, Chalmers University of Technology French Engineer Diploma from Institut Superieur d'Electronique de Paris


17-Sep-01 4/61

ABSTRACT

One of the potential future development areas in the GSM/EDGE standard is the increase of user peak transfer rates. For a wide area coverage system like GSM/EDGE, the design of higher data rate scheme can be based on the transmission on multiple carriers, instead of one single carrier. This work evaluates the feasibility of a multicarrier concept in GSM/EDGE and aims at minimizing modifications to current specifications. At first, through a presentation of GSM/EDGE evolution towards 3G, key definitions and concepts are defined. Then, based on relevant assumptions and constraints, the multicarrier concept is studied from a physical layer perspective. Multiple carrier reception and signal bandwidth are analyzed in order to determine the multicarrier capability of MSs. Furthermore, the study shows that frequency hopping is not easily applicable to multicarrier MSs with limited bandwidth. Although interleaving data over several carriers might provide diversity gain, it is found that link adaptation and power control constraints would reduce performance and introduce important modification in the specification. Moreover, the enhancements of radio protocols for supporting multicarrier are evaluated. Based on the physical layer study, multicarrier classes are defined. A new three-phase access procedure is also designed for allocating multicarrier resources, and generally for signaling future MS capabilities. At last, the study shows that, radio bearer data should preferably be (de)multiplexed in the PDCP layer over (from) the different carriers.

KEYWORDS

3GPP, allocation, assignment, EDGE, EGPRS, frequency hopping, GERAN, GSM, inter-frequency interleaving, MAC, multicarrier, multislot, radio access, radio resource, RLC, temporary block flow.


17-Sep-01 5/61

ACKNOWLEDGMENT

Helsinki, April 2001 This Master Thesis was carried out in the Radio Communications Laboratory at Nokia Research Center, Helsinki as part of the GSM evolution research.

I am grateful to both Mr Markus HAKASTE, R&D Manager and Dr Pekka SOININEN, Head of the Radio Communications Laboratory for the opportunity to work and complete my studies here.

I wish to express my gratitude to Mr. Eero NIKULA, the supervisor of my work, for his interest, valuable comments and guidance. I am also indebted to Mr. Guillaume SÉBIRE, for the time devoted to my work, his accurate answers and for his help. My thanks go also to all the members of the team in which I was involved, and particularly to Mr. Thierry BELLIER and Mr. Benoist SÉBIRE, who helped me with their advise. I am also thankful to Ms. Jaana KOSKELA and Ms. Kirsti GESTERBERG, the secretaries of the Laboratory, for their help.

I wish also to express my gratitude to Ms. Katharina REISSING, Nokia Student Exchange Coordinator, for the activities with other trainees for discovering northern countries.

Finally, I wish to thank my parents and my girlfriend for their support.

Kiitos paljon! Tack så myket!

Frédéric NOËL


17-Sep-01 6/61

TABLE OF CONTENTS

SYMBOLS .................................................................................................................................8

ABREVIATIONS......................................................................................................................8

1 INTRODUCTION...........................................................................................................10

2 PACKET DATA TRANSFER .......................................................................................11 2.1 PACKET TRANSMISSION PRINCIPLES ...........................................................................11

2.1.1 From data to radio waves........................................................................................11 2.1.2 Packet switching .....................................................................................................11 2.1.3 Multiplexing techniques..........................................................................................11

2.2 GENERAL PACKET RADIO SERVICE (GPRS) & ENHANCED GPRS (EGPRS) ..........12 2.2.1 Network architecture...............................................................................................12

2.2.1.1 GPRS network architecture................................................................................. 12 2.2.1.2 EGPRS Network Architecture............................................................................. 13

2.2.2 Protocol architecture ...............................................................................................14 2.2.2.1 Medium Access Control protocol........................................................................ 15 2.2.2.2 Radio Link Control protocol............................................................................... 16

2.2.3 Radio interface ........................................................................................................17 2.2.3.1 Radio Access Capability ..................................................................................... 17 2.2.3.2 GPRS Radio Interface ........................................................................................ 17 2.2.3.3 EGPRS Radio Interface ...................................................................................... 18 2.2.3.4 GSM/EDGE main characteristics ........................................................................ 20

2.3 GSM/EDGE RADIO ACCESS NETWORK (GERAN) ...................................................20 2.3.1 3G mobile services..................................................................................................20 2.3.2 Network Architecture..............................................................................................21 2.3.3 Protocol architecture ...............................................................................................21

2.3.3.1 Radio Resource Control protocol......................................................................... 22 2.3.3.2 Packet Data Convergence Protocol...................................................................... 23

3 MULTICARRIER FEASIBILITY STUDY.................................................................24 3.1 SCOPE OF THE STUDY....................................................................................................24 3.2 RADIO PROTOCOL ELEMENTS AFFECTED .....................................................................24

4 PHYSICAL LAYER .......................................................................................................25 4.1 RADIO FREQUENCY ......................................................................................................25

4.1.1 Receiver model .......................................................................................................25 4.1.2 Frequency offset between carriers ..........................................................................26 4.1.3 Bandwidth of the multicarrier allocation................................................................27

4.1.3.1 Bandwidth limitation .......................................................................................... 27 4.1.3.2 Bandwidth occupation ........................................................................................ 27

4.2 PHYSICAL CHANNEL ALLOCATION ...............................................................................28 4.3 CHANNEL CODING AND INTERLEAVING .......................................................................30

4.3.1 Logical channels .....................................................................................................30 4.3.2 Channel coding .......................................................................................................30 4.3.3 Interleaving .............................................................................................................30

4.4 FREQUENCY HOPPING...................................................................................................31 4.5 PERFORMANCE REQUIREMENTS AND RESULTS............................................................32


17-Sep-01 7/61

4.5.1 RF performance.......................................................................................................32 4.5.2 Link level performance ...........................................................................................32 4.5.3 Radio performance ..................................................................................................33

5 RADIO PROTOCOLS FOR MULTICARRIER.........................................................34 5.1 MULTICARRIER CAPABILITY .......................................................................................34

5.1.1 MS multicarrier capability ......................................................................................34 5.1.2 Network multicarrier capability..............................................................................34

5.2 PROTOCOL ARCHITECTURE SCENARIOS .....................................................................35 5.2.1 Case 1: Multicarrier with global RLC and MAC entities .......................................35 5.2.2 Case 2: Multicarrier with carrier dedicated RLC and MAC entities ......................36

5.3 MEDIUM ACCESS CONTROL (MAC) ............................................................................37 5.3.1 Multiple carrier assignment ....................................................................................37 5.3.2 PACCH operations ..................................................................................................37 5.3.3 Temporary Flow......................................................................................................37 5.3.4 Multicarrier TBF establishment ..............................................................................38

5.3.4.1 TBF establishment initiated by the Mobile Station................................................ 38 5.3.4.2 TBF establishment initiated by the Network......................................................... 40

5.3.5 Assignment procedures in RRC Cell_shared mode for multicarrier [14]..............40 5.3.5.1 Carrier assignment to an existing block flow ........................................................ 41 5.3.5.2 Carrier reassignment .......................................................................................... 41 5.3.5.3 Carrier release.................................................................................................... 41

5.3.6 Link quality measurements .....................................................................................41 5.4 RADIO LINK CONTROL (RLC).....................................................................................42

5.4.1 Link Adaptation ......................................................................................................42 5.4.2 Block Sequence Number and RLC Window Size ..................................................42 5.4.3 TBF Release ............................................................................................................43 5.4.4 RLC/MAC control messages ..................................................................................43

5.5 PACKET DATA CONTROL PROTOCOL (PDCP) ............................................................43 5.6 RADIO RESOURCE CONTROL (RRC) ...........................................................................44

6 SUMMARY OF RADIO PROTOCOLS CHANGES..................................................45

7 CONCLUSION................................................................................................................46

8 REFERENCES ................................................................................................................47

ANNEX A : GSM/EDGE ACCESS METHODS ..................................................................48 A.1. UPLINK ACCESS FOR MS IN PACKET IDLE MODE........................................................48 A.2. DOWNLINK ACCESS......................................................................................................49

ANNEX B : MULTISLOT CAPABILITY............................................................................51 B.1. MS CLASSES FOR MULTISLOT CAPABILITY..................................................................51 B.2. NETWORK REQUIREMENTS FOR SUPPORTING MS MULTISLOT CLASSES ....................52

ANNEX C : EGPRS RLC WINDOW SIZE ..........................................................................55

ANNEX D : MODIFIED PACKET DOWNLINK ASSIGNMENT....................................57


17-Sep-01 8/61

SYMBOLS

A Interface between BSC and 2G MSC Gb Interface between BSC and 2G SGSN Iu-cs Interface between BSS and 3G MSC Iu-ps Interface between BSS and 3G SGSN Iur-g Interface between different BSS Um Interface between MS and BSS

ABREVIATIONS

2G 2nd Generation 3G 3rd Generation 3GPP 3rd Generation Partnership Project 8-PSK Octal Phase Shift Keying ARQ Automatic Repeat reQuest BEP Bit Error Probability BLER Block Error Rate BSS Base Station Subsystem BSSGP Base Station System GPRS Protocol BTS Base Transceiver Station CN Core Network CS Coding Scheme Dm Control Channel (ISDN terminology applied to mobile service) ECSD Enhanced Circuit Switched Data EDGE Enhanced Data rates for Global Evolution EGPRS Enhanced General Packet Radio Service ETSI European Telecommunications Standards Institute GERAN GSM/EDGE Radio Access Network GGSN Gateway GPRS Support Node GMSK Gaussian Minimum Shift Keying GPRS General Packet Radio Service GSM Global System for Mobile communications HIIARQ Hybrid type II ARQ IP Internet Protocol IR Incremental Redundancy LA Link Adaptation LAN Local Area Network LAPDm Link Access Protocol on the Dm channel LLC Logical Link Control MAC Medium Access Control MCS Modulation and Coding Scheme MS Mobile Station NSS Network and Switching Subsystem OSI Open System Interconnect PACCH Packet Associated Control CHannel PDA Packet Downlink Assignment PDCH Packet Data CHannel PDTCH Packet Data Traffic CHannel PDCP Packet Data Convergence Protocol


17-Sep-01 9/61

PDU Packet Data Unit PLMN Public Land Mobile Network PSTN Public Switched Telephone Network PTCCH Packet Timing advance Control CHannel PTR Packet Timeslot Reconfigure PUA Packet Uplink Assignment QoS Quality of Service RAB Radio Access Bearer RB Radio Bearer RLC Radio Link Control RR Radio Resource RR Radio Resource Control RTP Real Time Protocol SAP Service Access Point SDU Service Data Unit SGSN Serving GPRS Support Node SMS Short Message Service SNDCP Sub-Network Dependent Convergence Protocol SPSCH Shared Physical Sub Channel TBF Temporary Block Flow TCH Traffic Channel TCP Transmission Control Protocol TDMA Time Division Multiple Access TFI Temporary Flow Identifier TS Timeslot UDP User Datagram Protocol UMTS Universal Mobile Telecommunications System USF Uplink State Flag UTRAN UMTS Terrestrial Radio Access Network WCDMA Wideband Code Division Multiple Access WS Window Size


17-Sep-01 10/61

1 INTRODUCTION

Over the past decade, the advent of mobile communications, laptops, and internet, have stirred up the society. Habits have been modified by the addition of mobility to usual devices. A greedy appetite of fast information and service has appeared. Those changes pave the way to a mobile information society. The increase of data rates will continue this evolution. Starting in a new era, high-speed connections will become accessible anywhere.

In the forefront of research and development, wireless communication standards have set and will keep preparing mobile information technologies. Already, for developing mobile data transfer, the European Telecommunications Standards Institute (ETSI) has introduced the General Packet Radio Service (GPRS) to the Global System for Mobile communications (GSM). As an enhancement of GSM data services and in particular, of packet data services, the addition of a new high- level modulation resulted in the Enhanced Data rates for Global Evolution (EDGE) including the Enhanced GPRS (EGPRS). The next step toward a mobile information society is the GSM/EDGE radio access network (GERAN) specified by the 3rd Generation Partnership Project (3GPP). GERAN will offer fully capable 3G services as in the Universal Mobile Telecommunication System (UMTS). In the future, further developments will have to be carried out for getting connections with higher speed from anywhere.

Recent developments in wide area coverage radio communication standards suggest that any competitive cellular standard should provide high user peak data rates to support demanding media. The need arises for non-real time packet data services in priority, e.g. web browsing, but relatively high real time data rates are also needed. As a solution, the use of multiple 200kHz carriers per user is a technique that has to be prospected. Therefore, this document aims at evaluating the feasibility and the impacts of a multicarrier technique for increasing GSM/EDGE data rates. After presenting GSM/EDGE standard from a packet data transfer viewpoint, the feasibility of multiple carriers will be assessed. The evaluation of multicarrier feasibility contains a physical layer study with RF requirements, frequency allocation, physical channel allocation, frequency hopping, performance requirements, and a protocol study with multicarrier capability definition, possible protocol architecture scenarios and protocol modifications.


17-Sep-01 11/61

2 PACKET DATA TRANSFER

Higher data rates in GSM/EDGE are primarily needed for non-real time data. Based on the success of the Internet Protocol (IP), telecommunication standard bodies have developed packet switched services for cellular systems. The first standard to introduce packet switched data into GSM terminals is GPRS. In order to offer higher data rates to wireless packet transfer, EGPRS has been standardized. Current GSM/EDGE release 5 standardization work focuses on GERAN, which will support third generation services through connections to 3G core network. Before looking at those standards, some general principles of packet transmission have to be presented.

2.1 PACKET TRANSMISSION PRINCIPLES

2.1.1 From data to radio waves

Packet radio transmission principles consist in converting packet data into radio blocks for sending them on radio path. A packet is a network data unit composed of a data message and control information, which provide the source and the destination identification as well as error recovery capability. When data has to be transmitted, the network layer transfers packet data to radio protocols. Then, the encapsulated data is segmented and mapped into radio blocks. A radio block is a sequence of four normal bursts ready for being transmitted by physical element on radio path [1]. Now, the transmission of packet data on radio link is ruled by packet switching and multiplexing mechanisms.

2.1.2 Packet switching

Whereas circuit-switching reserves fixed communication resources for the duration of a connection even though these may not be needed all the time, in packet switching, the resources are allocated dynamically according to the data source. Packet switching can adapt to changing conditions during a given connection and resources are used only when a packet is transmitted on the radio link. Besides, circuit-switched connection requires more bookkeeping than a packet-switched connection [2].

2.1.3 Multiplexing techniques

As the frequency spectrum for radio transmission allocated by regulatory agencies is limited, multiplexing techniques are used for sharing radio resources simultaneously and dynamically among mobile users [3]. Multiple access techniques are used for sharing simultaneously the available bandwidth. Depending on how the available bandwidth is allocated to the users, the system can be called either narrowband or wideband. In a narrowband multiple access system, the available radio spectrum is divided into a large number of narrowband channels.

GSM/EDGE is a narrowband system, which use a combination of frequency division (FDMA) and time division multiple access (TDMA). The frequency band is first divided into a set of 200kHz carriers. Each set is then decomposed into frames divided into 8 timeslots [4]. Bursts are sent in timeslots for conveying data. In full rate transmission, a channel can be defined as slots that can be used for transmitting bursts every 8 burst periods (8x15/26ms). Frequency division duplexing (FDD) is used to provide uplink and downlink channels on two different frequencies. With time division duplexing, the large gap of power levels at the transceiver between reception and transmission would disturb decisions. Packet channel can be dedicated to one user or shared between several users for a defined duration. The packet switching transmission unit is the Radio Block, which occurs every 20 ms.


17-Sep-01 12/61

Figure 1: Radio blocks in a Packet Data Channel (PDCH)

2.2 GENERAL PACKET RADIO SERVICE (GPRS) & ENHANCED GPRS (EGPRS)

The (E)GPRS systems bring new services based on packet switched bearer to the existing GSM system. When a MS is attached to the (E)GPRS network, the network protocol address (IP, X.25) can be activated for providing the user with a direct access to public data networks. As an example, an Internet connection can be established by an (E)GPRS mobile phone. From the end user's point of view, the (E)GPRS network is an Internet sub-network that has wireless access. From the Internet's point of view, the (E)GPRS network is just one sub-network among many others. Typical (E)GPRS applications are non-real time applications with best effort data like e-mail, web browsing, enhanced short messages and broadcasting.

EGPRS, the packet switched part of EDGE, is an enhancement of GPRS with the introduction of high- level modulation (8-PSK) modulation and Incremental Redundancy (also known as Hybrid Type II ARQ). The aim of EDGE is to boost the network capacity and the data rates. The main features of GPRS and EGPRS are presented in the following sections.

Whereas, the General Packet Radio Service has been standardized and has entered the deployment phase, Enhanced GPRS has recently been standardized but has not entered the deployment phase yet. EDGE radio interface was simultaneously standardized in ETSI and in TIA/EIA-136. Consequently, EDGE can be deployed in both GSM and IS-136 networks worldwide as GSM/EDGE and TDMA/EDGE respectively [5].

2.2.1 Network architecture

2.2.1.1 GPRS network architecture

In order to reuse the network subsystem with other radio access technologies a strict separation between the radio subsystem and network subsystem is maintained.

In the network subsystem, GPRS brings new elements to GSM network as shown in Figure 2 [6,7]. First, the Serving GPRS Support Node (SGSN) plays an essential role in the GPRS


17-Sep-01 13/61

network by providing the interface between the mobile radio network and the IP backbone of GPRS. The SGSN also performs essential functions such as subscriber mobility management, packet processing, ciphering, and compression. Secondly, the Gateway GPRS Support Node (GGSN) provides the interface between the GPRS network and external IP networks, such as the Internet. In this way the GGSN plays an essential role by allowing GPRS mobile users to enjoy safe and reliable access to external data services. Finally, the Border Gateway (BG) provides a direct connection to the GPRS networks of other operators. This allows operators to avoid using the public Internet when transferring data to other GPRS networks, and thus gives GPRS subscribers a secure connection to their home network when they are roaming to other GPRS networks.

In the radio subsystem, the BSS is upgraded with new GPRS protocols for the Gb interface (between the BSS and the SGSN) and enhanced radio protocols for the air interface [6]. The Gb interface connects the BSS and the SGSN, allowing the exchange of signaling information and user data. Both air and Gb interfaces allow many users to be multiplexed over the same physical resources.

Figure 2: GPRS Network Architecture [7]

While current GSM system was originally designed with an emphasis on voice sessions, the main objective of the GPRS is to offer an access to standard data networks, such as TCP/IP and X.25, with fast reservation to begin transmission of packets. These other networks consider GPRS just as a normal sub-network. A GGSN in the GPRS network will behave as a router and hide the GPRS specific features from the external data network. Packets originating from one user may take different routes through the network to receiver.

2.2.1.2 EGPRS Network Architecture

EGPRS network is similar to GPRS network. Figure 3 shows that hardware and software upgrades are needed to support the new air interface modulation and the increased data rates. The transceivers will support today’s mobile terminals with GSM modulation, as well as new user equipment providing enhanced data services with a new modulation and GSM modulation. [5]


17-Sep-01 14/61

Figure 3: EDGE built on existing GSM/GPRS networks [5]

GSM/EDGE and GSM coverage areas are similar since EDGE can use GMSK modulation, yet EDGE data rate improvement with 8-PSK has a smaller coverage area than GSM coverage area because of 8-PSK propagation [5].

2.2.2 Protocol architecture

(E)GPRS systems introduce new protocols for the GSM Phase 2+ network. As a consequence, the interworking between the new network elements is done with new specific (E)GPRS protocols [6]. However, there is number of existing protocols used at the network layer, namely TCP/UDP and IP. Figure 4 shows the transmission plane used in the (E)GPRS system. Since internal signaling in the (E)GPRS system is handled by protocols, which carry both data and signaling, no distinction is made between user and control planes.

Relay

Network Services

GTP

Application

IP / X.25

SNDCP

LLC

RLC

MAC

GSM RF

SNDCP

LLC

BSSGP

L1bis

RLC

MAC

GSM RF

BSSGP

L1bis

Relay

L2

L1

IP

L2

L1

IP

GTP

IP / X.25

Um Gb Gn Gi

MS BSS SGSN GGSN

UDP /TCP

UDP /TCP

Network Services

Figure 4: (E)GPRS Protocols in user plane between MS and BSS from

GSM/EDGE Release 4.

From the MS to the BSS, data is passing through protocols providing various functions for enabling data transmission. First, applications run on top of IP or X.25, as packet protocol offered to the subscriber by the GPRS system. Then, descending in the protocol stack, the Sub-network Dependent Convergence Protocol (SNDCP) provides mapping and compression functions between the network layer and lower layers, as well as, segmentation, reassembly


17-Sep-01 15/61

and multiplexing. Starting in the Radio Protocols, the Logical Link Control (LLC) layer guarantees a secure and reliable logical link between the MS and the SGSN to upper layers and conveys signaling, SMS and SNDCP packets. Then, between the MS and the BSS, the RLC and the MAC, which are detailed in the following sections, form together the data link layer for the Um interface. In addition, the physical layer, based on GSM physical layer with the addition of packet channels, enable the transmission over the air interface.

Within the core network, between the BSS and the SGSN, the Base Station System GPRS Protocol (BSSGP) transfers data and provides control information. Below the BSSGP, the Network Services have load sharing and redundancy functions [8]. Between the SGSN and the GGSN, the new layer is the GPRS Tunneling Protocol (GTP), used to tunnel data and signaling between the GPRS support nodes. The other layers, the L1, L1bis and L2 are vendor dependent OSI layer protocols.

2.2.2.1 Medium Access Control protocol

The Medium Access Control (MAC) protocol handles the functions related to the management of the shared transmission resources, e.g. the packet data physical channels and the radio link connections onto the packet data channels [1,9].

RLC

MAC

Physical

TBF

PDCH

data flowservice access point

PDCH PDCH

Figure 5: Service Access Points between radio protocols for a multislot MS

One of the Medium Access Control procedures is to support the provision of Temporary Block Flow (TBF) to the RLC layer for the duration of a data transfer [9]. As shown in Figure 5, TBFs allow unidirectional transfers of signaling and user data between the network and a mobile station [1]. Radio resources may be allocated to a TBF on one or more timeslots and comprises a number of RLC/MAC blocks carrying one or more packets.

In order to distinguish TBFs, which are sharing common physical resource, a unique Temporary Flow Identity (TFI) is assigned by the network to each TBF [1,9]. A TFI is unique among concurrent TBFs in each direction and is used instead of the MS identity in the RLC/MAC layer during data transfer [10]. It can be repeated from one PDCH to another, if the corresponding TBF are not sharing the same PDCHs. The same TFI value may be used concurrently for TBFs in opposite directions. The TFI is assigned with a resource assignment


17-Sep-01 16/61

message that precedes the transfer of packets belonging to one TBF to/from the MS. The TFI is included in the header part of each RLC/MAC data or control block belonging to a particular TBF in order to address the peer RLC entities.

The MAC protocol multiplexes the transmission over the physical layer of upper layer packet data units from one or more mobile station [1]. Therefore, the MAC is responsible for configuring the mapping of logical channels onto the appropriate physical channels, e.g. Packet Data Traffic Channel (PDTCH) on a Shared Physical Sub Channel (SPSCH). Different resource allocation mechanisms are offered for uplink: the dynamic allocation (and the extended dynamic allocation for multislot), and the fixed allocation using close-ended TBF or open-ended TBF for defining the allocated radio blocks (plus, the exclusive allocation for dual transfer mode, i.e. both data and speech).

In dynamic and extended dynamic allocation, Uplink State Flag (USF) is used on downlink PDCHs to enable the multiplexing of uplink radio blocks from different MSs [1]. As soon as a MS detects a USF allocated to it in downlink, the MS can transmit on the uplink resource pointed by the USF. The USF points either to the next uplink Radio Block or the sequence of four uplink Radio Blocks starting with the next uplink Radio Block. The GPRS and EGPRS MSs can be multiplexed dynamically on the same PDCH by utilising the USF.

In fixed allocation, an uplink part of the PDCH is reserved only for one MS during a certain period of time. It can be used to multiplex GPRS and EGPRS MSs on the same PDCH on the uplink.

2.2.2.2 Radio Link Control protocol

The Radio Link Control (RLC) provides procedures for segmentation and reassembly of LLC packet data units into RLC/MAC blocks [1]. The RLC operates either in acknowledged or unacknowledged mode depending on the desired reliability of the radio links.

In Acknowledged mode of operation, the Radio Link Control (RLC) provides reliable radio links to the upper layers by using Backward Error Correction (BEC) procedures and Link Adaptation. Such mode is used when no delay requirement exists, for example in file downloading or email transfer. On one hand, upon reception failure, BEC procedures enable the selective retransmission of RLC data blocks. On the other hand, Link Adaptation offers a dynamic selection of (modulation and) coding schemes in order to maximize the system throughput according to the time-varying channel conditions. In GPRS, the Link Adaptation is based on the signal-to- interference ratio as an estimate of the block error rate for choosing the best coding scheme. In EGPRS, the LA decision is based on the bit error probability measurement as the quality measurement for choosing the best MCS [11]. The RLC may also notify its upper layer of errors that cannot be resolved by itself in normal exception handling procedure [9].

In EGPRS acknowledged mode, upon reception failure, the retransmission of a data block includes additional redundancy compared to its earlier transmissions. Incremental Redundancy (IR) is a BEC that enables the combination of information with the earlier (re)transmission(s) in order to correct errors. In fact, the right protection of the data block is obtained incrementally, after transmitting additional redundancy and combining it with the prevailing (re)transmission(s). Incremental Redundancy is a physical layer performance enhancement mandatory in the MS in the downlink and optional in the BSS in the uplink [1].

The flexibility of radio blocks acknowledgement is increased with the use of RLC window mechanism by both network and MS side at transmission and reception [9]. For a TBF, the


17-Sep-01 17/61

acknowledgment window size is set by the network to 64 radio blocks in GPRS and to a value between 64 to 1024 radio blocks in EGPRS according to the timeslot allocation (Annex C). As uplink and downlink resources may be asymmetric, the window size can be set independently on uplink and downlink.

In Unacknowledged mode of operation, the RLC transmits higher layer PDUs without guaranteeing delivery to the peer entity. Such mode is used when there is a strong requirement to preserve time relation between information blocks, for example in video streaming. No backward error correction procedure is supported in this mode. However, a sequence number check is done to detect RLC blocks loss and to guarantee in-order delivery to the receive-end.

2.2.3 Radio interface

2.2.3.1 Radio Access Capability

In (E)GPRS system, MSs have two new Radio Access Capability features: the service mode of operation and the multislot capability.

Each (E)GPRS MS belongs to a class depicting the services it can operate [6]:

§ Class-A MS may operate (E)GPRS and other GSM services simultaneously.

§ Class-B MS can monitor control channels for (E)GPRS and other GSM services simultaneously, but it may only operate one set of services at one time.

§ Class-C MS exclusively operates (E)GPRS services.

(E)GPRS MS may have a multislot capability, which enables them to use between 1 to 8 timeslots [12]. To define the multislot capability, multislot classes defined in Annex B are indicated in the Radio Access Capability [13]. As uplink and downlink channels are reserved separately, various multislot resource configurations may be allocated in different directions.

2.2.3.2 GPRS Radio Interface

GPRS offers four alternative channel Coding Schemes, CS-1 to CS-4, from highly redundant coding to almost no coding. The more redundant the code is, the more protected against channel fading the information is. In radio channels varying with time, the networks should dynamically choose a suitable CS for transmission in order to maximize the system throughput, i.e. the traffic occupancy [1]. This procedure is called Link Adaptation (LA). In GPRS, the LA decision is based on the channel measurement reports sent by the MS to the network.

With the coding parameters presented in Table 1 for GPRS CS, the "theoretical" peak data rate reaches 171.2 kbps (8x21.4), using all the eight channels of a carrier. However, the capabilities of the network or the multislot capability of the mobile station may set a lower maximum data rate. The multislot capability defines also some procedural differences to mobile stations, e.g. full-duplex and half-duplex modes of operation.

Scheme Modulation Rate Convolution Puncture Code Rate Data rate [kbps] Blocks per 20 ms CS-1 ½ No 0.5 9.05 1 CS-2 ½ Yes 0.64 13.4 1 CS-3 ½ Yes 0.74 15.6 1 CS-4

GMSK

- No 1 21.4 1


17-Sep-01 18/61

Table 1: GPRS channel coding schemes

2.2.3.3 EGPRS Radio Interface

The major change to GSM standard to support higher data rates is the introduction of a new modulation, known as 8-PSK (Octal Phase Shift Keying). 8-PSK signals are able to carry three bits per modulated symbol instead of one bit per symbol with GMSK modulation [1]. Thus data rates can be increased up to three folds in a smaller coverage area. To differentiate GMSK from 8-PSK, blind detection is used in EGPRS based on the different phase characteristics of GMSK and 8-PSK modulations [1].

In order to adapt to channel fading fluctuation and to send data with optimal coding schemes, EGPRS offers various Modulation and Coding Schemes (MCS). GMSK and 8-PSK modulations can be used for user data transfer. GMSK being a more robust modulation than 8-PSK, it is used in most difficult channel conditions for user data transmissions and control signalling. However, IR also allows 8-PSK to operate in bad channel conditions.

Nine different modulation and coding schemes, MCS-1 to MCS-9, are defined for the EGPRS Radio Blocks (4 bursts, 20ms) carrying RLC data blocks. A general description of the MCSs is given in Figure 6 [1].

MCSs are divided into different families A, B and C. Each family has a different basic unit of payload: 37 (and 34), 28 and 22 octets respective ly. Different code rates within a family are achieved by transmitting a different number of payload units within one Radio Block. For families A and B, 1, 2 or 4 payload units are transmitted. For family C, only 1 or 2 payload units are transmitted.


17-Sep-01 19/61

Figure 6: General description of the payload formats for EGPRS

When four payload units are transmitted (MCS-7, MCS-8 and MCS-9), the total payload is split into two separate RLC blocks (i.e. with separate sequence numbers and BCSs). These blocks, in turn, are interleaved over two bursts only, for MCS-8 and MCS-9. For MCS-7, these blocks are interleaved over four bursts. The other MCSs carry one RLC block, which is interleaved over four bursts. When switching to MCS-3 or MCS-6 from MCS-8, 3 or 6 padding octets, respectively, are added to the data octets.

To allow Incremental Redundancy, the header part of the Radio Block is independently coded from the data part of the Radio Block. Three different header formats are used, one for MCS-7, MCS-8 and MCS-9, one for MCS-5 and MCS-6 and one for MCS-1 to MCS-4. The two first formats are for 8-PSK modes, the difference being in the number of Sequence Numbers carried (2 for MCS-7, -8 and -9, 1 for MCS-5 and -6). The third format is common to all GMSK modes. The header is always interleaved over four bursts [1,9].

The details of the EGPRS coding schemes are shown in Table 2:

Scheme Code rate

Header Code rate Modulation

RLC blocks per Radio Block (20ms)

Raw Data within one Radio Block Family

Data rate kb/s

MCS-9 1.0 0.36 2 2 x 592 A 59.2 MCS-8 0.92 0.36 2 2 x 544 A 54.4 MCS-7 0.76 0.36 2 2 x 448 B 44.8 MCS-6 0.49 1/3 1 592 or 544+48 A 29.6 or 27.2 MCS-5 0.37 1/3

8PSK

1 448 B 22.4 MCS-4 1.0 0.53 1 352 C 17.6 MCS-3 0.80 0.53 1 296 or 272+24 A 14.8 or 13.6 MCS-2 0.66 0.53 1 224 B 11.2 MCS-1 0.53 0.53

GMSK

1 176 C 8.8

Table 2: EGPRS Modulation and coding schemes

For using Incremental Redundancy, two or three punctur ing schemes are designed (P1 to P3) per MCS in order to maximize the block combination performance after data retransmission. The P1 puncturing scheme is used for the first transmission of the data block and P2 or P3 are used for the possible following retransmissions of the data block. As the first puncturing scheme (P1) is always used, it is important that the performance with P1 is same or better than with P2 or P3.


17-Sep-01 20/61

2.2.3.4 GSM/EDGE main characteristics

The Table 3 summarizes the main characteristics of GSM/EDGE systems, which apply to multicarrier [1,3,4,6].

GSM/EDGE Main multiple access parameters Multiple access method TDMA Duplexing method FDD Number of carriers per link 124 carriers considering 100kHz guard bands at upper and lower ends Channel frequency spacing 200kHz Physical layer structure Time slot structure 8 slots/TDMA frame Frame length 4.615 ms Multirate concept Various Coding Schemes and Multislot configurations Number of users per physical channel 1-32 in packet switched, 1(full-rate) - 4(quarter rate) in circuit switched Forward Error Correction codes convolutional codes, puncturing Interleaving block rectangular for packet data, diagonal for speech & circuit switched data Data modulation GMSK/8-PSK Modulating symbol rate 270,833 ksymbol/s Detection of modulation Coherent based on training sequence Additional diversity means Frequency Hopping per TDMA frame Other Radio Transmission features Power control Slow power control, 30 dB dynamic range Handover Synchronous or asynchronous handover Channel allocation Fixed and Dynamic Channel Allocation supported

Table 3: GSM/EDGE main characteristics

2.3 GSM/EDGE RADIO ACCESS NETWORK (GERAN)

Part of the GSM/EDGE standardization work aims at harmonizing GERAN with UMTS Terrestrial Radio Access Network (UTRAN), enabling GERAN to connect to the same 3G core network and provide the same set of services as UTRAN [14,15].

2.3.1 3G mobile services

As a part of the radio access network harmonization, services provided by GERAN are aligned with the four UMTS traffic classes: conversational, streaming, interactive, and background. All those traffic classes are using dedicated connections called Radio Access Bearer (RAB) [15]. First, with conversational traffic, speech or video is carried between end-users. The QoS requirements are mainly set by human perception. Failure in meeting these requirements causes a degradation of the perceived service quality to a level that is often unacceptable. Second, streaming traffic is unidirectional user data transfer having no absolute delay requirement. However, there is a strict requirement to preserve time relation between information entities. The stream can be time aligned in the receiving end by buffering data. Then, interactive traffic consists of sessions where client is requesting data from a remote server (e.g. web browsing). Generally, a request-response pattern exists, therefore, it is important to guarantee correct delivery of requested data. Finally, background traffic has relaxed delay requirement for applications like short messages, e-mails and file transfers.


17-Sep-01 21/61

Thus, GERAN may provide simultaneously real time and non-real time services, e.g. voice call, email transfer and web browsing.

2.3.2 Network Architecture

The basic principles applied to the design of GERAN architecture are the separation of the radio related and non-radio related functionality between the core network and the radio access network (Figure 7), and the support of terminals built for previous standard releases [14]. GERAN and UTRAN harmonization is maximized but the backward compatibility with earlier releases of GSM/EDGE has to be maintained in GERAN.

GSM/UMTS Core Network

GERAN

Gb

A

Iu

MS Um

Iur-g

BSC

BTS

BTS

BSS

BSS

MS

Iur-g (FFS)

UTRAN RNC

Figure 7: GERAN reference architecture [14].

GERAN's key issue is the common Iu interface between the Radio Access Network (RAN) and the 3G core network that enables GERAN. The Iu interface connects GERAN to 3G core network. Thus, GERAN provides the same set of services, support the same radio access bearers and QoS mechanisms as in UTRAN.

The A and Iu-cs interfaces connect the RAN towards the circuit switched domain of the 2G and the 3G core network respectively. The Gb and Iu-ps interfaces connect the RAN towards the packet switched domain of the 2G and 3G core network respectively. As this study focus on packet data, the Gb and Iu-ps interfaces are the ones involved in the higher data rates study. Finally, the Iur-g interface may connect two GERAN networks.

2.3.3 Protocol architecture

The new architecture implies significant modifications to the GERAN radio protocols. Also the efficient support of IP based services requires new functionality in the radio protocols.

In Figure 8, GERAN protocol architecture is divided into Control (C-) and User (U-) planes [14]. The RLC and MAC protocols and the physical layer carry data from both C- and U-plane. C- and U-plane functionality are separated in Radio Resource Control (RRC) and Packet Data Convergence Protocol (PDCP) respectively.

In GSM/EDGE Release 5, the Physical and MAC layers enable the multiplexing of user data from different core network interfaces. As an MS may have several Radio Bearers, more than one TBF may be allocated to a single MS in any direction. Except for those main points, the Physical, the MAC and RLC layers are similar to their definitions in (E)GPRS.


17-Sep-01 22/61

C-plane signalling U-plane information

RRCPDCP PDCP

M A C

PDCP

PHY

control

control

RLC RLC RLC RLC RLC

GC Nt DC

GC Nt DC

Duplication avoidance

RLC

control

M A C

Logical channels

control

LAPDmLAPDm

DL

control

RA

Figure 8: Radio Interface protocol architecture [14]. In this figure, the LAPDm inherited from GSM still exist on BCCH.

2.3.3.1 Radio Resource Control protocol

The Radio Resource Control (RRC) layer handles the control plane signaling of Layer 3 between the MSs and GERAN. Some RRC functions as paging, control of ciphering and others, derives from the LLC layer and RR management defined in GSM. In Figure 8, the RRC has local inter- layer connections controlling the configuration of low layers [14].

To support 3G services, new RR concepts from UTRAN are introduced in GERAN RRC, e.g. the Radio Bearer, the RRC connection and the RRC mobility management concept [14]. Some other RRC functions are broadcast of information, paging, notification, routing of higher layer PDUs, control of requested QoS, initial cell selection and reselection in idle mode, support for location services, control of security functions and MS measurement re-porting and control of the reporting.

Over Iu interface, several configurations of the protocol stack called Radio Bearer (RB) are defined to realize each RAB. The RRC realizes the QoS requirements of the RAB by establishing a Radio Bearer (RB) between MS and GERAN. For each RB RRC allocates either a dedicated or a shared channel. RRC and MAC are responsible for the allocation of dedicated and shared channels respectively. GERAN RRC state machine is presented in 3GPP TS 43.051.

The RRC also has connection mobility functions, which mean that the GERAN tracks the location of MS without involving the core network. RRC connection mobility has two levels of accuracy: Cell and GERAN Registration Area (GRA). In the cell level of accuracy, the MS makes cell updates after each GERAN cell reselection and respectively the network tracks the MS's location on cell level. For inactive users, the MS may fall back to the GRA level of accuracy. A GRA is a specified set of cells, which can be identified in the Broadcast Control Channels. When the MS is not connected, e.g. in the state named idle mode in 3GPP TS


17-Sep-01 23/61

43.051, the MS is identified by non-access stratum identities and the GERAN has no own information about the individual MS.

2.3.3.2 Packet Data Convergence Protocol

PDCP functions are to adapt the Data Link layer to Network layer data units and vice-versa. It operates in two modes: transparent and non-transparent [14].

In Transparent Mode, the PDCP layer does not change the incoming service data units (SDU). No header is added and the possible TCP/IP or RTP/UDP/IP headers in the data are left without modification.

In Non-Transparent Mode, header adaptation mechanism is used by the PDCP in order to optimize the transmission over the air interface e.g., by removing redundant header information. Two different mechanisms may be used by PDCP, header compression and header removal. Header compression consists of compressing transport and network level headers in such a way that the decompressed headers are semantically identical to the original uncompressed headers. It is suited for standard internet applications, especially fo r multimedia applications. Header removal consists in removing the Transport and network level headers (e.g. RTP/UDP/IP). Based on information submitted at call set-up and based on information derived from lower layer (link & physical), the receiving entity can regenerate the headers. The primary application of header removal is the optimized speech bearer, and the regenerated header may not always be semantically identical to the original header.


17-Sep-01 24/61

3 MULTICARRIER FEASIBILITY STUDY

3.1 SCOPE OF THE STUDY

The aim of this study is to develop a multicarrier concept to enable higher data rates in GERAN with minimal modifications to existing GSM/EDGE standard. Most of the findings for supporting multicarrier are enclosed in the Physical Layer and the Radio Protocols parts.

The main requirements and constraints of this study are:

• Modifications to GSM/EDGE specifications should be minimized,

• Higher data rates are provided by assignment of multiple 200kHz carriers to a single user,

• The peak data rate should be at least 2 Mbps,

• The study is applicable to GSM/EDGE Release 5, which is presently being standardized.

The Table 4 shows the achievable gross data rates for a user per number of timeslots and carriers. Stealing bits/symbols are excluded in the calculation.

Data rate in kbps per number of Time Slots Modulation Carriers

1 2 4 8 Single 200 kHz carrier 22.8 45.6 91.2 182.4

2 carriers 45.6 91.2 182.4 364.2 3 carriers 68.4 136.8 273.6 547.2 4 carriers 91.2 182.4 364.2 729.6 5 carriers 114.0 228.0 456.0 912.0

GMSK

6 carriers 136.8 273.6 547.2 1094.4 Single 200 kHz carrier 68.4 136.8 273.6 547.2

2 carriers 136.8 273.6 547.2 1094.4 3 carriers 205.2 410.4 820.8 1641.6 4 carriers 273.6 547.2 1094.4 2188.8 5 carriers 342.0 684.0 1368.0 2736.0

8-PSK

6 carriers 410.4 820.8 1641.6 3283.2

Table 4: Gross data rates for the multicarrier.

In order to support 2 Mbps data rate for a single connection, at least five carriers should be combined, assuming some overhead due to the radio protocol headers. For example, a connection using MCS-9 on eight time slots has a data rate equal to 473.6 kbps although the peak gross data rate is 547.2 kbps.

3.2 RADIO PROTOCOL ELEMENTS AFFECTED

The introduction of multicarrier affects various elements of GSM/EDGE radio protocols. The elements, which are studied in this report, are here presented.

Starting from the lowest layer, current physical layer, which was designed for single carrier transfers, has to be upgraded for supporting multicarrier transfers. At first, the multicarrier receiver model has to be designed based on the single carrier one. Low cost and low


17-Sep-01 25/61

complexity criteria shall be taken into account. As a result, the discovered hardware restrictions shall be taken into account in the rest of the study. Secondly, the radio parameters limiting the multicarrier capability, like the bandwidth of the multicarrier allocation and the offset between carriers, have to be studied. Consequently, radio and hardware critical parameters could synthesize the multicarrier capability. Then, to enable the simultaneous support of multislot and multicarrier, mere rules shall apply when assigning channels to a multicarrier MS. Next, channel coding and interleaving schemes shall be studied in order to support the multicarrier concept. In addition, the possibility, brought by the multicarrier concept, to interleave data over several carriers shall be evaluated. Besides, the compatibility of the multicarrier concept with frequency hopping shall be studied. Finally, some multicarrier RF, link level and radio performance requirements or results shall be presented.

After the physical layer study, modifications to layer 2 protocols for supporting multicarrier transfers have to be studied. At first, multicarrier classes shall be designed in order to enable MS to inform the network of its multicarrier capability. Secondly, different scenarios of protocol architecture upgrade for supporting multicarrier shall be evaluated. Then, MAC, RLC, PDCP and RRC modifications shall be studied.

4 PHYSICAL LAYER

4.1 RADIO FREQUENCY

As the need of multicarrier first appears in downlink transfers, reception at the MS is studied in this section.

4.1.1 Receiver model

From the RF front end point of view, a multicarrier signal can be seen either as several narrowband signals or as a wideband signal. Those two possibilities are presented for multicarrier signal reception in following figures. Figure 9 shows a multicarrier receiver model composed of one RF branch. In this model, the multicarrier signal is received as a wideband signal and separated after analog-to-digital conversion into parallel baseband processing chains.

Demodulator

EQblock

DECblock

Bits fromcarrier 1

Bits fromcarrier N

Demodulator

EQblock

DECblock

GSM

band

filter

RF

down

conversion

Analog

Digital

convertor

Multi-carrierband

Analogfilter

Narrow-band filter

Narrow-band filter

carrier 1

carrier N

Figure 9: Multicarrier receiver with one RF branch.

EQ Blocks consist of channel estimation and equalisation. DEC blocks consist of de- interleaving, de-puncturing and decoding.

Figure 10 presents a multicarrier receiver composed of several RF branches receiving different carriers of the multicarrier signal.


17-Sep-01 26/61

Demodulator

EQblock

DECblock

Bits fromcarrier 1

Bits fromcarrier N

Demodulator

EQblock

DECblock

RF downconversion

RF downconversion

GSMbandfilter

A/Dconvertor

A/Dconvertor

GSMbandfilter

Narrow-band filter

Narrow-band filter

carrier 1

carrier N

Figure 10: Multicarrier receiver with one RF branch per carrier. EQ Blocks consist of channel estimation and equalization.

DEC blocks consist of de- interleaving, de-puncturing and decoding.

As the receiver structure in Figure 10 would lead to a large number of multiple parallel RF branches increasing the cost of the receiver, the wideband RF receiver structure shown in Figure 9 is taken as a basis for the further discussion in this thesis.

4.1.2 Frequency offset between carriers

Frequency offsets between carriers are defined from the middle of one carrier to the middle of the other one as shown in Figure 11. Then, two adjacent carriers are 200kHz offset apart.

frequency

carrier band200kHz

200 kHzoffset

400 kHzoffset

Figure 11: Frequency separation between carriers

In multicarrier allocations, offsets are limited by the two following factors. On one hand, small offset produces spectrum overlapping, especially when the signal level of one carrier is overwhelming the signal levels of other carriers in the MS allocation. To reduce this spectrum overlapping, the current Gaussian pulse shaping filter would have to be replaced by a new pulse shaping filter. Such a change in the basis of GSM/EDGE would affect the whole standard. On the other hand, large carrier separation results in an increase of the power consumption due to the wider bandwidth covered by the analog baseband filter and AD converter. Besides, in GSM900 and GSM1800, the most common separation between consecutive frequencies in a cell allocation is respectively 400kHz and 800kHz. Therefore, offsets higher or equal to 400kHz are the most typical offset values that have to be considered for multicarrier and small offsets are preferred.


17-Sep-01 27/61

4.1.3 Bandwidth of the multicarrier allocation

4.1.3.1 Bandwidth limitation

In a multicarrier MS receiver, the multicarrier band analog filter is the first element located after the RF down conversion. It enables the rejection of interfering signals located outside the multicarrier band. The width of this filter determines the bandwidth of the signals entering in the receiver. Considering a filter with 2.6MHz edge to edge width (1.3MHz lowpass filter) for a multicarrier MS would be equivalent to cover a band of 13 GSM carriers as shown in Figure 12.

Figure 12: Bandwidth and carriers in the GSM900 band covered by a 2.6MHz wide analog filter.

Yet, the maximum band covered by an analog filter may be smaller than the operator band in one direction, which is a fraction of the total GSM band. Consequently, the limitation in bandwidth of the multicarrier MS, will determine the number of carriers which may be involved in its multicarrier allocation. Therefore, the bandwidth supported by the MS has to be taken into account in the definition of the multicarrier capability.

4.1.3.2 Bandwidth occupation

This section focuses on the physical layer concept of the multicarrier. As concluded previously, 200 kHz RF carriers to be allocated for one user are separated by at least 400kHz.

Figure 13 shows in frequency domain two examples of multicarrier signal with three adjacent carriers. In the first, 400 or 600 kHz offsets separate carriers. In the second, 600 kHz offset separate carriers.

frequency

200kHz

1.2MHz400 kHz

offset

frequency

200kHz 600 kHzoffset

1.4MHz

Figure 13: Multicarrier signals with bandwidth of 1.2MHz and 1.4MHz

GSM 900 Downlink band

935 MHz

0 1 2 3 5 4 6 7 8 9 10 11 12 13 14

935.2

Filter covering 2.6MHz

935.8 935.6 935.4 937 938 936

15 16

(MHz)


17-Sep-01 28/61

In the previous figures, the multicarrier signals, composed of three carriers, cover different bandwidths. Therefore, as Figure 14 shows, the bandwidth occupation of the multicarrier allocation depends on the offset chosen between carriers and the number of carriers in the multicarrier allocation.

Figure 14: Bandwidth of MS multicarrier allocation depending on the number of carriers and the offset. At maximum, the bandwidth is equal to 3.2MHz.

Consequently, the multicarrier capability is partly defined by the offsets supported by a multicarrier MS, the maximum number of carrier, equivalent to the number of processing chains in the receiver, and the maximum bandwidth of the analog filter. Those parameters of the multicarrier capability have to respect this inequality:

∑ ≤+=

1-usedcarrier ofnumber

1Nfilter analog bander multicarri theof Bandwidth200kHz 1-N andN carriers between usedoffset

4.2 PHYSICAL CHANNEL ALLOCATION

The multicarrier concept requires some modifications of the physical resource allocation to avoid simultaneous transmission and reception on all carriers and to reduce power consumption in the MS.

With one antenna (section 4.1.1), a multicarrier MS cannot receive and transmit simultaneously on all carriers. To enable uplink and downlink transfers on all carriers without overlapping, a multislot window mechanism could be used. This mechanism would delimit the multicarrier resources used for each unidirectional transfer as shown in Figure 15. On uplink or downlink, the size of the multislot window should be smaller than, or equal to, the


17-Sep-01 29/61

maximum number of slots indicated by the multislot class (Annex B) for transmission or reception respectively.

Allocation of physical resources should minimize the power consumption in MS. By transmitting consecutive bursts as in multislot, power ramping is done only at the beginning and at the end of the series of consecutive bursts, which reduce power consumption. In frequency domain, as the number of carriers assigned is equal to the number of chains activated in the receiver (section 4.1.1), the smaller the number of carriers used, the less the power consumed by the receiver chains. To sum up, the use of packed slots in frequency and time entails minimum power consumption.

For illustrating the use of packed slots and the multislot window, as an example, a multislot class 8 multicarrier MS has resources allocated for downlink on 12 timeslots on 5 carriers and for uplink 2 timeslots on 2 carriers in Figure 15. In this example, it is assumed that the MS do not have to perform adjacent cell signal level measurement. Multislot class 8 MS can have at most 4 receiving timeslots and 1 transmitting timeslots per carrier, multislot classes are depicted in Annex B.

Downlinkmultislot window

multicarrier user other users

Full downlink multislotcapability = Rx

Full Uplink multislotcapability = Tx

Uplinkmultislot window

RF9

RF7

RF5

RF3

RF1

TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7

RF9

RF7

RF5

RF3

RF1

TS0 TS1 TS2 TS3 TS4 TS5 TS6 TS7

400kHz

400kHz

400kHz

400kHz

400kHz

400kHz

400kHz

400kHz

UPLINK

DOWNLINK

Figure 15: Time slot (TS) - frequency channel (RF) allocation of a multislot class 8 multicarrier MS below its maximum multislot capability.

Similarly to the multislot class, multicarrier classes need to be defined, for multicarrier MS in 3GPP TS 45.002 as shown in the following section 5.1.


17-Sep-01 30/61

4.3 CHANNEL CODING AND INTERLEAVING

4.3.1 Logical channels

As the multicarrier concept is an enhancement of GSM/EDGE standard and the GSM/EDGE physical channel structure should not be modified by the introduction of the multicarrier concept, the traffic and control channels used for the multicarrier are the ones defined in the GSM/EDGE standard.

4.3.2 Channel coding

The channel coding schemes used in the multicarrier concept are kept unchanged from GERAN Release 5.

4.3.3 Interleaving

For the multicarrier concept, inter-TDMA frame and intra-burst interleaving are assumed as basic assumptions. However, an interleaving within several carriers allocated to a multicarrier MS could be considered. Such inter- frequency interleaving gives frequency diversity, similarly to frequency hopping today.

To enable the multiplexing of multicarrier MSs using inter- frequency interleaving with other MSs, the RLC/MAC header shall not be interleaved. Since header and data are interleaved together in GPRS [15], inter- frequency interleaving cannot be implemented in GPRS MSs. In EGPRS, header and data have different interleaving schemes [15]. Therefore, it is possible to interleave separately data over several carriers as shown in Figure 16. As the RLC/MAC header is not changed, all EGPRS MS will be able to detect the transmitted USF. Thus, in EGPRS, the MS supporting inter- frequency interleaving can be multiplexed with other MSs.

Figure 16: EGPRS inter- frequency interleaving

In each MCS, a modulation and a coding scheme are associated for offering a data rate performance. Each coding scheme implies which modulation to use. 8-PSK cannot be used for a MCS designed for GMSK, and vice-versa. Therefore, inter-frequency interleaving shall be achieved separately for each modulation, GMSK and 8-PSK.

In (E)GPRS, the same output power shall be used on all four bursts of a radio block [20]. Yet, with the use of inter-frequency interleaving, the encoded bits of radio blocks from different carriers are interleaved together and each bit would be sent at the power defined for its radio


17-Sep-01 31/61

block. As a consequence, in a burst, each bits would be sent at different power, on contrary to power control aims to reduce the power consumption and to improve the spectral efficiency [4]. Therefore, to support correctly the inter-frequency interleaving, the power control should be achieved per burst instead of radio block.

Some disadvantages appear against the use of inter- frequency interleaving. At first, when (E)GPRS control messages, which are GMSK modulated, have to be sent, or when RLC blocks are retransmitted in GMSK, no 8-PSK blocks can be sent at the same time. Secondly, the use of current link quality measurements would reflect conditions of frequency interleaved channels. Thus, Link Adaptation could not be done per carrier. But, Link Adaptation could only select one MCS for all resources used by the multicarrier interleaved resources during transmission and retransmission. Consequently, Link Adaptation performances would be reduced significantly.

4.4 FREQUENCY HOPPING

Frequency hopping (FH) consists in changing regularly the carrier frequencies of physical channels. In multicarrier, frequency hopping may not be supported because of the limited bandwidth covered by the MS (part 4.1.3). Indeed, frequency hopping can only be activated when the maximum multicarrier bandwidth supported by the MS is smaller than the bandwidth between the lowest and the highest frequencies available in the cell. To avoid that the loss of frequency hopping widens to all frequencies, multicarrier and single carrier resources have to be allocated on different frequencies of the cell. Due to this partition, the frequency hopping gain will decrease with the reduction of resources [3]. As long as radio network planning is not dynamic, the separate frequency allocation for multicarrier has to be dedicated permanently in each cell. Figure 17 presents an example of split in single carrier and multicarrier frequencies.

frequency

timeframe period

Hopping resources for single carrier Mobile Stations

Resources for multicarrier

RF9

RF7

RF5

RF3

RF1

200kHzcarrierband

400kHzoffset

frequenciesin the cell

Figure 17: Frequency split between single carrier and multicarrier users

In the multicarrier band, frequency hopping can be activated when the size of the band is smaller than the bandwidth supported by all multicarrier MSs in the cell. Otherwise, the fluctuating distance between the frequencies at both ends of a multicarrier MS allocation may exceed the bandwidth limitation indicated in the multicarrier class of the MS. Despite, cyclic hopping seems providing constant distance between the frequencies at both ends of a multicarrier MS allocation. The roll-over of the allocation prevents using cyclic hopping. The


17-Sep-01 32/61

roll-over is the fact of having parts of the MS allocation at both ends of the multicarrier allocation when cyclic hopping is used, take place as shown in Figure 18. If multicarrier frequencies do not hop, non-hopping single carrier users can be assigned resources in the multicarrier band.

frequency

time

allocated carrier

200kHzcarrierband

400kHzoffset

Hoppingcycle

frame period

RF5

RF3

RF1

RF7

frequency band of themulticarrier MS allocation

frequency band extension of the multicarrierMS allocation due to roll-over effect

Frequencies in themulticarrier allocation

Figure 18: frequency band variation of the multicarrier allocation due to the roll-over effect in the cell allocation.

4.5 PERFORMANCE REQUIREMENTS AND RESULTS

4.5.1 RF performance

To reject high power interferer within a multicarrier allocation, demanding requirement are required for the design of the Analogue to Digital Converter (ADC), the dynamic and the narrowband analog filtering. The complexity of the RF design could be reduced thanks to the avoidance of high power interferer located in multicarrier allocations.

4.5.2 Link level performance

In the link- level field, multicarrier concept unveils inter- frequency interleaving as one potential enhancement that has to be evaluated. Figure 19 shows the effects of inter- frequency interleaving using independent channel models and 400kHz offset between carriers over1, 3, 5 MCS-5 carriers.


17-Sep-01 33/61

Figure 19: Inter- frequency interleaving performance results using MCS-5 in a Typical Urban environment at 50km/h speed (TU50) with no Frequency Hopping.

At a BLER of 10-2, the inter-frequency interleaving provides significant gains in performance of 2,7 dB with 3 carriers and 4 dB with 5 carriers. However, with the addition of link adaptation constraints presented in part Interleaving4.3.3, inter- frequency interleaving performances may reduce significantly.

4.5.3 Radio performance

The following points have to be taken into account in evaluating the radio performance of multicarrier:

• Interference scenarios taking into account adjacent interferer and carrier to interference ratio.

• Due to light coding in high rate MCS, EGPRS MCS-9 and MCS-8 have better performance without frequency hopping. Therefore, MS using MCS-9 and MCS-8 should preferably have resources in the non-hopping multicarrier band when it exists in the cell.

• Multicarrier connections, as multislot connections, reduce the trunking efficiency of the system because those connections use several physical channels simultaneously. This has an impact on the overall trunked system capacity [3].


17-Sep-01 34/61

5 RADIO PROTOCOLS FOR MULTICARRIER

5.1 MULTICARRIER CAPABILITY

5.1.1 MS multicarrier capability

Based on multislot capability definition (Annex B), the MS multicarrier capability can be defined within multicarrier classes. The term multicarrier class would refer to the different mobile station capabilities to transmit and receive on different combinations of multiple 200kHz carrier.

For defining multicarrier classes, following points have to be taken into account:

• the MS maximum capability to receive and transmit over several carriers simultaneously,

• the maximum bandwidth supported by the MS for the multicarrier allocation,

• the MS multislot class, and optionally,

• the capability to have inter-frequency interleaving within the multicarrier allocation.

Any MS shall belong to a multicarrier class.

Maximum number of carriers

Maximum multicarrier allocation bandwidth

(Optional) Inter-Frequency interleaving

000 = 1 carrier 001 = 2 carriers 010 = 3 carriers 011 = 4 carriers 100 = 5 carriers 101 = 6 carriers 110 = 7 carriers 111 = 8 carriers

00000= carrier width 00001= 2x(carrier width) 00010= 3x(carrier width) 00011= 4x(carrier width)

.

.

. 11101= 30x(carrier width) 11110= 31x(carrier width) 11111= 32x(carrier width)

0=Yes

1=No

Table 5: Multicarrier classes coded on 8 bits.

In GSM/EDGE, the carrier width is 200 kHz large. Multicarrier capability may vary with the network and the MS. Therefore the network shall ensure that the resource allocation used for a mobile station is compliant with the multicarrier class of the mobile station. To do so, the Multicarrier class may be provided to the network in the MS Radio Access Capability information element (3GPP 24.008). This information element defines the way the network will handle the operations with the MS.

5.1.2 Network multicarrier capability

As GSM/EDGE Radio Access Network can manage connections on several carriers simultaneously, GSM/EDGE networks are capable to communicate on multicarrier resources [17]. Indeed, for the network, a multicarrier MS can be considered as several single carrier MS. Consequently, some modifications in the network has to be done in order to allow a MS


17-Sep-01 35/61

to use several carriers and several block flows. Finally, as the physical layer should achieve inter- frequency interleaving, at the base station, an upgrade enabling the connection between TRXs may be required for supporting inter-frequency interleaving in a cell.

5.2 PROTOCOL ARCHITECTURE SCENARIOS

In the multicarrier concept based on GSM/EDGE Release 5, a Radio Bearer data flow has to be demultiplexed onto physical resources allocated on several carriers. A solution for providing carrier demultiplexing is to upgrade the MAC layer, which performs (de)multiplexing to (from) physical resources. Besides, the RLC layer would also have to be modified for supporting the increase in data rate. Another strategy is to consider that the multicarrier radio protocol architecture is composed of single carrier entities. This way, changes in current specifications would be reduced. As the RLC and the MAC layer have been designed for single carrier resources, the next layer, which is the PDCP, could achieve the carrier demultiplexing. Finally, an RLC instance cannot perform multiplexing or splitting function [14]. Therefore, two different protocol architectures have to be stud ied, corresponding to the location of the user data stream demultiplexing to multiple carriers. This demultiplexing can be located either in MAC layer or in the PDCP layer.

5.2.1 Case 1: Multicarrier with global RLC and MAC entities

Considering the protocol architecture presented in Figure 20, on the user plane, the RB user data flow is transferred to the PDCP layer, next, to the RLC layer. Then, the RLC blocks are conveyed to the MAC layer through a multicarrier Temporary Block Flow. Finally, the MAC de-multiplexes RLC/MAC blocks onto available logical channels which are modulated and transmitted by the physical layer. On the control plane, the RRC has control management functions and configure the different user plane protocol layers. Besides, the MAC synchronizes multicarrier data transfer.

RLCRLC

User-plane data

Multicarrier signal

PDCP

MAC

Physical

PDCP

MAC

TBF

Logical channels

RRC

Control-plane data

user flowcontrol flowservice access point

Figure 20: Radio protocols architecture for multicarrier in packet switched domain with one RLC entity and one MAC entity.


17-Sep-01 36/61

Retransmission on a different carrier is one major advantage of this architecture. Yet, if frequency hopping is activated, retransmitting on a different carrier may not provide additional gain in frequency diversity compared to a retransmission on the same carrier.

A drawback of this architecture is that the RLC layer has not been designed for supporting multicarrier data. In fact, it has to be upgraded for supporting a theoretical maximum data rate of 59.2kbps x 8TS x 8carriers equal to 3788,8kbps and equivalent to 9.472kBytes per 20ms (8 carriers is the maximum due to the coding of the number of carriers on 3 bits, cf. 5.1.1).

5.2.2 Case 2: Multicarrier with carrier dedicated RLC and MAC entities

Considering the protocol architecture presented in Figure 21, on the user plane, the RB user data flow is transferred to the PDCP layer. Then, the PDCP demultiplexes its packet data units to an available RLC/MAC entity. Each entity is dedicated to a carrier. Inside each RLC/MAC entity, the RLC blocks are conveyed to the MAC layer through a Temporary Block Flow as in current standard. Then, each MAC transfers RLC/MAC blocks through logical channels to the Physical layer and configures the mapping of those logical channels onto the physical resources dedicated to its RLC/MAC entity.

User-plane data

Multicarrier signal

PDCP

RLC

MAC

Physical

PDCP

RLC

MAC

TBF

RLC RLC

MAC MAC

TBF TBF

Logical Channels

RRC

Control-plane data

user flowcontrol flowservice access pointRLC/MAC stack

Figure 21: Radio protocols architecture for multicarrier in packet switched domain with carrier dedicated RLC and MAC entitie s.

Since RLC and MAC entities are similar to single carrier RLC and MAC, this architecture requires limited modification to current standard. Yet, in order to transfer PDCP Packet Data Units to several RLC/MAC entities, a new multiplexing function has to be implemented in the PDCP layer. A PDCP PDU buffer will also be needed for guaranteeing the order of PDCP SDUs. Consequently, UTRAN and GERAN PDCP would differ. Besides, a RB shall be allowed to have several RLC instances carrying its data.


17-Sep-01 37/61

5.3 MEDIUM ACCESS CONTROL (MAC)

In GSM/EDGE release 5, more than one Radio Access Bearer can be established to a MS at a time. One RB is used to realize each RAB and one TBF may be established for carrying data belonging to this Radio Bearer. Currently, the peak throughput is defined up to 2048 kbps for a RAB in the Channel Request Description information element [9].

5.3.1 Multiple carrier assignment

During multicarrier resource allocation, the network sends Packet Downlink Assignment (PDA) and Packet Uplink Assignment (PUA) messages for allocating several carriers to a user.

Two strategies are available for assigning those PDCHs. On one hand, one packet assignment can allocate all the carriers involved in the multicarrier configuration. On the other hand, the network can send several packet assignments, each packet assigning one carrier of the multicarrier configuration. Table 6 summarizes the assets and drawbacks of each strategy.

Packet Assignment Multiple carrier allocation with one packet assignment for all carriers

Multiple carrier allocation with one packet assignment per carrier

Fields repetition in the message

No fields are repeated but long message.

Some fields repeated in each Packet Assignment

Modifications

New design of Packet Uplink/Downlink Assignment IE

with a carrier structure and a multicarrier frequencies parameter.

No modification needed, the carriers are allocated independently.

Error resistance considering that Packet assignment are coded

over several RLC blocks

Poor error resistance. When an error occurs, there is no allocation. The whole packet has to be resent.

High error resistance. When a packet is lost, some PDCHs may be

allocated on other carriers.

Table 6: Packet assignment.

As a consequence to the previous table, it is suggested to use one assignment per carrier rather than a global assignment for allocating multicarrier resources.

5.3.2 PACCH operations

In multicarrier, as in multislot, the occurrences of Packet Associated Control Channels depend on the link direction and the MAC mode of the Radio Bearer in uplink [1,9,10]. For downlink transfer, PACCHs may occur on any of the allocated PDCHs. If no PDCH is allocated, the PACCH is sent on PCCCH. For transfer, in dynamic and extended dynamic allocation, the network shall transmit PACCH messages on the PDCH carried on the lowest numbered timeslot in the carrier allocation; in fixed allocation, the MS may transmit message on any PDCHs allocated.

5.3.3 Temporary Flow

In GSM/EDGE Release 5, different types of Radio Access Bearers can be multiplexed on one MS. For each RAB, one Radio Bearer, e.g. one configuration of the radio protocols, is selected for realizing the required QoS. Currently, each Radio Bearer can have one TBF to


17-Sep-01 38/61

which is associated a unique Temporary Flow Identifier. Yet, a TFI change may take place before a TBF is reconfigured to other PDCHs where its TFI is in use by another TBF. To reconfigure a TBF, the network sends a Packet Timeslot Reconfigure message to the MS assigning the new TFI value and parameters of the TBF [9].

In the protocol architecture scenario with common RLC and MAC entities shown in section 5.2.1, the TFI value used by a MS for the multicarrier TBF in one direction is the same on all PDCHs over several carriers used by the MS.

In the protocol architecture scenario with carrier dedicated RLC and MAC entities shown in section 5.2.2, a RB shall be allowed to have several TBFs corresponding to the different RLC entities.

5.3.4 Multicarrier TBF establishment

A TBF may be established when traffic has to be sent by the Mobile Station or the Network [9]. For initiating a multicarrier TBF establishment or several single carrier TBFs establishments to a multicarrier MS, the network has to know the MS multicarrier capability before assigning several carriers to the MS in respect to this multicarrier capability. When an RRC connection is established, using the identification information of the MS from the contention resolution procedure, the network checks in its SGSN whether the MS Radio Access Capability has been stored from a previous Attach and Routing Area Update procedure [8]. If the MS Radio Access Capability information is available, the network can initiate the establishment of a multicarrier TBF. If no RRC connection is established or the MS Radio Access Capability information is not available within the network, the network reserves uplink resources for receiving the full radio access capabilities from the MS.

5.3.4.1 TBF establishment initiated by the Mobile Station

In uplink, if there is a packet downlink transfer in progress, the mobile station may request establishment of an uplink transfer by including a Channel Request Description information element in the Packet Downlink ACK/NACK message on an uplink activated PACCH. The Channel Request Description information element specifies the peak throughput and QoS needed for the TBF establishment. On receipt of a Channel Request Description information element in Packet Downlink ACK/NACK message, the network may assign multicarrier radio resources to the MS. The resources are assigned in respect with the MS multicarrier capability by transmitting Packet Uplink Assignment messages or, if a TBF is active, Packet Timeslot Reconfigure messages on the PACCH, or may reject the request.

If there is no packet downlink transfer in progress, different access method can be considered. First, with the one phase access method used in single carrier, limited resources may be allocated for control data [9]. As multicarrier aims at transmitting large amount of user data, this access method cannot be used for multicarrier. Secondly, in the two-phase access method used for transmitting user data on a single carrier, the network may reserve two radio blocks for receiving, in the first, a Packet Resource Request and, in the second, the Additional Radio Access Capacity (ARAC) message or a dummy RLC/MAC control block. If the network already has the MS Radio Access Capability information, current two-phase access, presented in Figure 22, can be used for establishing TBF with multicarrier resources.


17-Sep-01 39/61

MSPacket Channel Request

Packet Uplink Assignmentallocating two radio blocks

PUA(s) allocating resourceson additional carriers

Network

PAGCH

PACCH

PRACH

PACCH

PACCH

Packet Resource Request

Additional Radio Access Capabilitywithout multicarrier class(Optional)

Figure 22: Two-phase access method for multicarrier allocation.

If the network needs the full radio access capability information of the MS with the multicarrier class, current two-phase access, presented in Figure 22, may be used. However, with the addition of the Multicarrier Capability in MS Radio Access Capability information Element (MS RAC IE), there might not be enough room in the PRR+ARAC messages [11]. In that case, an additional mechanism is needed to allow for the transmission of the whole MS RAC IE.

For extending the current single carrier one-phase access method to multicarrier, two solutions can be considered. On one hand, using the codes reserved for future use in the “Number of radio block allocated” field could increase the number of radio blocks reserved by the network in the Packet Uplink Assignment (PUA) message. Yet, sending two blocks instead of one usually would result in having longer and less robust access method. On the other hand, the multicarrier class could be sent after the two-phase access in a new Packet Resource Request message resulting in a three-phase access.

Figure 23 presents the three-phase access method for multicarrier allocation. As, the network does not have the "a priori" knowledge of the MS multicarrier capability; it may assign resources on a single carrier even though it may not fulfil the throughput requested initially by the MS. Then, the MS may send, on an allocated PACCH, another Packet Resource Request including its multicarrier class in order to have its allocated resource increased. Finally, the network assigns additional carriers, according to the MS multicarrier class, by sending a packet assignment message for each additional carrier on the activated PACCH.


17-Sep-01 40/61

Packet ResourceRequest

Packet ChannelRequest

Packet Uplink Assignment

PUA(s) allocating PDCHs on

additional carriers

NetworkMS

PAGCH

PACCH

PACCH

PRACH

PDTCH

PACCH

Packet Resource Request including MSmulticarrier class

Possible data transfer

Packet Uplink Assignmentallocating PDCHs on one carrier

PACCH

Current two-phaseaccess method for

allocating the first carrier

Third phase for allocatingadditional carriers

(Additional Radio Access Capabilities)

PDTCHData transfer

Figure 23 : Three phase access method (extended Two-phase access method) for multicarrier allocation.

5.3.4.2 TBF establishment initiated by the Network

The network initiates a multicarrier packet transfer to a MS with its RRC in Idle mode by sending one or more packet paging request messages on the downlink PPCH. The MS may respond to one packet paging request message by initiating a mobile originated packet transfer as defined in the previous section.

The network directly assigns PDCHs to a MS with its RRC in Connected mode, according to the MS multicarrier capability, in Packet Downlink Assignment messages on an activated PACCH, or PCCCH if there is no uplink packet transfer in progress. Each packet downlink assignment message points on a carrier and includes the carrier in the list of PDCHs that will be used for downlink transfer on it.

5.3.5 Assignment procedures in RRC Cell_shared mode for multicarrier [14]

With the protocol architecture with one RLC entity and one MAC presented in section 5.2.1, a TBF may exist on several carriers. Consequently, channels on an additional carrier may be allocated to an existing TBF, for example in the access methods previously introduced for the multicarrier allocation. Moreover, an allocated carrier may be reconfigured within a TBF without affecting other allocated carriers of the TBF. Finally, following the increase of data rate due to link adaptation, some allocated resources to a TBF may not be needed anymore. Therefore, a carrie r could be released without affecting the other carriers of the TBF and


17-Sep-01 41/61

without releasing the TBF. Yet, for reducing the resources of a TBF, it may be preferable to release the TBF, reconfigure current RB, and to establish a new TBF.

5.3.5.1 Carrier assignment to an existing block flow

For a multicarrier MS, with respect to its multicarrier class, a Packet Downlink/Uplink Assignment message or a Packet Timeslot Reconfigure message may be sent by the network on any activated PACCH to assign an additional carrier to an existing TBF. The multicarrier MS shall use the TFI indicated in the message on current resources of the TBF and the new resources. The frequency of the new carrier shall be indicated with the Frequency Parameters of the message. In order to distinguish a carrier assignment to an existing block flow resulting in an allocation increase from a carrier reassignment, a "multicarrier TBF extension" field has to be added to current messages. On the PACCH, legacy single carrier MS will misinterpret such message since the TFI used identify a multicarrier TBF. The new assigned resource should be added to the TBF allocation directly or after the TBF starting time, if it is present. The previously assigned resources to this TBF should not be released. Allocated channels on a carrier of a TBF shall be valid until a new assignment for this TBF on this carrier is received or the TBF using those resources is terminated.

5.3.5.2 Carrier reassignment

For a multicarrier MS, each allocated carrier to a TBF may be reconfigured without affecting the resources of the TBF on other carriers. To do so, the Packet Timeslot Reconfigure message has to be sent by the network on the PACCH of the carrier intended to be reconfigured. In the Packet Timeslot Reconfigure message, the present Frequency Parameter and Timeslot Allocation indicate the new resource allocated to the TBF on the indicated carrier. The resources, which do not continue in the TBF allocation, should be released without ending the TBF.

5.3.5.3 Carrier release

Within the multicarrier allocation of a TBF, a carrier may have to be dropped without releasing the TBF, for example, because of an increase in the available throughput after a Link Adaptation procedure, or, because of a reduction in the throughput needed for transmission. In order to release a carrier without releasing the TBF, a new signaling message sent by the network has to be designed. Yet, a release of the TBF will most likely occur, followed by the setup of a new RB for realizing the RAB and the establishment of a new TBF.

If a carrier has to be released by MSs using it, the network broadcasts in the cell a Packet PDCH Release to indicate that all the timeslots on this carrier are no longer available for packet data channels.

5.3.6 Link quality measurements

In GSM/EDGE, when a MS send a Packet Resource Request or a Packet Downlink ACK/NACK message on PACCH, those messages may include a Channel Quality Report information element [9,19]. In GPRS, this information element reports the signal level and variance received by the MS on a carrier, the averaged signal level on the common channels and the interference level per timeslot. In EGPRS, this information element compulsorily reports the mean and coefficient of variation of the Bit Error Probability (BEP) measurements for each modulation averaged on all timeslots in the TBF and the averaged signal level on the common channels. Optionally, it may include per-timeslot measurements, i.e. the mean value of BEP measurements for each modulation and the interference level.


17-Sep-01 42/61

With the protocol architecture scenario presented in section 5.2.1, a TBF may have resources on several carriers. Whereas, the link quality measurements aim at estimating the effect of fading like frequency selective fading, the calculation of the averaged mean BEP over a multicarrier TBF would average the frequency selective fading. Therefore, the mean BEP measurement should be averaged per timeslots on each carrier of a TBF, instead of all carriers of a TBF. Consequently, current BEP mechanism could be reused. Thus, Link Adaptation would be made on a carrier basis (See next section).

5.4 RADIO LINK CONTROL (RLC)

The concept of multicarrier slightly affects the RLC protocol. Although no new RLC function is added to the RLC layer defined in the GSM/EDGE standard, the multicarrier capability requires some modifications, mainly for supporting the increase in data rates and the possibility to have multicarrier TBF.

5.4.1 Link Adaptation

In current Link Adaptation (LA), one MCS is defined for all channels allocated to a TBF based on their link quality measurements (presented in part 5.3.6). Yet, with the protocol architecture presented in section 5.2.1, a TBF may have multicarrier resources. As a part of the fading affecting the propagation channel is frequency selective, the LA should independently define a MCS for the allocated channels of each carrier in a multicarrier TBF.

Although, for EGPRS, the measurement reports already include BEP measurements for each timeslot of the allocated carrier, having a LA mechanism defining MCS for each timeslot would require longer signaling messages. Indeed, an MCS would have to be commanded (Packet Uplink Assignment, Packet Uplink ACK/NACK, Packet Timeslot Reconfigure) fo r every single timeslot of the allocation, whereas today one single MCS is commanded that is used by all the timeslots of the allocation.

5.4.2 Block Sequence Number and RLC Window Size

A TBF is established between two RLC end points where the data unit is the RLC block. The RLC end points of a TBF use a block sequence number (BSN) field coded on 11 bits (in EGPRS) for sorting the RLC blocks. Indeed, in order to sort the RLC block at the reception, each RLC data block contains a BSN, which corresponds to the order of the RLC blocks at the transmission. Moreover, the RLC end points of a TBF use a window mechanism for acknowledgment in order to reduce the number of unacknowledged RLC blocks kept by the sender in buffers. In fact, each receiver has a receive window of size (WS) and each transmitter has a transmit window of size WS defined in Annex C for EGPRS. To enable the transmission of several blocks without having to wait for the acknowledgment of each block, the window size must be large enough compared to the resource used.

In the protocol architecture presented in section 5.2.1, one RLC entity controls the RLC blocks coming from all the allocated carriers. Thus, the number of blocks to acknowledge increase with the number of timeslots allocated. Therefore, the WS has to be extended in proportion to the data that has to be acknowledged. Yet, the WS is limited to the half of the highest BSN value in order to have the acknowledgement window in one piece. Consequently, the BSN has to be extended. Downlink RLC/MAC headers do not contain any spare bits that would allow the BSN extension. Thus, the BSN extension would require new EGPRS RLC/MAC headers and new MCS's. Stealing flags in the four bursts of each radio blocks distinguish the EGPRS RLC/MAC header types. However, at similar protection level,


17-Sep-01 43/61

there is no code left for new EGPRS RLC/MAC header type. Therefore, new MCSs cannot be created for supporting multicarrier BSN without modifying current physical layer [16]. Besides, Packet Downlink Ack/nack would not be sufficient for acknowledging that larger window.

In the protocol architecture presented in section 5.2.2, the data flow in each RLC/MAC entity is similar to the data flow in the single carrier MS. Therefore, it is not necessary to change the BSN and the RLC Window Size.

5.4.3 TBF Release

With the protocol architecture presented in section 5.2.1, the TBF release may occur on any carrier of the TBF.

5.4.4 RLC/MAC control messages

Currently, control messages are designed for single carrier TBF. With the protocol architecture presented in section 5.2.2, each RLC/MAC entity can be considered as a single carrier MS, therefore most of the current control messages and procedures will not need modification.

If the architecture presented in section 5.2.1 using multicarrier TBF is selected, control messages will have to be modified for supporting multicarrier. Indeed, since the MAC layer provides the RLC with a multicarrier TBF, at the reception of current control messages, the RLC layer can not know which carrier is concerned by each control message. Therefore, in carrier dedicated RLC/MAC control messages present ed in Table 7, a carrier identifier field shall be used for distinguishing, on uplink or downlink, the carrier concerned. In Annex D, an example of modified RLC/MAC control message is presented. Such a change would also allow control messages for all the multicarrier resources of a TBF to be sent on a "master" PACCH.

Direction RLC/MAC control messages sent on PACCH

Downlink

§ Packet Downlink Assignment

§ Packet Power Control / Timing Advance

§ Packet Uplink Ack/Nack

§ Packet Uplink Assignment

§ Packet Timeslot Reconfigure

Uplink

§ Packet Downlink Ack/Nack

§ EGPRS Packet Downlink ACK/NACK

§ Packet Measurement Report

§ Packet Enhanced Measurement Report

Table 7: List of RLC/MAC Control Messages, which shall include a carrier identifier.

5.5 PACKET DATA CONTROL PROTOCOL (PDCP)

Considering the different protocol architecture scenarios presented in part 5.2, the first scenario with common RLC and MAC has no impact on the PDCP layer since the functions needed are the same as in the single carrier case [14,21]. However, the second scenario with


17-Sep-01 44/61

separate RLC/MAC entities requires a multiplexing function in the PDCP layer. This new function should enable the PDCP to deliver PDCP PDUs to the first RLC/MAC entity available for transmission. For identifying the RLC/MAC entities, different access points should be designed between the PDCP and the different RLC/MAC entities. Finally, a PDCP PDU buffer will also be needed for guaranteeing the order of PDCP SDU. Consequently, UTRAN and GERAN PDCP would differ.

5.6 RADIO RESOURCE CONTROL (RRC)

The Radio Resource Control (RRC) is a control plane protocol for the radio resource management whose functions are partially defined in the standard. Therefore, the possible impacts of the multicarrier concept on the RRC are left for further study. Yet, slight modifications of some RRC functions can be forecast for:

§ The assignment, reconfiguration and release of multicarrier radio resources for the RRC connection.

§ For multicarrier Downlink transfer, the power control at the MS and the power measurement at the BSS are made separately for each carrier. For multicarrier Uplink transfer, as the transmitter may have one antenna, the signal power amplification would be achieved once. Therefore, the power control should be valid for all carriers and power measurements could be averaged.

§ The control of the MS measurement reporting for each carrier in the allocation.

§ Considering the different protocol architecture scenarios presented in part 5.2, a RB shall be allowed to have several RLC instances carrying its data.


17-Sep-01 45/61

6 SUMMARY OF RADIO PROTOCOLS CHANGES

§ Multicarrier MS receiver should preferably include one antenna and multip le receiving chains.

§ Radio planning shall have a separation between hopping frequencies for single carrier MSs, and non-hopping (or, in some conditions, restricted hopping) frequencies for multicarrier MSs.

§ Inter- frequency interleaving can be an optional feature for multicarrier MS. Link adaptation and power control constraints have to be taken into account. Only data shall be interleaved and a separate interleaving is required for GMSK and 8-PSK.

§ Changes per layer and per radio protocol scenario:

Layer Subject Multicarrier with global RLC and

MAC entities. Data multiplexing at the MAC layer

Multicarrier with dedicated RLC and MAC entities.

Data multiplexing at the PDCP layer

R R C

Radio Bearer - A RB can have several TBFs

P D C P

UTRAN/GERAN misalignment

- Addition of a multiplexing function

and a PDCP PDU buffer

Link Adaptation MCS assignment per carrier instead of

current assignment per TBF -

R L C BSN and RLC

window

Need of new MCS for sending information of increased BSN and

RLC window in the header part -

TBF Several carriers can be allocated to a TBF

Link quality measurements

Measurement per carrier instead of current measurement per TBF -

M A C

Control messages Multicarrier class -

Control procedures TBF establishment procedure for multicarrier MS

Control messages Addition of a carrier identifier field in some control messages

RLC &

MAC

MCS Need to create new MCS -

Phys.


17-Sep-01 46/61

7 CONCLUSION

The increase of GSM/EDGE user peak data rate based on transmission over multiple carriers, instead of one single carrier, can be considered as a trivial concept. However, to ensure the feasibility of a multiple carrier technique, the modifications to GSM/EDGE standard have to be minimized, and the backward compatibility with the previous releases has to be kept. In this report, those issues were studied from the radio protocol perspective.

At first, the context and scope of the multicarrier study were defined. Basically, reference concepts of packet data transfer and GSM/EDGE standard were presented.

The second focus area was to describe the multicarrier physical layer. As a basic case, it was assumed that the multicarrier MS receiver would use one antenna. In the MS receiver, the number of decoding chains and the bandwidth of the multicarrier filter would determine the MS multicarrier capability. Concerning link level performance, current frequency hopping could not be applicable to a multicarrier MS with such receiver. Thus, a separation between hopping frequencies for single carrier MSs and non-hopping frequencies for single carrier and multicarrier MSs was advised. Although interleaving data over several carriers might theoretically provide performance gain, consequent performance degradation with link adaptation and power control reduce the interest fo r such interleaving method. Concerning the allocation of physical resources, consequent time slots should preferably be allocated to multicarrier MS to minimize the MS complexity in terms of supported time slots.

The main focus was to study changes to current radio protocols and signaling for supporting the multicarrier concept. Based on the physical layer conclusions, multicarrier classes were defined. Concerning TBF establishment, as the addition of a multicarrier class may not always be applicable to the current access methods, a three-phase access procedure was defined for allocating multicarrier resources, and generally for signalling future MS capabilities. Moreover, to transfer data in multiple carriers, radio bearer data should be (de)multiplexed over (from) the different carriers. The (de)multiplexing function could take place either in the MAC layer or in the PDCP layer resulting in two conceivable scenarios. With multicarrier demultiplexing located in the MAC layer, radio bearer data should be transferred in a single block flow. To support single block flow transfer onto multicarrier resources, large changes of RLC, MAC and Physical layers would be required. On contrary, with the PDCP demultiplexing of Radio Bearer data in multiple block flows, few modifications to current GSM/EDGE standard would be needed. However, the introduction of a multiplexing functionality in PDCP for the (de)multiplexing use would result in a misalignment between GERAN and UTRAN.

On the radio protocol perspective, the multicarrier in GSM/EDGE is a feasible technique for increasing user data rates. To assess the overall feasibility of the multicarrier concept in GSM/EDGE, further studies are needed especially in the areas of MS RF parts and radio performance.


17-Sep-01 47/61

8 REFERENCES

[1] ETSI, GPRS General description of GPRS radio interface, GSM 03.64 v8.5.0, 07/2000.

[2] Jean Walrand, Communication Networks a first course, WCB McGraw-Hill 1998.

[3] Theodore Rappaport, Wireless Communications Principle & Practice, Prentice Hall edit ion, 1998.

[4] Michel Mouly, Marie-Bernadette Pautet, GSM System for Mobile communications, Edition Michel Mouly & Marie-Bernadette Pautet,1992.

[5] Nokia web site, http://www.nokia.com/networks/17/edge.html, 2000.

[6] ETSI, GPRS Service description Stage 2, GSM 03.60 version 7.4.0, 2000.

[7] Nokia web site, http://www.nokia.com/networks/17/gprs/index.html, 2000.

[8] 3GPP TSG GERAN, Base Station System (BSS) - Serving GPRS Support Node (SGSN); BSS GPRS Protocol (BSSGP), 3GPP TS 08.18 v8.6.0, 2001.

[9] ETSI, GPRS MS-BSS RLC/MAC protocol, GSM 04.60 v8.5.0, 07/2000.

[10] 3GPP TSG GERAN, MAC design for Geran'00, 3GPP Tdoc 2g00-103, 2000.

[11] ETSI SMG2, Concept proposal for enhanced GPRS, Tdoc SMG2 657/99, 06/ 1999.

[12] 3GPP, Mobile radio interface layer 3 specification, 3G TS 24.008 v3.4.1, 2000.

[13] 3GPP, Multiplexing and multiple access on the radio path, TS 45.002 v4.0.1, 10/2000.

[14] 3GPP TSG GERAN, GERAN Overall Description stage 2, 3GPP TS 43.051 v4.0.0, 2000.

[15] Eero Nikula, Shkumbin Hamiti, Janne Parantainen, Timo Rantalainen, Benoist Sébire, Guillaume Sébire, Evolution of GSM/EDGE towards 3G mobile services, Nokia Research Center, 2000.

[16] ETSI, Channel Coding, GSM 05.03 v8.5.0, 2000.

[17] Harri Posti, Performance requirement and design specification of multicarrier GSM base station radios, thesis at the University of Oulu, dept. of Electrical Engineering, 2000.

[18] ETSI, Associated Control Channels for GERAN Radio Access Bearers, Tdoc 2e00, 2000.

[19] Yuping Zhao, Guillaume Sébire and Eero Nikula, Advanced Measurement for EGPRS Link Quality Control, Nokia Research Center 2000.

[20] 3GPP TSG GERAN, Radio subsystem link control, 3GPP TS 45.008 v4.0.1, 2000.

[21] 3GPP, Packet Data Convergence Protocol specification, 3GPP TS 25.323v 4.0.0, 2001


17-Sep-01 48/61

ANNEX A : GSM/EDGE ACCESS METHODS

The information enclosed in this annex are coming form 3GPP 03.64 v8.5.0 related to GSM/EGDE release 4.

A.1. UPLINK ACCESS FOR MS IN PACKET IDLE MODE

An MS initiates a packet transfer by making a Packet Channel Request on PRACH or RACH. The network responds on PAGCH or AGCH respectively. It is possible to use one or two phase packet access method as shown in Figure 24.

MS Network

Packet Channel Request


Packet Resource Request


PRACH (or RACH)

PAGCH (or AGCH)

PACCH

PACCH

(Optional)

(Optional)

Figure 24: Access and allocation for the one or two phase packet access, uplink packet transfer.

In the one phase access, the network responds the Packet Channel Request with the Packet Uplink Assignment reserving the resources on PDCH(s) for uplink transfer of a number of Radio blocks. The reservation is done according to the information about the requested resources that is comprised in the Packet Channel Request. On RACH, there are only two cause values available for denoting GPRS, which can be used to request limited resources or two phase access. On PRACH, the Packet Channel Request may contain more adequate information about the requested resources and, consequently, uplink resources on one or several PDCHs can be assigned by using the Packet Uplink Assignment message.

In the two-phase access, the Packet Channel Request is responded with the Packet Uplink Assignment, which reserves the uplink resources for transmitting the Packet Resource Request. The network or a mobile station can initiate a two-phase access. The network can order the MS to send Packet Resource Request message by setting parameter in Packet Uplink Assignment message. Mobile station can require two-phase access in Packet Channel Request message. In this case, the network may order MS to send Packet Resource Request or continue with a one-phase access procedure.

The Packet Resource Request message carries the complete description of the requested resources for the uplink transfer. The MS can indicate the medium access method, it prefers to be used during the TBF. The network responds with the Packet Uplink Assignment reserving resources for the uplink transfer and defining the actual parameters for data transfer (e.g. medium access mode).


17-Sep-01 49/61

A.2. DOWNLINK ACCESS

The network initiates a packet transfer to an MS that is in the Standby state by sending one or more packet paging request messages on the downlink PPCH or PCH. The MS responds to one packet paging request message by initiating a mobile originated packet transfer. This mobile originated packet transfer allows the MS to send a packet paging response to the network. The packet paging response is one or more RLC/MAC data blocks containing an arbitrary LLC frame. The message sequence described in Figure 25 below is conveyed either on PCCCH or on CCCH. After the packet paging response is sent by the MS and received by the network, the mobility management state of the MS is Ready.

MS Network

Packet Paging Request (or Paging Request) PPCH (or PCH)

Packet Channel Request (or Channel Request)PRACH (or RACH)

Packet Uplink Assignment (or Immediate Assignment)PCCCH (or CCCH)

Packet Resource RequestPACCH(Optional)

Packet Uplink AssignmentPACCH(Optional)

Packet Paging Response (LLC frame)PDTCH

…

Figure 25: Paging message sequence for Paging, downlink packet transfer.

The network using a packet downlink assignment message initiates the transmission of a packet to an MS in the Ready state. In case there is an uplink packet transfer in progress, the packet downlink assignment message is transmitted on PACCH. In case there is PCCCH allocated in the cell, the Packet Downlink Assignment message is transmitted on PCCCH. Otherwise, the Immediate Assignment message is transmitted on CCCH. The packet downlink assignment message includes the list of PDCH(s) that will be used for downlink transfer. The Timing Advance and Power Control information is also included, if available. Otherwise, the MS may be requested to respond with a Packet Control Acknowledgement. The MS multislot capability needs to be considered.

Figure 26 shows an example of message sequence for (multislot) downlink data transfer with one resource reallocation and possible RLC Data Block re-transmissions.


17-Sep-01 50/61

MS Network

Packet Downlink Assignment PACCH, (P)CCCH

Packet Control AcknowledgementPACCH

Data BlockPDTCH

Packet Downlink Ack/NackPACCH

(Optional)

Data BlockPDTCH

Data Block (polling)PDTCH

Data BlockPDTCH

Packet Control AcknowledgementPACCH

Data BlockPDTCH

Data BlockPDTCH

Data BlockPDTCH

Packet Downlink Ack/Nack (final)PACCH

Data BlockPDTCH

Data Block (last, polling)PDTCH

Packet Downlink Assignment (polling)PACCH

Figure 26: An example of downlink data transfer


17-Sep-01 51/61

ANNEX B : MULTISLOT CAPABILITY

B.1. MS CLASSES FOR MULTISLOT CAPABILITY

When a MS supports the use of multiple timeslots it shall belong to a multislot class defined below [13].

Multislot Maximum number of slots Minimum number of slots Type class Rx Tx Sum Tta Ttb Tra Trb

1 1 1 2 3 2 4 2 1 2 2 1 3 3 2 3 1 1 3 2 2 3 3 2 3 1 1 4 3 1 4 3 1 3 1 1 5 2 2 4 3 1 3 1 1 6 3 2 4 3 1 3 1 1 7 3 3 4 3 1 3 1 1 8 4 1 5 3 1 2 1 1 9 3 2 5 3 1 2 1 1 10 4 2 5 3 1 2 1 1 11 4 3 5 3 1 2 1 1 12 4 4 5 2 1 2 1 1 13 3 3 NA NA a) 3 a) 2 14 4 4 NA NA a) 3 a) 2 15 5 5 NA NA a) 3 a) 2 16 6 6 NA NA a) 2 a) 2 17 7 7 NA NA a) 1 0 2 18 8 8 NA NA 0 0 0 2 19 6 2 NA 3 b) 2 c) 1 20 6 3 NA 3 b) 2 c) 1 21 6 4 NA 3 b) 2 c) 1 22 6 4 NA 2 b) 2 c) 1 23 6 6 NA 2 b) 2 c) 1 24 8 2 NA 3 b) 2 c) 1 25 8 3 NA 3 b) 2 c) 1 26 8 4 NA 3 b) 2 c) 1 27 8 4 NA 2 b) 2 c) 1 28 8 6 NA 2 b) 2 c) 1 29 8 8 NA 2 b) 2 c) 1

Table 8: MS classes for multislot capability

Type 1 MS are not required to transmit and receive at the same time Type 2 MS are required to be able to transmit and receive at the same time. a) = 1 with frequency hopping. = 0 without frequency hopping. b) = 1 with frequency hopping or change from Rx to Tx. = 0 without frequency hopping and no change from Rx to Tx. c) = 1 with frequency hopping or change from Tx to Rx. = 0 without frequency hopping and no change from Tx to Rx. Sum: Sum is the total number of uplink and downlink TS that can actually be used by the MS per

TDMA frame. The MS must be able to support all combinations of integer values of Rx and Tx TS where 1 <= Rx + Tx <= Sum (depending on the services supported by the MS). Sum is not applicable to all classes.

Tta: Tta relates to the time needed for the MS to perform adjacent cell signal level measurement and

get ready to transmit.


17-Sep-01 52/61

Ttb: Ttb relates to the time needed for the MS to get ready to transmit. Tra: Tra relates to the time needed for the MS to perform adjacent cell signal level measurement and

get ready to receive. Trb: Trb relates to the time needed for the MS to get ready to receive.

B.2. NETWORK REQUIREMENTS FOR SUPPORTING MS MULTISLOT CLASSES

The multislot class of the MS will limit the combinations and configurations allowed when supporting multislot communication.

It is necessary for the network to decide whether the MS needs to perform adjacent cell power measurement for the type of multislot communication intended and whether the service imposes any other constraints before the full restrictions on TS assignments can be resolved.

This is best shown by example:

For a multislot class 5 MS in circuit switched configuration (adjacent cell power measurements required) five basic configurations of channels are possible which can occur in six different positions in the TDMA frame. The service itself may determine that asymmetry must be downlink biased, in which case the last two solutions would not be allowed.


17-Sep-01 53/61

Rx

Tx

Rx=2

Tt=1 Tx=2

Rx

Tx

Rx=2

Tt=1

Tx<2

Rx

Tx

Rx<2 Tt>1 Tx=2

These five combinations can be repeated at the six other positions that can be fitted

within the same TDMA frame

Rx

Tx

Rx=2 Tt>1

Tx<2

Tra>3

Tra=3 Tra=3

Tra=3

Rx

Tx

Rx<2 Tx=2 Tra>3

Tt=1

All possible timeslots used Downlink biased assymetry

Alternative downlink biased assymetry Uplink biased assymetry(not prohibited by multislot class)

Alternative uplink biased assymetry(not prohibited by multislot class)

Figure 27: Basic configurations of channels are possible for a multislot class 5 MS

For a multislot class 13 MS when adjacent cell power measurements are not required and the service does not constrain the transmit and receive timeslots to use the same timeslot number. Many configurations of channels are possible as long as the 5 constraints of the MS are catered for.


17-Sep-01 54/61

Rx

Tx

Rx=3

Tx=3

Rx

Tx

Trb=1 Tt

=1

There is no requirement for relative timingof Tx timeslots in relation to Rx timeslots

Many configurations are possible

Rx

Tx

Rx

Tx

Rx

Tx

This configuration could be used forHSCSD or GPRS

Figure 28: Basic configurations of channels are possible for a multislot class 13 MS


17-Sep-01 55/61

ANNEX C : EGPRS RLC WINDOW SIZE

For EGPRS, the window size (WS) shall be set by the network according to the number of timeslots allocated in the direction of the TBF (uplink or downlink). The allowed window sizes are given in Table 9. Preferably, the selected window size should be the maximum, or follow the definition in Table 10.

The acknowledgment window size may be set independently on uplink and downlink [9]. MS shall support the maximum window size corresponding to its multislot capability, Annex B. The selected window size shall be indicated within PACKET UL/DL ASSIGNMENT and PACKET TIMESLOT RECONFIGURE using the coding defined in Table 9.

Timeslots allocated (Multislot capability) Window size

Coding 1 2 3 4 5 6 7 8

64 00000 96 00001

128 00010 160 00011 192 00100 Max 224 00101 256 00110 Max 288 00111 320 01000 352 01001 384 01010 Max 416 01011 448 01100 480 01101 512 01110 Max 544 01111 576 10000 608 10001 640 10010 Max 672 10011 704 10100 736 10101 768 10110 Max 800 10111 832 11000 864 11001 896 11010 Max 928 11011 960 11100 992 11101 1024 11110 Max

Reserved 11111 x x x x x x x X

Table 9: Allowed window sizes in EGPRS TBF mode for different multislot allocations.

Although for each multislot allocation, the selected window size could preferably be the maximum, a smaller window size may be selected in order to optimize e.g. the number of (multislot) users and network memory consumption.

However, for each MS, in order to meet a performance, which corresponds to the number of timeslots allocated to this MS, the selected window size shall not be smaller than a minimum window size for this particular multislot allocation.


17-Sep-01 56/61

For each network, the round-trip delay has a direct implication on the performance, hence on the definition of the minimum window sizes. Consequently, no generic minimum window sizes are suggested. However, for information, the table below lists the window size ranges recommended with a round-trip delay of about 120ms.

Timeslots allocated (Multislot capability) Window size

Coding 1 2 3 4 5 6 7 8

64 00000 Min 96 00001 Min 128 00010 160 00011 Min Min 192 00100 Max 224 00101 Min 256 00110 Max 288 00111 320 01000 Min 352 01001 Min 384 01010 Max 416 01011 448 01100 480 01101 512 01110 Max Min 544 01111 576 10000 608 10001 640 10010 Max 672 10011 704 10100 736 10101 768 10110 Max 800 10111 832 11000 864 11001 896 11010 Max 928 11011 960 11100 992 11101 1024 11110 Max

Reserved 11111 x X x x x x x X

Table 10: Allowed window sizes in EGPRS TBF mode for different multislot allocations with preferable minimum value in order to avoid reducing the RLC window efficiency.


17-Sep-01 57/61

ANNEX D : MODIFIED PACKET DOWNLINK ASSIGNMENT

This message is sent on the PCCCH or PACCH by the network to the mobile station to assign downlink resources to the mobile station.

For a mobile station assigned to operate in the fixed allocation MAC mode, the network may assign regularly repeating intervals during which the mobile station shall measure neighbor cell power levels. A mobile allocation or reference frequency list received as part of this assignment message shall be valid until a new assignment is received or each TBF of the MS are terminated.

Message type: PACKET DOWNLINK ASSIGNMENT

Direction: network to mobile station

Classification: non-distribution message

Table 11: PACKET DOWNLINK ASSIGNMENT information elements < Packet Downlink Assignment message content > ::= < PAGE_MODE : bit (2) > { 0 | 1 <PERSISTENCE_LEVEL : bit (4) > * 4 } { { 0 < Global TFI : < Global TFI IE > > | 10 < TLLI : bit (32) > } { 0 -- Message escape { < MAC_MODE : bit (2) > < RLC_MODE : bit (1) > < CONTROL_ACK : bit (1) > < TIMESLOT_ALLOCATION : bit (8) > < Packet Timing Advance : < Packet Timing Advance IE > > { 0 | 1 < P0 : bit (4) > < BTS_PWR_CTRL_MODE : bit (1) > < PR_MODE : bit (1) > } { { 0 | 1 < Frequency Parameters : < Frequency Parameters IE > > } { 0 | 1 < DOWNLINK_TFI_ASSIGNMENT : bit (5) > } { 0 | 1 < Power Control Parameters : < Power Control Parameters IE > > } { 0 | 1 < TBF Starting Time : < Starting Frame Number Description IE > > } { 0 | 1 < Measurement Mapping : < Measurement Mapping struct > > } { null | 0 bit** = <no string> -- Receiver backward compatible with earlier version | 1 -- Additional contents for Release 1999 { 0 | 1 < EGPRS Window Size : < EGPRS Window Size IE >> < LINK_QUALITY_MEASUREMENT_MODE : bit (2) > { 0 | 1 < BEP_PERIOD2 : bit(4) > }} { 0 | 1 <Packet Extended Timing Advance : bit (2)> } { 0 | 1 < COMPACT reduced MA : < COMPACT reduced MA IE >> } { 0 | 1 < Carrier Identifier : bit (3) > } < padding bits > }} // -- truncation at end of message allowed, bits ‘0’ assumed ! < Non-distribution part error : bit (*) = < no string > > } ! < Message escape : 1 bit (*) = <no string> > } ! < Address information part error : bit (*) = < no string > > } ! < Distribution part error : bit (*) = < no string > > ; < Measurement Mapping struct > ::= < Measurement Starting Time : < Starting Frame Number Description IE > > < MEASUREMENT_INTERV AL : bit (5) > < MEASUREMENT_BITMAP : bit (8) > ;

Table 12: PACKET DOWNLINK ASSIGNMENT information element details

PAGE_MODE (2 bit field) This field is defined in clause 12.20.


17-Sep-01 58/61

PERSISTENCE_LEVEL (4 bit field for each Radio Priority 1…4) This field is defined in subclause 12.14, PRACH Control Parameters.

Referenced address struct This information element contains the address of the mobile station addressed by the message.

Global TFI This information element contains the TFI of the mobile station's downlink TBF or Uplink TBF. This field is defined in clause 12.10.

TLLI (32 bit field) This field is defined in subclause 12.16.

MAC_MODE (2 bit field) This information field indicates the medium access method to be used during the TBF.

bit 2 1 0 0 Dynamic Allocation 0 1 Extended Dynamic Allocation 1 0 Fixed Allocation, not half duplex mode 1 1 Fixed Allocation, half duplex mode

RLC_MODE (1 bit field) This field indicates the RLC mode of the requested TBF.

0 RLC acknowledged mode 1 RLC unacknowledged mode

CONTROL_ACK (1 bit field) This field shall be set to '1' if the network establishes a new downlink TBF for the mobile station whose timer T3192 is running. Otherwise this field shall be set to '0'.

TIMESLOT_ALLOCATION (8 bit field) This field is defined in subclause 12.18.

Packet Timing Advance This information element is defined in subclause 12.12.

P0 (4 bit field) For description and encoding, see the Packet Uplink Assignment message.

BTS_PWR_CTRL_MODE (1 bit field) For description and encoding, see the Packet Uplink Assignment message.

PR_MODE (1 bit field) For description and encoding, see the Packet Uplink Assignment message.

Power Control Parameters This information element is defined in subclause 12.13.

Frequency Parameters This information element is defined in subclause 12.8.


17-Sep-01 59/61

DOWNLINK_TFI_ASSIGNMENT (5 bit field) This information element, if present, assigns the TFI to the mobile station to identify to downlink TBF described by this message. TFI is encoded as defined in subclause 12.15.

TBF Starting Time The TBF Starting Time field contains a starting time that indicates the TDMA frame number during which the assigned TBF may start. If no downlink TBF is in progress, the mobile station need not monitor the TFI field of downlink RLC data blocks until the indicated TDMA frame number. After the indicated TDMA frame number, the mobile station shall operate as during a downlink TBF. If a downlink TBF is already in progress, the mobile station shall continue to use the parameters of the existing TBF until the TDMA frame number occurs. When the indicated TDMA frame number occurs, the mobile station shall immediately begin to use the new parameters assigned. This information element is defined in subclause 12.21.

Measurement Starting Time The Measurement Starting Time field contains a starting time that indicates the frame number during which the first assigned measurement period shall occur. The mobile station must make one or more neighbor cell power measurements during the assigned frame number and during the following 3 TDMA frames. This information element is defined in subclause 12.21.

MEASUREMENT_BITMAP (8 bit field) This information field indicates the timeslots assigned for use during measurement periods. The field as a bitmap where each bit corresponds with a timeslot number. Bit 1 corresponds to TS0; Bit 2 to TS1...

0 the MS shall receive downlink data during this timeslot 1 the MS shall make measurements during the timeslot

MEASUREMENT_INTERVAL (5 bit field) The Measurement Interval field indicates the number of block periods from the start of one assigned measurement period to the beginning of the next measurement period.

bit 5 4 3 2 1 0 0 0 0 0 make measurements during every block period 0 0 0 0 1 make measurements during every other block period 0 0 0 1 0 make measurements during every 3rd block period . . . 1 1 1 1 1 make measurements during every 32nd block period

EGPRS Window Size This information element is defined in subclause 12.5.2.


17-Sep-01 60/61

LINK_QUALITY_MEASUREMENT_MODE (2 bit field)

This field determines the measurements to be included within the EGPRS Timeslot Link Quality Measurements IE.

bit 2 1

0 0 The MS shall not report either interference measurements (γ values) or per slot BEP measurements.

0 1 The MS shall report available interference measurements (γ values) for timeslots 0 through 7. The γ value is defined in 3GPP TS 05.08. No per slot mean BEP measurements shall be reported.

1 0 The MS shall report mean BEP on each allocated time slot. The MS shall report the mean BEP measurement corresponding to the modulation for which it has received a larger number of blocks since the previous report. The MS shall make BEP measurements only on Radio Blocks intended for it. No interference measurements (γ values) shall be reported.

1 1 The MS shall report mean BEP on each allocated time slot. The MS shall report the mean BEP measurement corresponding to the modulation for which it has received a larger number of blocks since the previous report. The MS shall make BEP measurements only on Radio Blocks intended for it. In addition to mean BEP, the MS shall report interference measurements (γ values) for no more than four time slots. The MS shall first report available interference measurements for time slots 0, 1, 2, and 3; the MS shall next report available interference measurements for time slots 4, 5, 6, and 7; and the MS shall alternate between these two groups for subsequent reports. If any of the interference measurements are unavailable for reporting for reasons specified in 3GPP TS 05.08, the MS shall substitute least-recently-reported and available interference measurements for time slots not already included in the report.

All other values are reserved.

Packet Extended Timing Advance (2 bit field) This field is defined in subclause 12.12b.

COMPACT reduced MA

This information element is defined in subclause 12.29.


17-Sep-01 61/61

BEP_PERIOD2 (4 bit field) This field contains a constant which is used for filtering channel quality measurements in EGPRS. BEP_PERIOD2 when present, or if not, when received in a previous message of the same TBF session, shall be used instead of BEP_PERIOD. For details see 3GPP TS 05.08.

Range: 0 to 15

Carrier Identifier (3 bits field)

This field enable the numbering and the identification of a carrier within a multicarrier TBF.

bit 3 2 1 0 0 0 Carrier number 1 0 0 1 Carrier number 2 . . . 1 1 1 Carrier number 8

edge multicarrier

Documents