overview of the gsm system and protocol architecture

Overview Of The GSM Systemand Protocol ArchitectureWe can use GSM as a basic framework to define and developthe standards for handling the mobility-specific functions ofnext-generation PCNs.

Moe Rahnema

IEEE Communications Magazine April 1993

GSM and PCNs

0163-6804/93/$03.00 1993 IEEE

Global system for mobile telecommunication (GSM)comprises the CEPT-defined standardization of theservices, functional/subsystem interfaces, andprotocol architecture, based on the use of worldwidestandards produced by CCITT and CCIR, for a pan-European digital land mobile system primarily intend-ed to serve users in motor vehicles. The digital mobileradio networks, for which GSM represents the Euro-pean standards, provide powerful message signal-ing capabilities that facilitate and enhanceroaming, compared to the first generation analoguesystems, through automatic network location detec-tion and registration.

GSM provides terminal mobility, with person-al mobility provided through the insertion of asubscriber identity module (SIM) into the GSM net-work (mobile station). The SIM carries the personalnumber assigned to the mobile user. The GSM-basedcellular mobile networks are currently in widespreaduse in Europe. At the present time, the next gen-eration of personal communication services(PCS) beyond GSM is also being considered.These third generation systems, known as univer-sal personal communication networks (PCN) will beusing lower power handsets to provide personalmobility to pedestrians, as well. The PCS low-powerhandsets are expected to eliminate the need tohave different handsets for wide-area (cellular)and local (cordless) applications. The universal PCSwill also provide a higher quality of personal-servicemobility across the boundaries of many different net-works (mobile and fixed, wide- and local-area).

Many network capabilities, however, such asmobility management, user security protection, andresource allocation, addressed in GSM, are also someof the critical requirements and issues in UPC net-w o r k s of the future. GSM is expected to play amajor role in the specification of the standardsfor UPC. In the United Kingdom, PCN is alreadybeing designed and deployed with close adher-ence to the GSM standards other than the differ-ent operating frequencies (GSM operates at 900MHz and the United Kingdom PCN operates at 1800MHz). Generally, GSM may be viewed as a frame-work for studying the functions and issues that

are specific to cellular type personal communicationnetworks, whatever the means of implementationmight be.

In applying and extending GSM to the next gen-eration personal communication networks, how-ever, one should be careful in differentiating someof the implementation specifics unique to the GSMnetwork architecture and application from the func-tions and issues that would be more or less gener-ally applicable and relevant to cellular networking.It is with this point in mind that the reader shouldview GSM as a framework or platform on whichto build his or her vision of how GSM may be usedas a guide to design and build the next generationnetworks. In that regard, a good understanding ofthe GSM standards and network functions isessential for the professional working on the nextgeneration personal communication networks. Thisarticle is intended to assist with this objective.

The Cellular Concept

C ellular mobile communication is based on theconcept of frequency reuse. That is, the limit-ed spectrum allocated to the service is partitionedinto, for example, N non-overlapping channelsets, which are then assigned in a regular repeat-ed pattern to a hexagonal cell grid. The hexagonis just a convenient idealization that approximatesthe shape of a circle (the constant signal levelcontour from an omnidirectional antenna placedat the center) but forms a grid with no gaps oroverlaps. The choice of N is dependent on many trade-offs involving the local propagation environment,traffic distribution, and costs. The propagation envi-ronment determines the interference received fromneighboring co-channel cells which in turn gov-erns the reuse distance, that is, the distanceallowed between co-channel cells (cells using thesame set of frequency channels).

The cell size determination is usually based onthe local traffic distribution and demand. The morethe concentration of traffic demand in the area,the smaller the cell has to be sized in order to availthe frequency set to a smaller number of roaming

MOE RAHNEMA is a principal communicationengineer at Motorola SatelliteCommunications.


subscribers and thus limit the call blocking proba-bility within the cell. On the other hand, thesmaller the cell is sized, the more equipment willbe needed in the system as each cell requires the nec-essary transceiver and switching equipment,known as the base s tation subsystem (BSS),through which the mobile users access the net-work over radio links. The degree to which theallocated frequency spectrum is reused over the cel-lular service area, however, determines the spectrumefficiency in cellular systems. That means thesmaller the cell size, and the smaller the numberof cells in the reuse geometry, the higher will bethe spectrum usage efficiency. Since digital mod-ulation systems can operate with a smaller signalto noise (i.e., signal to interference) ratio for thesame service quality, they, in one respect, would allowsmaller reuse distance and thus provide higher spec-trum efficiency. This is one advantage the digital cel-lular provides over the older analogue cellular radiocommunication systems. The interested reader mayrefer to [1,2] for the details on spectrum efficien-cy analysis in cellular network.

It is worth mentioning that the digital systems havecommonly used sectored cells with 120-degree orsmaller directional antennas to further lower theeffective reuse distance. This allows a smallernumber of cells in the reuse pattern and makes alarger fraction of the total frequency spectrum avail-able within each cell. Currently, research is beingdone on implementing other enhancements suchas the use of dynamic channel assignment strate-gies for raising the spectrum efficiency in certaincases, such as high uneven traffic distributionover cells.

The Network Infrastructure

T he cellular concept of networking is based on thesuperposition of a distributed star type net-work architecture on the existing fixed landline tele-phony communication infrastructure. The basicnetwork architecture is illustrated in Fig. 1. The tele-phony network is used to provide not only thecommunication links between a mobile user anda fixed landline user, but also to provide the con-nectivity between the mobile users roaming in remote-ly located cells or in the domain of mobile networksoperated by different service providers. TheBSSs, provide management of the radio resources,and the switching between the radio channels andthe TDM slots on their connections with themobile switching center (MSC). MSCs link groupsof neighboring BSSs through point-to-point land-line or microwave-based E1 trunks. The MSCacts as the nerve center of the system. It controlscall signaling and processing, and coordinates thehandover of the mobile connection from onebase stat ion to another as the mobile roamsaround. Each MSC is in turn connected to thelocal public switched telephony network (PSTN,or ISDN) to provide the connectivity between themobile and the fixed telephony users, as well as thenecessary global connectivity among the MSCs ofthe cellular mobile network. This is intended to makeit possible for any mobile user to communicatewith any other mobile or fixed telephony user inthe world. Thus, the global connectivity providedby the existing landline telephony infrastructureis used to link up the cellular mobile subscribers

throughout the world.Direct links between certain l o c a l M S C s

may also be provided to allow the communicationbetween two mobile users to bypass the telepho-ny network when there is considerable trafficflow between the mobile users roaming in theareas under the coverage of those MSCs. Thus,the communication path between any two mobileusers roaming under the coverage of two localMSCs may or may not switch through the public tele-phony network. It depends on the connectivityprovided between the two MSCs. The MSC may alsoconnect to public data networks (PDN), such asthe packet-switched networks, to provide the mobileswith access to data services.

Network Databases andStandardization

G SM defines a number of network databases thatare used in performing the functions of mobil-ity management and call control in a public landmobile network (PLMN). These elements includethe location registers consisting of the home loca-tion register (HLR), and the visiting location reg-i s t e r (VLR), the equipment identity register (EIR),and the authentication center (AC). The HLR main-tains and updates the mobile subscribers loca-tion and his or her service profile information.The VLR maintains the same information local-ly, where the subscriber is roaming. The VLR isdefined as a stand-alone function (see following para-graph), but is usually viewed by vendors as part ofthe MSC. These registers are called service con-trol points (SCP) in the terminology used in intel-ligent networking (IN). The EIR is used to listthe subscribers equipment identities, which are usedfor identification of unauthorized subscriber equip-ment, and hence denial of service by the network.The AC provides the keys and algorithm formaintaining the security of subscriber identities, andfor encrypting information passed over the air inter-face. The MSC is equipped with a service switch-ing point (SSP) module which is used to querythe databases such as a location register to identi-fy where a mobile subscriber is located and whathis or her service profile is, for the routing, andprocessing of calls to (or by) the subscriber.

The GSM specifications have defined logicallyseparate functions and standard interfaces for eachof the databases, to allow each function to be imple-mented on a physically separate network compo-nent. The interfaces are specified via the mobileapplication part (MAP) that uses the transactioncapability applications part (TCAP) of (SS7). Theseare all elements of an IN. GSM is considered an

n Cellular network infrastructure.


IN application and GSM providers are consider-ing the GSM implementation as experience inintelligent networking.

Numbering Plan

T he numbering consists of at least one internationalISDN number allocated to either the mobile sub-scriber, if the mobile is card operated, or to the mobilestation, otherwise. The mobile station ISDN (MSIS-DN) conforms to the CCITT E.164 recommenda-tion, and should, in each country, comply to thatcountrys ISDN numbering plan. The MSISDN num-ber basically consists of a country code (CC), anational destination code (NDC), which speci-fies a PLMN within that country, and a subscribernumber (SN). This structure is shown in Fig. 2.

The MSISDN number is used for dialing by acalling subscriber from the PSTN/ISDN, and is usedto route the call to the gateway MSC of the GSM net-work. The GSM MSC then uses the MSISDN to inter-rogate the appropriate HLR for the re-routinginformation required to extend the call to the mobilesvisiting MSC.

The rerouting information is specified by themobile station roaming number (MSRN) which isobtained from the HLR and is used to progress thecall to the called mobile. The MSRN is a tempo-rary number, allocated by the VLR (associated withthe mobiles visiting MSC) and sent to the mobilesHLR either on location update (discussed in alater section) or on a per call basis. The MSRNhas the same structure as the MSISDN numbersin the visiting location area where it is allocated.

For provision of mobile packet data services, amobile international data number conforming toCCITT recommendation X.121 may be specified.GSM recommendation 03.70 discusses the require-ments for the numbering interworking functionsrequired in this case.

Addressing and Call RoutingThe MSISDN number is used for the routing ofcalls within the PSTN/ISDN networks. The detailsof call routing requirements are discussed inGSM recommendation 03.04. The followingparagraphs provide a summary discussion of pos-sible scenarios involved in call routing.

National Calls from the Fixed NetworkA local or transit exchange, when receiving a calldestined for a mobile, recognizes the NDC, androutes the call to a gateway MSC. The gatewayMSC performs the HLR query for the MSRN,which it then uses to reroute the call.

International Calls from the FixedNetworkWhen a local or transit exchange receives an inter-national call and recognizes the international pre-fix, it routes the call to the nearest ISC. The ISCrecognizes that the NDC indicates a PLMN. If it cansupport HLR query (i.e., if it has TCAP signalingconnectivity to the HLR) it queries the HLR andreceives the called subscribers roaming number androutes the call to the visiting MSC. If not, it routesthe call to the ISC of the home PLMN of thecalled subscriber.

National Calls from Within the PLMNWhen a local exchange (MSC) receives a call destinedfor a mobile, it queries the mobiles HLR for the roam-ing number of the mobile. On receipt of the MSRN,it routes the call to the called mobiles visiting MSC.

Addressing Other Components of aPLMNOther components of a PLMN, which may beaddressed for the routing of various signalingmessages, are the MSCs, and the location regis-ters. If these elements are addressed from withinthe same PLMN, the SS7 point codes (PC) can beused. Otherwise, for interPLMN routing, globaltitles (GT) derived, for instance, from the mobilecountry code (MCC) and the national destinationcodes (NDC) are used.

Radio Channel Structure in GSM

I n GSM, the radio channels are based on a TDMAstructure that is implemented on multiple frequencysubbands (TDMA/FDMA). Each base station isequipped with a certain number of these preassignedfrequency/time channels.

CEPT has made available two frequency bandsto be used by the GSM system. These are: 890-915MHz for the direction mobile to base station, and935-960 MHz for the direction base station tomobile terminal. These bands are divided into124 pairs of carriers spaced by 200 kHz, starting withthe pair 890.2 MHz. Each cell site has a fixedassignment of a certain number of carriers, rang-ing from only one to usually not more than 15c h a n n e l s.The cell ranges in size from 1 to several km.

The assigned spectrum of 200 kHz per chan-nel is segmented in time by using a fixed alloca-tion, time-division multiple access (TDMA) scheme.The time axis is divided into eight time slots oflength 0.577 ms. The slots numbered from timeslot 0 to 7 form a frame with length 4.615 ms. T h erecurrence of one particular time slot in eachframe makes up one physical channel.

The TDMA scheme uses a gross bit rate of about270 kb/s (with a Gaussian minimum shift keyingmodulation, GMSK) and requires sophist i c a t e dadaptive receiver techniques to cope with the t r a n s-mission problems caused by multipath fading.T h eTDMA factor of 8 in combination with a carrierspacing of 200 kHz would correspond to the earli-er analog system using single-channel per-carrier w i t ha 25 kHz carrier spacing. The GSM digital syst e mallowed operation at lower carrier to interference(C/I) ratio by using the gains provided by digital voicecompression along with channel coding (powerfulerror correction). The reduced C/I ratio in turnallowed the use of shorter channel reuse dis-tances to achieve spectrum efficiencies competi-tive to that achieved by the analog systems.

The TDMA structure is applied in both the for-w a r d(base station to mobile) and the reverse ( m o b i l eto base station) directions. The numbering, however,is staggered by three time slots, to prevent the mobilestation from transmitting and receiving at the

n The structure for the GSM MSIS-

In GSM, the radiochannels arebased on aTDMA structure that is i m p l e m e n t e don multiplef r e q u e n c ys u b b a n d s( T D M A /F D M A ) .


same time. These time slots are used to carryuser, and signaling or control information inb u r s t s . The bursts are slightly shorter than theslots, namely .546 ms, to allow for burst timing align-ment errors, delay dispersion on the propagationpath, and for smooth switch on/off of the transmitter.

GSM defines a variety of traffic and signal-ing/control channels of different bit rates. Thesechannels are assigned to logical channels derivedfrom multiframe structuring of the basic eightslotted TDMA frames just discussed. For thispurpose, two multiframe structures have beendefined: one consisting of 26 time frames (result-ing in a recurrence interval of 120 ms), and one com-prising 51 time frames (or 236 ms).

The 26 multiframe is used to define trafficchannels (TCH), and their slow and fast associat-ed control channels (SACCH and FACCH) thatcarry link control information between the mobileand the base stations. The TCH have been definedto provide six different forms of services, that is, full-rate speech or data channels supporting effective bitrates of 13 kb/s (for speech), 2.4, 4.8, and 9.6 kb/s;and the half-rate channels with effective bit-ratesof 6.5 (for speech) and kb/s, 2.4 kb/s, and 4.8 kb/sfor data (note that the gross bit rates on thesechannels are higher due to required channel coding,22.8 kb/s for full -rate speech). The full-rateTCHs are implemented on 24 frames of the mul-tiframe, with each TCH occupying one time slot fromeach frame. The SACCH is implemented onframe 12 (numbered from 0), providing eight SACCHchannels, one dedicated to each of the eight TCHchannels. Frame 25 in the multiframe is currentlyidle and reserved to implement the additionaleight SACCH required when half-rate speech chan-nels become a reality. The FACCH is obtainedon demand by stealing from the TCH, and is usedby either end for signaling the transfer character-istics of the physical path, or other purposes suchas connection handover control messages. The steal-ing of a TCH slot for FACCH signaling is indi-cated through a flag within the TCH slot.

The 51-frame multiframe has a more complexstructure and we will refer the reader to GSMRecommendation 05.0 for the specific positionsof the various logical channels in the multiframe.The 51-frame structure, however, is used to derivethe following signaling and control channels.

S D C C H Stand-alone dedicated control chan-nel is used for the transfer of call control signal-ing to and from the mobile during call setup. Likethe TCHs, the SDCCH has its own SACCH andis released once call setup is complete.

B C C H Broadcast control channel is used inthe BSS to mobile direction to broadcast systeminformation such as the synchronization parame-ters, available services, and cell ID. This channelis continuously active, with dummy bursts substi-tuted when there is no information to transmit,because its signal strengths are monitored by mobilesfor handover determination.

SCH Synchronization channel carries informa-tion from the BSS for frame synchronization.

FCCH Frequency control channel carries infor-mation from the BSS for carrier synchronization.

CCCH Common control channels are used fortransferring signaling information between allmobiles and the BSS for call origination and call-paging functions. There are three common con-trol channels: PCH: paging channel used to call (p a ge) a mobile

from the system. RACH: random access channel used by the mobiles

trying to access the system. The mobiles usethe slotted Aloha scheme over this channel forrequesting a DCCH from the system at call ini-tiation.

AGCH: access grant channel used by the sys-tem to assign resources to a mobile such as a DCCHchannel.Note that the AGCH and the PCH are never used

by a mobile at the same time, and therefore are imple-mented on the same logical channel. All the con-trol signaling channels, except the SDCCH, areimplemented on time slot 0 in different TDMAframes of the 51 multiframes using a dedicatedRF carrier frequency assigned on a per cell basis.The multiframe structure for the SDCCH and itsassociated slow associated control channel(SACC) is implemented on one of the physical chan-nels (TDM slots and RF carriers) selected by the sys-tem operator.

Mobility Management

M obility management is concerned with the func-tions of tracking the location of roamingmobiles and registering the information in appro-priate network elements, and handling connec-tion handoffs for users in the communication process.These functions are discussed in the followingsections.

Connection HandoffsThis may be done between channels in the samecell, between channels in different cells under thesame BSS coverage, or between cells under thecoverage of different BSSs, and even differentMSCs. In GSM, the BSS may autonomously han-dle the connection handoffs in the same cell, orbetween cells under its own coverage. This is calledinternal connection handoffs. The MSC is involvedin managing connection handoffs that need totake place between cells under coverage of twodifferent BSSs. These are called external connec-tion handoffs. When the BSS indicates that an exter-nal handover is required, the decision of whenand whether an external handover should occuris then taken by the MSC. The MSC uses the signalquality measurement information reported by themobile stations (MSs) which are pre-processed atthe BSS for external handover determination.

The original MSC handling a call will alwayskeep control of the call in an external handoverto a different and even a subsequent MSC.

When the BSS performs an internal connec-tion handoff, it informs the MSC at the comple-tion of the process. The need for a connection handoffmay be indicated by the mobile user, throughmessaging on the FACH, for instance, or by theBSS as it keeps tracking the quality of the signalsreceived. The BSS monitors the quality of theradio signal received and also transmits suchresults to the MSC who keeps a more global viewon the radio channels belonging to its BSSs. The

Co m m o ncontrol channels areused fort r a n s f e r r i n gs i g n a l i n gi n f o r m a t i o nbetween allmobiles andthe BSS for call o r i g i n a t i o nand call-paging f u n c t i o n s .


MSC may also initiate the need for a connectionhandoff for traffic reasons in an attempt to bal-ance out the traffic load in the network.

Handling of Location InformationLocation information is maintained and used bythe network to locate the user for call routing pur-poses. The network registers the users locationin a register called the users, HLR, which is asso-ciated with an MSC located in the PLMN, to whichthe user is subscribed. Each BSS keeps broadcast-ing, on a periodic basis, the cell identities on thebroadcast control channels of the cells under itscoverage. The mobiles within each cell keep mon-itoring such information. As changes in locationare detected (from the last information recordedby them), they each report the new location tothe BSS which routes it to the VLR, of the MSCto which it is connected. The VLR , in turn, sends thelocation information to the users HLR, where itis also recorded. In the meant ime, the HLRdirects the old VLR to delete the old visitinglocation of the mobile from its data base, and alsosends a copy of the users service profile to thenew VLR. Location updating is performed by themobility management (MM) protocol sublayerthat will be discussed later in the article.

Call Routing and Signaling

A call may be initiated by a mobile user to anoth-er mobile or a fixed landline user, or in reverse,by a fixed landline user to a mobile. For routing a callto a mobile user, however, the network signalingneeds to first locate the mobile. We will illustrate thisfor the case when a call is initiated by a landline user,

and then comment on the scenario in which thecall is initiated by a mobile to another mobile.When the call is initiated by a mobile to a land-line user, the procedure is rather straightforward.

In the case of a call initiated by a landline user,the PSTN may use the mobile station ISDN num-ber, MSISDN, to route the call to the closest Gate-way MSC within the mobiles PLMN. The GMSC inturn uses the MSISDN to interrogate the mobilesHLR for the routing information required to extendthe call to the visiting MSC of the mobile at thet i m e.This visiting MSC (or more specifically the,V L Rwithin the local MSC) is identified in the mobilesHLR by the MSRN which specifies the visiting MSC.The MSRN is a temporary number allocated bythe VLR and sent to the HLR on location updat-ing, or call initiation. The MSRN should have thesame structure as the MSISDN numbers in the VLRarea where it is allocated. The VLR then initiatesthe paging procedure and the MSC pages the mobilestation with a paging broadcast to all BSSs of thelocation area, as the exact base station area of themobile may not be known. After paging response,the current BSS is located. The RR and MM con-nections are established, during which bothauthentication of the user (for access to the network),as well as cipher mode setting are performed.The VLR then sends the required parameters forcall setup to the MSC, and may also assign the mobilea new TMSI for the call. The MSC sends a setup mes-sage to the mobile station.

The mobile station, on receiving the set-upmessage performs a compatibility check and returnsa call-confirmed message to the network, which mayinclude the bearer capability of the mobile sta-tion. The BSS may at this point assign a trafficchannel, TCH, to the call, or may assign it at alater stage, the latest being on receipt of theconnect message from the mobile station. Ifuser alerting is carried out at the MS, an alerting mes-sage is sent to the calling subscriber. When, thesubscriber answers the call, the MS sends a con-nect message, which at the network side initiates thecompletion of the traffic channel allocation andswitch through of the connection. The connectmessage is progressed to the calling subscriber.The network also sends an acknowledgement tothe MS, that enters the active state.

n GSM protocol architecture.

n LAPDm address field format.


Protocol LayeringArchitecture

T he GSM protocol architecture used for theexchange of signaling messages pertaining tomobility, radio resource, and connection manage-ment functions is shown in Fig. 3. The protocollayering consists of the physical layer, the datalink layer, and the Layer 3. It is noted to theOSI-minded reader to be careful in not confusingthe Layer 3 protocol functions defined by GSM withwhat is normally defined to be the Layer 3 func-tions in the OSI model. The GSM Layer 3 proto-cols are used for the communication of networkresource, mobility, code format and call-related man-agement messages between the various network enti-ties involved. Since, in the OSI model, some of thesefunctions are actually provided by the higher lay-ers, the term message layer may be a moreappropriate term for refering to the Layer 3 in GSM.The message layer (Layer 3) protocol is made upof three sublayers called the resource manage-ment (RR) implemented over the link betweenthe MS and the BSS, the mobility management (MM),and connection management (CM) sublayersproviding the communication between the MSand the MSC. Layer 3 also implements the mes-sage transport part (MTP), level 3, and the sig-naling connection control part of the CCITT SS7 onthe link between the BSS and the MSC (the Ainterface) to provide the transport and address-ing functions for signaling messages belonging to thevarious calls routed through the MSC.In dis-cussing the functionality provided by the Layer 3in the GSM protocol stack, particular attention shouldbe paid to not confuse the details of this layers func-tionality with what is commonly provided by theLayer 3 of the OSI protocol stack. In GSM, theCM, and MM sublayers, for instance, provideactually some of the functionalities which arerealized by the transport, the session, and thepresentation layers of OSI, as will be seen later. Thefunctions of each protocol layer/sublayer is discussedin some detail in the following.

Physical LayerThe physical layer on the radio link was discussed inthe section on radio channel structure. The trafficchannels on the landside are formed from TDM slotsimplemented on 2.048 Mb/s links (E1 trunks).The signaling channels are basically logically mul-tiplexed on an aggregate of the TDM slots.

Link Layer on the Air InterfaceThe data link layer over the radio link (connect-ing the MS to the BSS) is based on a LAPD-like pro-tocol, labeled LAPDm, that has been modifiedfor operation within the constraints set by theradio path. In particular, LAPDm uses no flags (and therefore no bit stuffing) for frame delim-

itation. Instead, frame delimitation in LAPDm isdone by the physical layer that defines the trans-mission frame boundaries. LAPDm uses a LengthIndicator field to distinguish the informationcarrying field from fill-in bits used to fill thetransmission frame. LAPDm uses an addressfield to carry the service access point identifier,(SAPI), (3 bits in this case) which LAPD alsouses to identify the user of the service providedby the protocol. When using command/controlframes, the SAPI identifies the user for which a command frame is intended, and the user trans-mitting a response frame. The format for the addressfield is shown in Fig. 4. The 2-bit link protocoldiscriminator (LPD) is used to specify a particu-lar recommendation of the use of LAPDm, theC/R is a single bit which specifies a command orresponse frame as used in LAPD, and a 1-bit extend-ed address (EA) is used to extend the addressfield to more than one octet (the EA bit in thelast octet of the address should be set to 1, other-wise to 0). The 8-bit is reserved for future uses.

LAPDm uses a control field as is used inLAPD to carry sequence numbers, and to specifythe type of frame. LAPDm uses three types of framesused for supervisory functions, unnumberedinformation transfer and control functions (unac-knowledged mode), and numbered informationtransfer (multiframe acknowledged mode) asused in LAPD. LAPDm uses no cyclic redundan-cy check bits for error detection. Error correctionand detection mechanisms are, instead, provided bya combination of block and convolutional codingused (in conjuction with bit interleaving) in the phys-ical layer. The general frame format for LAPDmis shown in Fig. 5.

Link Layer on the A InterfaceOn the terrestrial link connecting the BSS to theMSC (the A interface), the MTP level 2 of theSS7 protocol is used to provide the OSI Layer 2 func-tions of reliable transport for the signaling messages,such as recovery from transmission errors througherror detection and retransmission.

Message Layer Protocols andFunctions

Radio Resource (RR) ManagementSublayer

The RR management sublayer terminates at the BSSand performs the functions of establishing physi-cal connections over the radio for the purpose oftransmitting call-related signaling information suchas the establishment of signaling and traffic chan-nels between a specific mobile user and the BSS. TheRR management functions are basically imple-mented in the BSS.

n LAPDm general frame format.

LA P D mis aL A P D- l i k eprotocol thathas beenmodified foro p e r a t i o nwithin thec o n s t r a i n t sset by theradio pass.


Mobility Management Sublayer (MM)

The MM sublayer is terminated at the MSC andthe related messages from or to the MS arerelayed transparently in the BSS using the DTAPprocess. The MM sublayer provides functionsthat can be classified into three types of proce-dures. These are called the MM specific procedures,the MM common procedures, and the MM con-nection-related procedures. These proceduresare discussed in the following.

MM Connection Related ProceduresThese are the procedures used to establish, main-tain, and release a MM connection between the MSand the network (MSC) over which an entity ofthe connection management (CM) sublayer canexchange information with its peer. More thanone MM connection may be active at the sametime to serve multiple CM entities. Each CMentity within the MS will have its own MM connection,and each connection is identified by the protocol dis-criminator, and a transaction identifier within therelated signaling messages exchanged. The trans-action identifier is sort of analogous to the callreference used by ISDN to identify signaling mes-sages from different calls on the D channel. Thusparallel calls can be supported by the same MS whichare then identified by a different value for thetransaction identifier parameter. Establishmentof a MM connection requires that no MM-specif-ic procedure (discussed later) be active.

The MM connections provide services to thedifferent entities of the upper connection man-agement (CM) sublayer which currently consistof the call control (CC), the short message ser-vices (SMS), and the call-independent supple-mentary services (SS). An MM connection is initiatedby a CM service request message which identifiesthe requesting CM entity and the type of servicerequired of the MM connection. The servicesprovided by the MM connections include such thingsas enciphering (for privacy of user information), andauthentication (of the users-access to the networkand the service requested) which would be actual-ly provided by the presentation, and application lay-ers in the OSI framework. Each of these serviceswould involve the exchange of multiple messagesbetween the MS and the network before the requiredMM connection is established and the requestingentity within the CM sublayer is notified.

Mobility Management SpecificProceduresThe MM specific procedures do not set up anMM connection. They can only be initiated when n oother MM-specific procedure is running, and no MMconnection is established. These proced u r e sconsist of location updating, and the IMSI attachprocedures. These are discussed in the following.

Location UpdatingLocation updating is the procedure for keeping thenetwork informed of where the mobile is roaming.Location updating is always initiated by the mobilestation on either detecting that it is in a new loca-tion area by periodically monitoring the locationinformation broadcast by the network on the broad-cast channel, and comparing it to the informationpreviously stored in its memory, or by receiving an

indication from the network that it is not knownin the VLR upon trying to establish an MM con-nection. Anytime, the network updates the mobileslocation, it sends it an updatedtemporary mobile sub-scriber i d e n t i f i c a t i o n ( T M S I ) , in ciphered mode,which is stored in the MS and used for subsequentmobile identification in paging and call initiatingoperations. The purpose of using the TMSI asopposed to the users IMSI is to keep the subscribersidentity confidential on the radio link. The TMSI hasno GSM- specific structure, and has significance onlywithin the location area assigned. The TMSI hasto be combined with the location area identifier (LAI)to provide for unambiguous identification outsidethe area where it is assigned.

IMSI AttachThe IMSI attach procedure is the complement of theIMSI detach procedure, a function of the MMcommon procedures (discussed later). Both of theseprocedures are network options whose necessityof usage are indicated through a flag in the sys-tem information broadcast on the BCCH chan-nel. The IMSI detach/attach procedures mark theMS as detached/attached in the VLR (and option-ally in the HLR) on MS power down or power upor subscriber information module (SIM) removed orinserted (The IMSI detach disables the locationupdating function to prevent unnecessary signal-ing overhead on the network). Any incomingcalls, in that case, are either rejected or forward-ed as may be specified by the user). The IMSI is usedto indicate the IMSI as active in the network.This procedure is invoked if an IMSI is activatedin a MS (power up, or SIM insertion) in the cov-erage area of the network, or an activated MS entersthe networks coverage area from outside. The IMSIattach procedure is then performed only if the storedlocation area at the time is the same as the onebeing broadcast on the BCCH channel of theserving cell. Otherwise, a normal location u p d a t-ing procedure is invoked regardless of whether thenetwork supports IMSI attach/detach procedures.

MM Common ProceduresThe MM common procedures can be initiated at anytime while a dedicated radio channel exists betweenthe network and the MS. They do not set up anMM connection, but can be initiated during anMM specific procedure, or while an MM connectionis in place. The MM Common procedures consistof IMSI detach, TMSI reallocation, and authenti-cation/identification. These are discussed next.

IMSI DetachThe IMSI detach procedure is invoked by the mobilestation to indicate inactive status to the network.No response or acknowledgement is returned tothe MS by the network on setting the active flagfor the IMSI.

The IMSI detach procedure is not started if at thetime a MM-specific procedure is active. In that case,the IMSI detach procedure is delayed, if possibleuntil the MM-specific procedure is finished, oth-erwise the IMSI detach request is omitted.

If at the time of a detach request, a radio con-nection is in existence between the MS and thenetwork, the MM sublayer will release any ongo-ing MM connections before the MM detach indi-cation message is sent.

Lo c a t i o nupdating is the procedure for keepingthe networkinformed of where the mobile is roaming.


TMSI Reallocation

The purpose of TMSI reallocation is to provide iden-tity confidentiality. That is, to protect the userfrom being identified and located by an intruder.This procedure must be performed at least at eachchange of the MSC coverage area. Reallocationin any other case is left to the network operator.

If the TMSI provided by a mobile station isunknown in the network, for instance, in the caseof a data base failure, the MS has to provide itsIMSI on request from the network. In this casethe identification procedure has to be performedbefore the TMSI procedure can be initiated.

AuthenticationThe purpose of the authentication procedure isto let the network verify the identity provided bythe user when requested, and to provide a new cipher-ing key to the mobile station. The cases when authen-tication procedures should be used are defined inGSM Recommendation 02.09. The authentica-tion procedure is always initiated and controlledby the network.

IdentificationThis procedure is used by the network to requesta mobile station to provide specific identificationparameters to the network, such as the usersinternational mobile subscriber or equipmentidentifiers (IMSI or IMEI). The mobile station shouldbe ready to respond to an identity request mes-sage at any time while RR connection exists betweenthe mobile and the network.

Connection ManagementSublayer (CM)

T he CM sublayer terminates at the MSC and con-tains entities that currently consist of CC includ-ing call-related supplementary services, SMS, andcall independent supplementary services support(SS). Once a MM connection has been established,the CM can use it for information transfer. TheCC entity uses the CCITT Q.931 protocol, with minormodifications, for the communication of call con-trol-related messages between the MS and the MSC.The SMS is a GSM-defined service that provides forspeedy packet mode (connectionless) commu-nication of messages up to 140 bytes between the MSand a third party service center. These messages canbe sent or received by the mobile station while a voiceor data call is in the active or inactive state. It is accept-able, however, if the service is aborted while acall is in a transitional state such as handover or busy-to-idle. The service center is responsible for thecollection, storage, and delivery of short mes-sages, and is outside the scope of GSM.

BSS Application Part (BSSAP)

T he BSS, in addition to providing the channel switch-ing and aerial functions, performs radio resourcemanagement, and interworking functions b e t w e e nthe data link protocols used on the radio and theBSS-MSC side for transporting signaling-relatedm e s s a g e s . These functions are provided by theBSS Management Application Process (BSSMAP),and the Direct Transfer Application Process (DTAP).

The BSSMAP is used to implement all proce-dures between the MSC and the BSS that requireinterpretation and the processing of informationrelated to single calls, and resource manage-ment. Basically, the BSSMAP is the process with-in the BSS that controls radio resources inresponse to instructions from the MSC (in thatsense, the BSSMAP represents the RR sublayerto the MSC). For instance, the BSSMAP is usedin the assignment and switching of radio channelsat call setup, and handover processes.

The DTAP process is used for the transparenttransfer of MM/CM signaling messages between theMS and the MSC. That is, the DTAP functionprovides the transport level protocol interworkingfunction for transferring Layer 3 signaling messagesfrom and to the MS to and from the MSC with-out any analysis of the message contents.

Signaling Transport Protocols

T he CCITT SS7 MTP and SCCP protocols areused to implement both the data link and theLayer 3 transport functions for carrying the call con-trol and mobility management signaling messageson the BSS-MSC link. The MM and CM sublayersignaling information from the mobile station is rout-ed over signaling channels (such as the DCCH,SACCH, FACCH) to the BSS from where they aretransparently relayed through the DTAP processto an SCCP, of CCITT SS7 type logical channel,assigned for that call, on the BSS-MSC link for trans-mission to the peer CC entity in the MSC for pro-cessing. Similarly, any call signaling informationinitiated by the MSC on the SCCP connection isrelayed through the DTAP process in the BSS to theassigned signaling channel, using the LAPDmdata link protocol, for delivery to the mobile station.

The interworking between the Layer 2 proto-col on the radio side and the SS7 on the BSS-MSClink is provided by a distribution data unit withinthe information field of the SCCP. These param-eters are known as the discrimination, and the datalink connection identifier (DLCI) parameters.The discrimination parameter (currently dedicat-ed one octet) uses a single bit to address a messageeither to the DTAP or the BSSMAP processes. TheDLCI parameter (sized one octet) is made up of twosubparameters that identify the radio channel type(such as the DCCH, SACCH, FACCH), and theService Access Point Interface(SA P I) value (in theLAPDm protocol) used for the message on the radiolink. The SCCP provides for the logical multiplexingof signaling information from different calls ontothe same physical channel (such as a single 64 kb/sslot of a 2.048 Mb/s E1 trunk) on the BSS-MSC link.For each call supported by a BSS, an SCCP logicalconnection is established on the BSS-M S C link. A n yinformation pertaining to a specificcall flows throughits associated SCCP connection and that is howsignaling information exchange pertaining t odifferent calls are identified in the BSS or MSC.

The connectionless service mode of the SCCPis also supported for the transfer of OA&M relat-ed messages as well as BSSMAP messages thatdo not pertain to any specific call (Note that BSSMAPmessages pertaining to specific calls, such as hand-off messages, are transmitted using the SCCPconnection established for the call). The SCCP rout-ing function uses the SubSystem Number (SSN)

The authenti-cation proce-dure allowsthe networkto verify theidentity pro-vided by theuser whenr e q u e s t e d ,and to pro-vide a newciphering keyto the mobiles t a t i o n .


in the Service Information Octet (SIO) within theMTP level 3 message to distinguish messagesaddressed to the OA&M function from thoseaddressed to either the DTAP or the BSSMAP appli-cation parts. The high-level address translation capa-bility of the SCCP, known as global title translation,may then be used to provide additional address-ing capabilities such as use of E.164 numberingfor addressing different OA&M entities. Theglobal title translation feature of the SCCP also pro-vides the MSC the capability to address signalingmessages to remote MSCs that may be located ina different PLMN.

The interworking functions between the CM, MMand BSSMAP entities and the correspondingentities of the SS7 (i.e., the ISDN-UP), MAP, SCCP,and the transactions capabilities application part(TCAP) is provided by the MSC.

Paging

P aging messages for mobiles are sent via theBSSMAP to the BSS as a connectionless mes-sage through the SCCP/MTP. The paging mes-sage may include the mobiles IMSI in order toallow derivation of the paging population num-ber. A single paging message transmitted to the BSSmay contain a list of cells in which the page is to bebroadcast. The larger the paging area is defined, thelower the frequency of location updates and hencethe associated traffic overhead on the network.On the other hand, large paging areas result inincreased use of transmitting power as well as theradio resources (channels). Therefore, the optimumsize for the paging area (location area) is detem-ined by a proper balance between the costs ofpaging and the costs of location updates.

The paging messages received from the MSC arestored in the BS, and corresponding paging messagesare transmitted over the radio interface at the appro-priate time. Each paging message relates to only onemobile station and the BSS has to pack the pages intothe relevant 04.08 paging message (include Layer3 information). Once a paging message is broad-cast over the radio channel(s), if a response messageis received from the mobile, the relevant signalingconnection is set up towards the MSC and thepage response message is passed to the MSC.

Summary Remarks

T he description of the GSM network functions,system architecture and protocols are spreadover a large number of GSM documents, each ofwhich contains many details with some of the crit-ical issues and highlights covered within those details.Therefore, it is not an easy task to extract outsome of the crucial concepts and design specifics,

and present it in some logical and well-relatedformat. I have tried my best, however, to achieve thisgoal in this article.

This article was meant to provide a concise, brief,but adequately detailed description of the GSM sys-tem and protocol architecture that can serve as aquick, rather self-contained conceptual frame-work for extending and relating the mobility-specificfunctions of the next generation personal com-munication networks to the GSM network functions,and the protocols used to achieve them. Finally, alist of references have been provided for anymore detailed information on the issues addressedin the article.

AcknowlegementsThe author would like to thank Bomber Bishopand David Leeper from Motorola, and PrapeepSherman from AT&T for their careful reading ofthe original manuscript and for providing usefulcomments.

References[1] W.C.Y. Lee, Spectrum Efficiency in Cellular, IEEE Trans. on Veh. Tech. ,

vol. 38, no. 2, May 1989.[2] W.C.Y. Lee, Spectrum Efficiency and Digital Cellular, 38th IEEE

Veh. Tech. Conf. Records, pp.643, June 1988..[3] GSM Recommendation 04.03, MS-BSS Interface: Channel Struc-

tures and Access Capabilities.[4] GSM Recommendation 05.01, Physical Link Layer on the Radio

Path (General Description).[5] GSM Recommendation 05.02, Multiplexing and Multiple Access

on the Radio Path.[6] Conference Proceedings, Digital Cellular Radio Conference, Hagen FRG,

Oct. 1988.[7] GSM Recommendation 002.02, Bearer Services Supported by a PLMN.[8] GSM Recommendation 09.01, General Aspects on PLMN Inter-

w o r k i n g . [9] GSM Recommendation 03.04, Signaling Requirements Related to

Routing of Calls to Mobile Subscribers.[10] GSM Recommendation 08.02, BSS-MSC Interface-Interface Principles.[11] GSM Recommendation 08.04, BSS-MSC Layer 1 Specifications.[12] GSM Recommendation 08.06, Signaling Transport Mechanisms

for BSS-MSC Interface.[13] GSM Recommendation 09.02, Mobile Application Part (MAP)

Specification.[14] GSM Recommendation 08.08, BSS-MSC Layer 3 Specifications.[15] GSM Recommendation 04.08, Mobile Radio Interface-Layer 3

Specifications.

BiographyMO E RA H N E M A received a B.S. degree in engineering science from theUniversity of Kentucky at Lexington in 1978 with honors. He receivedthe M.S. degree and the more advanced engineering degree in Avion-ics from MIT in 1981. From 1983 to 1984, he taught and studied com-munication sciences at Northeastern University from which he also receivedthe Engineer degree in electrical and computer engineering with Ph.D-level coursework. He worked as a senior communication design engi-neer at infinet in Andover, Mass from 1984 to 1985, where he designedthe digital signal processing firmware for a 4800 baud modem. From 1985to 1989, he worked as a member of the technical staff at GTE Laboratories,and developed a new system architecture for fast packet switchingbased on the slotted ring concept (published in IEEE Transactions on Com -m u n i c a t i o n s, April 1990). From 1989 to 1991, he worked as a princi-pal engineer at Arinc on the design and analysis of air/ground communicationnetworks for the airlines industry. He joined Motorola as a principalcommunication engineer in 1992, and since has been working on theIridium satellite project. His interests include wireless networks, com-munication systems, and digital signal processing.

The o p t i m u msize for thepaging a r e ais determinedby a properb a l a n c ebetween t h ecosts ofpaging andthe costs ofl o c a t i o nu p d a t e s .

overview of the gsm system and protocol architecture

Documents