gprs traffic modelling.doc

7/29/2019 GPRS Traffic Modelling.doc

1/55

Commercial in Confidence Page 1

GPRSTraffic Modelling

P-0234-0802

Rev: 1.0

Date: 31 May 2001

RADIO ENGINEERING SOLUTIONS


2/55

DOCUMENTREVISIONHISTORY

Rev Date Author Reason for Change1.0 Hamid Akhrif Initial Draft



3/55

CONTENTS

LIST OF FIGURES.5

LISTOF FIGURES

FIGURE 1 GPRS BILLING..............................................................................................................................7

FIGURE 2 CHARGINGFORGPRS..................................................................................................................9

FIGURE 3 PCU CIRCUIT/ PACKET DATA SEPARATION..............................................................................12

FIGURE 4 PCU CONFIGURATION................................................................................................................13

FIGURE 6 ADDITIONAL ELEMENTSOFTHE GPRS NETWORK....................................................................17

FIGURE 7GPRS ASSOCIATED INTERFACES................................................................................................19

FIGURE 8 UM AIRINTERFACE....................................................................................................................21

FIGURE 9 THE UM INTERFACE RLC LAYER..............................................................................................23

FIGURE 10 RLC BLOCKGENERATION.......................................................................................................25

FIGURE 11 THE UM INTERFACE MAC LAYER...........................................................................................26

FIGURE 12 MAC FRAMES GENERATION...................................................................................................28



4/55

FIGURE 13 THE UM INTERFACE LINKLAYER............................................................................................29

FIGURE 14 BURST FORMATTINGINTHE UM INTERFACE...........................................................................30

FIGURE 15 BURST INTERLEAVING IN THE UM INTERFACE........................................................................31

FIGURE 16 GPRS LOGICAL CHANNELS.....................................................................................................32

FIGURE 17 PBCCH/PCCCH LOGICAL CHANNELS....................................................................................33

FIGURE 18 PDTCH/PDCCH LOGICAL CHANNELS....................................................................................33

FIGURE 19 USING SPARE GSM CAPACITY................................................................................................37

FIGURE 20 UPLINKSTATE FLAG ...............................................................................................................41

FIGURE 21 MS- INITIATED TBF ESTABLISHMENT.....................................................................................44

FIGURE 22 NETWORK-INITIATED TBF ESTABLISHMENT...........................................................................45

FIGURE 23 GPRS CODING SCHEMES.........................................................................................................48

FIGURE 24 C/I COVERAGE.........................................................................................................................49

FIGURE 25 MAXIMUM THROUGHPUT VERSUS C/I IN FREQUENCY HOPPING............................................50

FIGURE 26 MAXIMUM THROUGHPUT VERSUS C/I WITH NO FREQUENCY HOPPING...............................51

1 Introduction

The scope of this document is to describe the impact of the implementation ofGPRS on a GSM network and highlight the traffic allocation and radio resourcemanagement.

2 Purpose of GPRS



5/55

GPRS provides packet data Radio access for GSM mobile phones. GPRS iswell adapted to burst data applications and it upgrades GSM data services toallow an interface with Local Area Networks (LANs), Wide Area Networks(WANs), and the internet.

GPRS uses the radio interface efficiently in two ways. Firstly, it enables a fastmethod for reserving radio channels. Secondly, the benefit of GPRS is thesharing of resources with circuit switched connection. GPRS packet can betransmitted in the free periods between circuit switched calls. Furthermore,GPRS provides immediate connectivity and high throughput.

On a general level, GPRS connections use the resources for only a short timewhen they are sending or receiving data. When the user is ready to receive newdata the terminal sends a request, and resources are again reserved only for theduration of transmitting the request and initiating a second data transfer. Thedata to be transferred is encapsulated into short packets with a header

containing the originating and destination address. No preset time slots are used.Instead, network capacity is allocated when needed and released when notneeded. This is called statistical multiplexing. In static time division, time slots arereserved for one user for the length of the connection, regardless of whether it isused or not, as with PCM lines and GSM voice and circuit switched data.

GPRS offers a very flexible range of bit rates, from less than 100bits/sec to over100 kbit/s. Applications that need less than one time slot benefit from GPRSsability to share one time slot among several users. Moreover, the high bit ratesthat GPRS provides by usingMultiple time slots give short response times, even if a lot of data is transmitted.

2.1 Benefits for the operator

GPRS has minimal effect on the handling of circuit switched calls, but theinteroperability of existing circuit switched features needs to be taken intoconsideration. Nevertheless, GPRS does offer additional benefits for the

operator:

- resources are better used, thus there is less idle time

- circuit switched traffic is prioritised, but quality is guaranteed by reservingtimeslots only for GPRS traffic

- new services, applications and business for the operator



6/55

- fast connection setup for end-users

- high bitrate in data bursts, up to 100kbit/sec (for end users)

2.1.1 Key GPRS User EnhancementsSpeed.

GPRS has a theoretical maximum data throughput of 171.2 kbps. This isabout 3 times faster than is generally available to consumers over thepublic telephone network and ten times faster that standard GSM CircuitSwitched Data.

Immediacy

GPRS is effectively always on subject to coverage and therefore datacan be sent and received without the need to dial-up the intendedrecipient.

Simultaneity

Using the correct GPRS Mode, it is possible to make voice callssimultaneously whilst transferring GPRS packet data.

New and Improved Applications

The potential increase in data rates available through GPRS enable morebandwidth-intensive applications (such as video streaming) to be migrated

onto the GSM network.These applications were not suited to the mobile environment in the pastdue to the low data rates available through CSD. GPRS also enablesmore effective use to be made of existing services such as SMS andWAP.

Affordability

GPRS Services will have to be priced at a level where they are attractiveto consumers but sufficient to recover infrastructure investment. Inaddition, billing tariffs will have to be sufficiently flexible to meet a widerange of user requirements.



7/55

2.1.2 Billing

FIGURE 1 GPRS BILLING

The GPRS specifications stipulate the minimum charging information thatmust be collected in the Phase 1 service description. These include:

destination and source addresses

usage of radio interface

usage of external Packet Data Networks

usage of the packet data protocol addresses

usage of general GPRS resources

location of the Mobile Station.

volume of packets sent and received

Current billing techniques can only handle charging for circuit-switchedservices (plus SMS) and therefore new billing methods must beintroduced to cope with packet data transfers.



8/55

GPRS call records are generated in the GPRS Support Nodes. TheGGSN and SGSN may not be able to store charging information but thischarging information needs to be processed. Therefore a ChargingGateway is introduced with the specific purpose of collecting andprocessing billing information in the form of Call Detail Records foronward transmission to billing systems.

It may well be the case that the cost of measuring packets is greater thantheir value. The implication is that there will NOT be a per packet chargesince there may be too many packets to warrant counting and chargingfor. For example, a single traffic monitoring application can generate tensof thousands of packets per day. Thus the Charging Gateway function ismore a policing function than a charging function since network operatorsare likely to tariff certain amounts of GPRS traffic at a flat rate and thenneed to monitor whether these allocations are exceeded.

There are a number of parameters against which charges can be raised.It is likely that any implemented billing scheme will be based on a

combination of these parameters:

Volume: This can be measured in number of packets sent/received or theamount of data in Kbytes.Flat Rate: This is similar to the business model used for fixed-line ISPs.Subscribers pay a fixed fee and then have unrestricted access to theservice. Such a service would benefit heavy users rather than those withoccasional access requirements.



9/55

FIGURE 2 CHARGINGFOR GPRS

Time: Subscribers could be billed for the amount of time spent connectedto the GPRS service, regardless of traffic volume. However, this conceptmay be difficult to implement due the always on-line concept of GPRS.

Transaction: For Internet access, ISP functions and services provide theirown accounting data. It is likely that this will be transaction-based (priceper response, price per piece of information for example.Content. Billing by access to value-added services such as specificinternet sites or downloads such as sports results etc.

The chosen billing tariff will be modified by additional user requirementssuch as enhanced security or specific quality of service requirements.

In order to broaden consumer base appeal, it is likely that a number oftariffs will become available to suit different usage models.

3 Base Station Subsystem GPRS Modifications



10/55

The Base Transceiver Station (BTS) will require a software upgrade, buttypically will require no hardware enhancements.

The Base Station Controller (BSC) will also require a software upgrade,as well as the installation of a new piece of hardware called a PacketControl Unit (PCU).

When either voice or packet data traffic is originated at the subscriberterminal, it is transported over the air interface to the BTS, and from theBTS to the BSC in the same way as a standard GSM call. However, atthe output of the BSC the traffic is separated; voice is sent to the mobileswitching center (MSC) as with normal GSM, and packet data is sent to anew network node called the SGSN, via the PCU over a networkinterface.

3.1 THE PACKET CONTROL UNIT (PCU)



11/55

The PCU is responsible for the following GPRS MAC and RLC layerfunctions:

LLC layer PDU segmentation into RLC blocks before downlinktransmission over the air interface

LLC layer PDU reassembly of RLC blocks into LLC PDUs after receipt on

the air interface uplink

Scheduling functions for uplink and downlink data transfersUplink and downlink Backward Error Correction (ARQ) functionsincluding RLC block ack/nack (uplink) and RLC block buffering andretransmission (downlink)

Radio channel management functions including access control, packetscheduling, congestion control, power management etc



12/55

FIGURE 3 PCU CIRCUIT/ PACKET DATA SEPARATION

The PCU provides a physical and logical data interface between the basestation system (BSS) and the SGSN in the GPRS packet data network. Itseparates packet data traffic from circuit switched traffic (voice or data) atthe BSC and directs the packet data traffic into the GPRS network. The

PCU terminates the radio end of the Gb interface.

PCU ConfigurationsAlthough the PCU resides logically at the BTS, it may be positioned in ageographically remote location. Its position determines the interconnectingfunctionality.

The A configuration is most common and is representative of the logicalconfiguration.If the PCU is physically located in the BSC (configuration B), packets aretransferred between the BTS CCU and the PCU with a fixed 320-bytelength every 20mS,and are known as PCU Frames.

The Abis interface is the same for both B and C configurations but requiresin-band signalling (C-bits), unlike configuration A.



13/55

FIGURE 4 PCU CONFIGURATION

The advantage with configuration A is a reduced requirement for channelcapacity between BTS and BSC/SGSN. However, configurations B and Crequired fewer PCUs in the system.

The Channel Coding Unit. The (CCU) performs the following functions:Channel coding functions (CS1-4, see Section 9.1)Forward Error Correction (FEC)Interleaving (see Section 9.1)Radio channel measurement functions

CCU Radio channel measurement functions including received qualitylevel, received signal level and information related to timing advancemeasurements.



14/55

4 GPRS Core Network Modifications

In order to transport packet data across an existing GSM infrastructure,new network nodes have been introduced, known as GPRS SupportNodes (GSNs). These nodes are responsible for the efficient routing of

packet data between a GPRS-enables MS and an external Packet DataNetwork (PDN).

There are two main types of GSN: the Serving GSN (SGSN) and theGateway GSN (GGSN). Also, a number of existing GSN entities need tobe modified to accommodate additional GPRS functionality. Theseinclude the MSC and HLR/VLR databases. Also an interface between theexisting Short Message Service (SMS) node and the new GPRS SMSnode has been defined.The HLR stores the user profiles, current SGSN and PDP address foreach GPRS subscriber in the network.



15/55

4.1 Serving GPRS Support Node

The SGSN can be viewed as a "packet-switching MSC;" it deliverspackets to and received packets from mobile stations (MSs) within its

service area.

SGSNs send queries to home location registers (HLRs) to obtain profiledata of GPRS subscribers.

An SGSN detects new GPRS MSs in its service area, process registrationand authentication and manages ciphering between MS and SGSN.

The SGSN also performs mobility management functions such as mobilesubscriber attach/detach, location management inside its GPRS Location

Area (LA) and logical link management towards the MS.For each MS, the SGSN also collects billing information related to networkusage (as does the GGSN).

The SGSN is connected to a PCU in the BSS via the Frame-Relaynetwork layer, the Home Location Register (HLR) and the GGSN fortransfer of packet data in from and out to external packet networks.



16/55

4.2 Gateway GPRS Support Node

The GGSN is the interface towards external Packet Switched Networks(PSNs), such as the IP-based Internet and X.25 networks. It is acting asan access server to the GPRS network and is responsible for routingincoming traffic to the correct SGSN, i.e. it is responsible for setting up alogical link to the mobile station (MS), through the SGSN.

The GGSN also translates between data formats, signalling protocols andaddress information to enable communication between the GPRS networkand differing external networks.

GGSNs are used as interfaces to external IP networks such as the publicInternet, other mobile service providers' GPRS services, or enterprise

intranets. GGSNs maintain routing information that is necessary to tunnelthe protocol data units (PDUs) to the SGSNs that service particular MSsusing the GPRS Tunnelling Protocol (GTP).

The GGSN also manages PDP contexts between itself and MSs forpacket data transfer in and out of the GPRS core network.



17/55

Other functions include network and subscriber screening and addressmapping (using a Domain Name Server). One (or more) GGSNs may beprovided to support multiple SGSNs

4.3 Additional Elements of the GPRS

Network

FIGURE 6 ADDITIONAL ELEMENTSOFTHE GPRS NETWORK

The Border Gateway (BG) is a network element that provides adirect connection to other operators' GPRS networks, thus allowingoperators to avoid using the public Internet to transfer data to otherGPRS networks. This means that, when roaming to another GPRSnetwork, subscribers can have a secure GPRS connection to theirhome network. This connection is provided via a GPRS "tunnel"over an inter-operator backbone network. The BG is alsosometimes referred to as the BGSN or BGGSN.

The Charging Gateway (CG) is a stand-alone network element forcontrolling the Charging Detail Records (CDRs) processing routinesin the GPRS network. It collects CDRs from GGSNs) and SGSNsand forwards them to a billing system (BS) after consolidating theCDRs and converting them to a suitable format. Quality checkingfunctions are also provided.



18/55

The Lawful Interception Gateway (LIG) fulfils an essentialfunction within the GPRS network by providing the ability tointercept GPRS mobile data calls for the purpose of lawenforcement, as required by national authorities. In many countries,

local regulatory authorities require a means of lawful datainterception before GPRS networks can be launched commercially.Data call interception in the GPRS network is a new methodcompletely different from GSM call interception. In GSM,interception is mainly voice-based audio recording, while in GPRSdata interception occurs between the GPRS terminal and theaccess point (the GGSN).

The Domain Name System (DNS) provides the Internet namingstructure for the GPRS network, by translating Web host addressesinto numerical IP addresses. Based on the URL requested by the

user, the DNS supplies the actual IP address for the correct accesspoint (in the GGSN) to the Internet.

Dynamic Host Control Protocol (DHCP) Server fulfils the functionof managing temporary IP address assignment within the GPRScore network.



19/55

5 GPRS-Associated Interfaces

FIGURE 7GPRS ASSOCIATED INTERFACES

The GPRS specification defined a number of compulsory interfaces,

optional interfaces and reference points. Some interfaces aredefined for transmission and some for signalling. An interfacediffers from a reference point in that an interface is defined whereGPRS-specific information is passed and processed in some way.

Three compulsory GPRS transmission interfaces are defined:

Um. MS to BSS interface.

Gb. BSS to Core Network (SGSN) interface.

Gn. GSN to GSN interface (primarily SGSN-GGSN).

Two reference points are defined:Gi. Core Network to External Public Packet-Switched Network(PPSN) reference pointThe R reference point lies between the TE and MS.

There are also a number of optional interface implementations ascan be seen from the above diagram:



20/55

Transmission Interfaces:Gc This optional SS7 interface is generally used by the GGSN tointerrogate the HLR to update its location database with the currentuser location.Gd This optional interface is used to exchange SMS messagesusing the GPRS network rather than the conventional GSM method.Gp This optional interface is used to interface to other PLMNGPRS networks

Signalling InterfacesGf This interface connects the SGSN to the Equipment IdentityRegister (EIR) for accessing authentication information.Gr This interface is used to pass subscriber profile informationbetween the SGSN and the HLR. For example, the SGSN informsthe HLR about the current location of an MS. When the MSregisters with a new SGSN, the HLR will send the MSs profile tothe new SGSN via the Gr interface.Gs This optional interface is generally used to exchangemessages between the SGSN and the MSC/VLR when performingcircuit-switched paging requests via the SGSN.



21/55

5.1 The Um Air Interface

FIGURE 8 UM AIR INTERFACE



22/55

Enhancements to the GSM Air interface to support GPRS include:

GPRS has defined a number of new logical channels formanaging the flow of packet data across the Um interface.

A new 52-frame multifame structure has been defined,based on radio blocks of 4 timeslots

Ability to allow up to 8 GPRS users to share a singletimeslot.

Ability to allocate multiple timeslots to a single GPRS user.

Four new channel coding schemes (CS-1 to CS-4) havebeen introduced that allow greater data throughput when airinterface radio path quality permits.



23/55

5.1.1 The Um Radio Link Control (RLC) Layer

FIGURE 9 THE UM INTERFACE RLC LAYER



24/55

The RLC protocol provides a reliable radio link for transporting datapackets passed to it from the Logical Link Control (LLC) layer. It is alsoresponsible for:

Transferring Logical Link Layer (LLC) Packet Data Units (PDUs)

between the LLC layer and the MAC layer. Segmentation and reassembly of LLC PDUs into RLC data blocks.

Backward Error Correction (BEC) procedures. The RLC adds a

Block Check Sequence (BCS) to each RLC Radio Block. Thisenables errors in each block to be detected. Correction ofdetected errors is achieved through selective block retransmission.This process is generally known as Automatic ReQuest forretransmission (ARQ).



25/55

5.1.2 Radio Link Control Block Generation

FIGURE 10 RLC BLOCK GENERATION

Packets generated at the LLC layer are passed to the RLC layer wherethey are segmented into equal size blocks and encapsulated within theRLC protocol header and BCS before being passed to the MAC layer.The payload size of each block is dependant upon the coding schemebeing used.



26/55

5.1.3 THE Um MAC LAYER

FIGURE 11 THE UM INTERFACE MAC LAYER



27/55

The Medium Access Control (MAC) layer protocol handles packetresource allocation and multiplexing. It is responsible for:

The efficient multiplexing of data and control signalling packets on

both the uplink and downlink. On the downlink, multiplexing is

controlled by a scheduling mechanism. On the uplink, multiplexingis controlled by resource (channel) allocation to individual users.

Mobile originated channel access, contention resolution between

channel access attempts including collision detection andrecovery.

Mobile Terminate channel access, scheduling of access attempts

including queuing of packet accesses.

Priority handling when QoS levels have been applied.



28/55

FIGURE 12 MAC FRAMES GENERATION

At the MAC layer, additional header information is added for controllingthe Air Interface traffic management. This information differs dependingon the direction of traffic flow.



29/55

5.1.4 THE Um LINK LAYER

FIGURE 13 THE UM INTERFACE LINK LAYER



30/55

The Link Layer divided into two parts:

Radio Frequency (RF) Layer. The RF layer is responsible for:

Carrier frequency characteristics

Modulation techniques employed

Tx / Rx characteristics and performance requirements.

GSM Radio Channel Structures

Physical Link Layer: The physical layer provides communicationbetween the MSs and the network. It is responsible for:

Forward Error Correction (FEC) Coding. FEC allows the detection

and correction of errors using a Frame Check Sequence (FCS). Insome circumstances, errored code words cannot be corrected, inwhich case, the error is reported only.

Burst Interleaving. Interleaving breaks a radio block into 4 parts

and burst transmits them over 4 consecutive TDMA frames. Thisreduces the vulnerability of data to burst errors on the radio path.

FIGURE 14 BURST FORMATTINGINTHE UM INTERFACE



31/55

FIGURE 15 BURST INTERLEAVING IN THE UM INTERFACE

The RLC Block, together with the MAC header (i.e. the MAC frame) formsthe Radio Block. This radio block is divided into four 114-bit data bursts.

In order to integrate Control channels onto the air interface, GPRS hasintroduced a new 52-frame multiframe structure. Each multiframe

contains 12 radio blocks and each radio block is divided into four 114-bitframes or data bursts.

If all 4 bursts of a radio block from the same user were to be transmittedconsecutively in the same TDMA frame, a burst of noise could disrupt thewhole transmission. Therefore to make the data transfer more robust,each burst of a 114-bit data block is transmitted in a timeslot ofconsecutive TDMA frames as illustrated above.

Each data burst is then divided into two 57-bit blocks and inserted into a158.25-bit radio burst (1 timeslot) for transmission across the radiointerface.



32/55

6 GPRS PHYSICAL AND LOGICAL CHANNELS

6.1 GPRS Logical Channels

FIGURE 16 GPRS LOGICAL CHANNELS

As with GSM, GPRS has defined a number of new logical channels thatperform a multiplicity of functions including signalling, system informationbroadcast, synchronisation, channel assignment, paging and payloadtransport. These logical channels can be divided into two categories;traffic and signalling/control. Note that, unlike conventional GSM, aGPRS handsets can be configured to access multiple traffic channels(PDTCHs) simultaneously.

All packet control signalling takes place over a physical channel dedicatedto packet data, called the Packet Data Channel (PDCH).



33/55

FIGURE 17 PBCCH/PCCCH LOGICAL CHANNELS

FIGURE 18 PDTCH/PDCCH LOGICAL CHANNELS



34/55

GPRS Logical Channels:

Packet Common Control Channels (PCCCH)- Packet Random Access Channel (PRACH) - UL- Packet Paging Channel (PPCH) - DL- Packet Access Grant Channel (PAGCH) - DL- Packet Notification Channel (PNCH) - DL

Packet Broadcast Control Channel (PBCCH) - DL Packet Data Traffic Channel (PDTCH) - UL/DL Packet Dedicated Control Channels (PDCCH)

- Packet Associated Control Channel (PACCH) - UL/DL- Packet Timing Advance Control Channel (PTCCH) - UL/DL

Allocation of GPRS Logical Channels

Dynamic allocation according to capacity on demand Not permanent allocation of PDCHs At least one PDCH channel acts as a master carries:

- Control signalling (PCCCH or CCCH)- Dedicated signalling (PACCH)- User data (PDTCH)

Other PDCHs used for:- User data- Dedicated signalling

Master-Slave Concept

Allocation of GPRS Logical Channels

The number of GPRS channels is monitored by the Territory UpgradeTimer (TUT)

- High value results to lower GPRS available capacity and increase ofspeech blocking- Small value results to higher GPRS available capacity but increase ofDL signalling

Depends on the number of carriers per cell- 1 carrier per cell (PBCCH and PCCCH information carried outthrough BCCH and CCCH)- More than one carrier per cell (PBCCH, PCCCH Channels Available)



35/55

Allocation of GPRS Physical Channels

1 2 3 4 5 6 7 88 BCCH

7

6 Speech Call

5

4 TBF 1

3 TBF 2

2 TBF 3

1

Temporary Block Flow (TBF) is released if there are no more packets to be sente.g. MS is not using its uplink TSL allocations

Load of TBFs per TSL is balanced across TSLs

Allocation of GPRS Physical Channels

TRX1

TRX2


BCCH

Speech Calls

Additional GPRS Capacity (CS Default, GPRS possible)

Default GPRS Capacity (GPRS Default, CS when needed)

Dedicated GPRS Capacity (GPRS only)


36/55

7 Timeslot Resource Allocation



37/55

FIGURE 19 USING SPARE GSM CAPACITY

GPRS can use traffic capacity on the GSM network away from the busyhour for non time-critical data transfers. Even during the busy hour, thereis spare capacity that GPRS can make use of.

The first graph above shows the demand for circuit-switched servicesover a 24-hour period and where GPRS packet traffic could be insertedwith no increase in radio resource requirements.

The second graph above shows the allocation of timeslots for a combinedcircuit/packet switched service over a single carrier. One Timeslot (TS) ispermanently assigned for packet traffic and six can be dynamicallyallocated for either voice of packet data with voice having the priority. (the

8th TS being for signalling). Therefore, when packet data is to be sent, itwill be allocated a single timeslot initially (shown in red). If packet datademand exceeds the capacity of a single TS, resources are allocated onone or more of the other 6 TSs (shown in blue), as voice traffic loadingallows.



38/55

7.1 Operator Time Slot Resource Allocation



39/55

7.2 LOAD SUPERVISION

The GPRS radio interface consists of asymmetric and independent uplinkand downlink channels. The downlink carries transmissions from thenetwork to multiple MSs and does not require contention arbitration. Theuplink is shared among multiple MSs and requires contention controlprocedures.

7.3 Uplink Timeslot Allocation

Four modes of operation are supported for the purposes of uplink timeslotallocation:

Dynamic

Extended Dynamic

Fixed

Exclusive

Dynamic and fixed allocation methods are mandatory for current GPRSnetworks.

The exclusive allocation mode is optional.

7.3 Uplink Time Slot Allocation



40/55

Dynamic Resource Allocation. Dynamic resource allocation refers tothe allocating of GPRS packet data traffic to existing timeslots on astatistical basis where circuit-switched traffic allows (assuming priority isgiven to circuit-switched traffic). Allocation is based to Uplink State Flag(USF) assignment (see below)

Fixed Resource Allocation. An alternative method of sharing thetimeslots is supported which does not use the USF. In this case, the MSis sent a fixed list of timeslots and radio blocks in each timeslot which areallocated for its use.

Exclusive Resource Allocation. Exclusive allocation is used to reservethe uplink part of the PDCH for only one MS during the life of the TBF.Therefore, with exclusive allocation, all the uplink blocks of the uplink partof the PDCH are available to the MS for transmission.

THE UPLINK STATE FLAG



41/55

The Uplink State Flag (USF) is used on PDCH to allow multiplexing ofRadio blocks from a number of MSs onto a single timeslot. The USF isused in dynamic and extended dynamic (EGPRS) access modes. TheUSF is transmitted only in the downlink direction but controls allocation ofuplink resources. It comprises 3 bits at the beginning of each downlinkRadio Block and uniquely identifies up to 8 different traffic streams that

can be multiplexed onto a single uplink PDTCH.

To control the dynamic multiplexing of radio blocks from different MSs, aUSF signal is transmitted on the downlink to tell each MS which radioblock it may use. In the initial assignment message on PAGCH, the MSreceives a list of the PDCHs each with a corresponding USF value. TheMS monitors the USF values in downlink transmission on the assignedPDCHs. The MS may only transmit in the radio blocks that currently havethe same USF value that was sent in the assignment message.

FIGURE 20 UPLINK STATE FLAG

In the Figure shown, three USF values (1, 2 or 3) have been assigned tothe uplink radio blocks.

User A was sent USF =1 (001) in its PAGCH assignment message

and so can only use blocks B0 to B4.

User B was assigned USF = 2 (010) and can use blocks B5 to B8.



42/55

7.4 DOWNLINK TIMESLOT ALLOCATION

The transmission of a packet to an MS in the Ready state is initiated bythe network using a packet downlink assignment message. Thismessage includes the list of PDCH(s) that will be used for downlink

transfer. The MS may be requested to respond with a Packet ControlAcknowledgment.The network sends the RLC/MAC blocks belonging to one TemporaryBlock Flow on the assigned downlink channels. Multiplexing theRLC/MAC blocks destined for different MSs on the same PDCH downlinkis enabled with an identifier, e.g. TFI, included in each RLC/MAC block.The interruption of data transmission to one MS is possible.

7.5 Temporary Block Flows (TBFs)

A Temporary Block Flow (TBF) is a physical connection used by the twoRadio Resource (RR) entities (e.g. the GPRS MS and the BSS) to supportthe unidirectional transfer of LLC PDUs on packet data physical channels.The TBF is allocated radio resources on one or more Packet DataChannels (PDCHs) and comprise a number of RLC/MAC blocks carryingone or more LLC PDUs.

A TBF is temporary and is maintained only for the duration of the datatransfer.



43/55

TEMPORARY FLOW IDENTIFIER

Each TBF is assigned a Temporary Flow Identity (TFI) by the network.The assigned TFI is unique among concurrent TBFs in each direction andis used instead of the MS identity in the RLC/MAC layer. The same TFI

value may be used concurrently for TBFs in opposite directions. The TFIis assigned in a resource assignment message that precedes the transferof LLC frames belonging to one TBF to/from the MS. The same TFI isincluded in every RLC header belonging to a particular TBF as well as inthe control messages associated to the LLC frame transfer (e.g.acknowledgements) in order to address the peer RLC entities.

Establishing a TBFEither an MS or the network can establish a TBF. The PCCCH is used, ifavailable, to request the establishment of the TBF. If no PCCCH isavailable in the cell, the CCCH can be used.

MS-Initiated TBF Establishment. The purpose of establishing a TBF is to

facilitate the transfer of LLC PDUs from the MS to the network. Accesscan be granted for one or two phase access:

One-Phase. One-phase access is used automatically when in

RLC acknowledged mode or the total data to be transmitted will fitinto 8 or less RLC/MAC blocks. If the total data to be transmittedwill not fit into 8 or less RLC/MAC blocks, 1 or 2-phase access canbe used.



44/55

Two-Phase (optional). Two-phase access is used when in RLC

unacknowledged mode, or can optionally be used if the total datato be transmitted will not fit into 8 or less RLC/MAC blocks.

The main difference between 1 and 2-phase access is that 1-phase is

simpler to implement but only has a limited ability to describe the packetresources required. 2-Phase access is used when additional resourcesare required as it provides a fuller description of the requirements.

7.5.1 MS-Initiated TBF Establishment

FIGURE 21 MS- INITIATED TBF ESTABLISHMENT

The outline procedure for establishing an MS-initiated TBF is as follows:An upper MS layer requests the transfer a LLC PDU.The MS initiates a packet access procedure by sending a PacketChannel Request message to the network.The network allocates radio resources in the form of one or more PDCHsby replying with a Packet Uplink Assignment message, including theUSF if dynamic allocation is used.Packet transfer is established if 1-Phase access is used.



45/55

If 2-phase access is used the MS sends a further Packet ResourceRequest message, providing a complete description of the resourcesrequired.

The network responds with a second Packet Uplink Assignmentmessage, allocating further resources if available.

7.5.2 Network-Initiated TBF Establishment

The purpose of establishing a TBF is to facilitate the transfer of LLC PDUsfrom the Network to the MS.

FIGURE 22 NETWORK-INITIATED TBF ESTABLISHMENT

The outline procedure for establishing a network-initiated TBF is asfollows:

The network-initiated TBF is started when the network has LLC PDUs tosend to the MS.



46/55

If the destination MS is in the STANDBY MM state, a Packet PagingRequest message is issued to the RA of the MS.

The MS responds to the received paging message by sending a PacketChannel Request message to the network.

The network then allocates resources using a Packet DownlinkAssignment message.

Finally, the MS issues a Packet Paging Response message implicitlywithin the first LLC PDU received by the network. The MS moves into

READY MM mode and packet transfer commences.

8 Radio Resource Operating Modes

Packet Idle Mode

In packet idle mode no packet data flow (i.e. no TBF) exists. Upper layerscan require the transfer of a LLC PDU which, implicitly, may trigger packetflow and transition to packet transfer mode.In packet idle mode, the MS listens to the PBCCH and to the paging sub-channel for the paging group the MS belongs to. If PCCCH is not presentin the cell, the mobile station listens to the BCCH and to the relevantpaging sub-channels.



47/55

Packet Transfer Mode

In packet transfer mode, the mobile station is allocated radio resourceproviding a TBF on one or more physical channels. Continuous transferof one or more LLC PDUs is possible. Concurrent TBFs may beestablished in opposite directions. Transfer of LLC PDUs in RLC

acknowledged or RLC unacknowledged mode is provided.

Dual Transfer Mode

In dual transfer mode, both packet and circuit-switched connections areestablished simultaneously. The MS has an ongoing circuit-switched

connection and is allocated radio resource providing a TBF on one ormore physical channels for packet data transfer. Concurrent TBFs maybe established in opposite directions. Transfer of LLC PDUs in RLCacknowledged or RLC unacknowledged mode is provided.While in dual transfer mode the MS performs all the tasks of dedicatedmode. In addition, upper layers can require:

the release of all the packet resources, which triggers the

transition to dedicated mode.

the release of the RR resources, which triggers the transition to

idle mode and packet idle mode.

Cell Handover. When handed over to a new cell, the MS leaves the

packet/dual transfer mode, enters the dedicated mode where it switchesto the new cell, may read the system information messages sent on theSACCH and may then enter dual/packet transfer mode in the new cell.

Establishing Circuit Switched Connections. In Packet Idle or transfermode, an A class GPRS MS may simultaneously establish a circuit-switched connection. B or C class GPRS MSs must leaves both packetidle mode and packet transfer modes before establishing a circuit-switched connection.



48/55

9 Network Planning Considerations

9.1 GPRS Coding Schemes

FIGURE 23 GPRS CODING SCHEMES

The use of radio as a data transmission medium is prone to errors from anumber of sources. Coding schemes are used to protect data from errorswhen crossing this radio path interface.Unfortunately, the more robust the protection, the greater the dataoverheads required. Whilst the raw data throughput remains the same,the actual information throughput reduces with a corresponding increase



49/55

in protection overheads. Therefore, to maximise the use of variablequality radio paths, a number of coding schemes are available. Eachcoding scheme has varying levels of protection to maximise theinformation throughput where good radio paths allow.

GPRS has introduced 4 new coding schemes for this purpose, CS-1 to

CS-4. For compatibility purposes, all coding schemes are mandatory forMobile Stations but only CS-1 is compulsory for BSSs.

9.2 C/I Coverage

FIGURE 24 C/I COVERAGE

Most existing GSM900 networks are capacity restricted now

- Characterised by minimal C/I & C/A operation

- Difficult to add new services due to restricted capacity

GSM1800 networks have most to offer

- Smaller cell sizes- Typically higher capacity availability


GSM onlycell

EnablingGPRS

GSMCoverage

e

Cs1111

Cs2222

Cs3111111

CS44


50/55

9.

3 Throughput Vs C/I

FIGURE 25 MAXIMUM THROUGHPUT VERSUS C/I IN FREQUENCY HOPPING


C/I levels impact throughput in packet networksDifferent services require different throughput levels, i.e. different C/I levels

Depends on several parameters such as:- Environment- Speed- Frequency hopping


51/55

FIGURE 26 MAXIMUM THROUGHPUT VERSUS C/I WITH NO FREQUENCY HOPPING



52/55

10 Limitations of GPRS

Slower Data Rates Than Anticipated

In order to achieve the theoretical maximum data rates would require a

single user to be allocated all 8 timeslots on specific air interface carrierwithout error protection. This is unlikely to happen for several reasons:

It is unlikely that an operator will allocate all 8 timeslots per carrier todata.Most planned GPRS handsets will only support up to 3 timeslots.

As a result, the theoretical data rates are only likely to be achievedthrough the implementation of EDGE or UMTS.

Cell Capacity

Although GPRS can utilise redundant capacity on the air interface, it mayrequire the allocation of timeslots possibly at the expense of voice users.

A conflict may arise between the service level required by voice users andthose of GPRS data users.



53/55

Sub-Optimal Modulation Technique

Current Air interface modulation uses Gaussian Minimum Shift Keying orGMSK (see Modulation Techniques). More efficient modulationtechniques are now available (such as 8PSK used in EDGE technology)that can significantly increase the available data rates (by a factor of 3)with no increase in the air interface resource requirements.

Transit Delays

GPRS availability is limited by network resources and the application ofhigh QoS levels are resource-intensive and may not be achievable in theearly stages of GPRS. It cannot guarantee the route taken by eachpacket and the retransmission delays incurred with errored packets.Therefore, transit delays cannot be guaranteed, making it less suited toreal-time applications that circuit-switched data technologies such asHSCSD.

No Store and Forward.

Unlike SMS, where messages are stored until the recipient is available toreceive them, GPRS makes no such provision. Packets that cannot be

delivered are generally discarded.



54/55



55/55

gprs traffic modelling.doc

Documents