Nhiều giao thức truy cập đối với truyền thông di động P2

frequency spectrum as efficiently aspossible, to limit infrastructure deployment and prevent some of the problems listed in relation with small cell radii; • handover procedures should b

Alex Brand, Hamid Aghvami Copyright  2002 John Wiley & Sons Ltd ISBNs: 0-471-49877-7 (Hardback); 0-470-84622-4 (Electronic)

2

CELLULAR MOBILE COMMUNICATION SYSTEMS:

FROM 1G TO 4G

The basic principles of cellular communications were explained in the introductorychapter, and terms such as cluster size and reuse efficiency introduced In the following,some more considerations on the advantages and limitations of the cellular conceptwill be made before reviewing first generation (1G) and second generation (2G) cellularcommunication systems Moving on to 3G, we will first discuss the initial requirementsaccording to which 3G systems were designed, and then to what extent they are likely

to be satisfied by first releases of ‘true’ 3G systems on the one hand, and evolved 2Gsystems on the other While special attention is paid to the GSM evolution route and toUMTS, the initially European, but now global proponent for 3G cellular communications,evolutionary routes from cdmaOne to cdma2000 are also reviewed

The emergence of the Internet witnessed in the 1990s is expected to have a fundamentalimpact on the telecommunications industry, including cellular communications We willexamine how this affects 3G, what new requirements arise and how these can be metthrough subsequent releases of UMTS Finally, looking further into the future, possiblemanifestations of 4G systems will also be discussed, including the role wireless LANtechnologies may play in this context

2.1 Advantages and Limitations of the Cellular Concept

It was outlined in the previous chapter how the cellular concept allows use of the sameset of frequencies in multiple cells, and how it is in theory possible to arbitrarily increasecapacity to match growing demand for wireless communication services through cellsplitting In practice, however, there are certain limitations With smaller and smallercells as a result of cell splitting, it becomes increasingly difficult to place base stations atthe locations that offer the necessary radio coverage [1] Furthermore, as the cell radiusdecreases, the handover rate increases This will place a costly burden onto the network

in terms of increased signalling load and, given the non-zero probability of handoverfailures, the call-dropping probability may increase, particularly for fast moving mobiles.Finally, it would be wasteful to deploy a large number of base stations covering smallcells in areas with low traffic density

All these considerations call for a network topology where different cell types aredeployed concurrently, to provide capacity, where required, and at the same time ensureuniversal coverage According to the topography and other conditions, base stations may

cover a macrocell (also called umbrella cell ), a microcell (typically to be found in city centres, on highways or even indoors), or a picocell (usually deployed indoors), with

diameters of several kilometres, up to one kilometre and possibly as little as a few tens ofmetres respectively These cells of different types, each cell-type constituting a separatelayer of a multi-layered system, will serve overlapping coverage areas In this way, afast moving mobile terminal, for instance, can be served by a macro- or umbrella cell,

in order to limit the number of handovers per call, while slower moving mobiles in thesame coverage area are normally supported by microcells

For traffic management purposes, these multiple layers can be organised in a

hierar-chical cellular structure (HCS, see e.g Reference [64]), in which overflow traffic from

microcells is handed ‘up’ to the hierarchically higher macrocellular layer A space segmentmay be added to the terrestrial segment, where satellite spot beams overlay clusters

of terrestrial macrocells [65], as illustrated in Figure 2.1 GSM, the most important 2Gcellular system in operation today, is one example of existing cellular systems that cansupport multiple cell layers

From these considerations, a first set of requirements for the efficient operation ofcellular communication systems can be inferred:

• the system should use available resources (i.e frequency spectrum) as efficiently aspossible, to limit infrastructure deployment and prevent some of the problems listed

in relation with small cell radii;

• handover procedures should be fast, require a minimum amount of signalling ably to be exchanged on dedicated signalling resources to avoid an impairment ofongoing communications), and be robust to avoid dropped calls; and

(prefer-• the system should be able to support multiple cell layers, for instance in the shape ofhierarchical cellular structures, to provide high-speed mobility and hot-spot capacity.This list will be expanded in the following, first with respect to the initial 3G require-ments identified in the early 1990s, according to which first releases of 3G systems weredesigned, then with respect to additional requirements identified later on, mainly as aresult of the Internet revolution Before we do that, the older 1G and 2G systems will bereviewed briefly

Macrocell Microcell Satellite spot

beam

Figure 2.1 Microcells, macrocells, and a satellite spot-beam

2.2 1G and 2G Cellular Communication Systems

2.2.1 Analogue First Generation Cellular Systems

In the late 1970s and early 1980s, various first generation cellular mobile communicationsystems were introduced, characterised by analogue (frequency modulation) voice trans-

mission and limited flexibility The first such system, the Advanced Mobile Phone System

(AMPS), was introduced in the US in the late 1970s1 Other 1G systems include the

Nordic Mobile Telephone System (NMT), and the Total Access Communications System

(TACS) The former was introduced in 1981 in Sweden, then soon afterwards in otherScandinavian countries, followed by the Netherlands, Switzerland, and a large number ofCentral and Eastern European countries, the latter was deployed from 1985 in Ireland,Italy, Spain and the UK

While these systems offer reasonably good voice quality, they provide limited spectralefficiency They also suffer from the fact that network control messages — for handover

or power control, for example — are carried over the voice channel in such a way thatthey interrupt speech transmission and produce audible clicks, which limits the networkcontrol capacity [7] This is one reason why the cell size cannot be reduced indefinitely

to increase capacity

Such constraints did not prevent these systems from enjoying considerable success withthe public, so that subscriber numbers were still growing in the mid 1990s In Italy, forinstance, they peaked at four million users in March 1998, corresponding to a penetration

of roughly 7%, and (while less impressive in absolute figures) penetration exceeded 10%

in most Scandinavian countries However, they have been increasingly thrust aside by2G systems in most parts of the world In the meantime, after closing down 1G systems,spectrum refarming from 1G to 2G has taken place in several countries

2.2.2 Digital Second Generation Systems

Capacity increase was one of the main motivations for introducing 2G systems in theearly 1990s Compared to the first generation, 2G offers [69]:

• increased capacity due to application of low-bit-rate speech codecs and lower

frequen-cy reuse factors (the cluster size can be as low as three compared to seven in analoguesystems for example, see also Subsection 2.3.2 on evolved 2G systems);

• security (encryption to provide privacy, and authentication to prevent unauthorised

access and use of the system);

• integration of voice and data owing to the digital technology; and

• dedicated channels for the exchange of network control information between mobileterminals and the network infrastructure during a call, in order to overcome the limita-tions in network control of 1G systems (note though, that handover-related signallingstill steals into the traffic channel in GSM)

1 According to Lee, the FCC released frequencies for cellular communication systems in the 800 MHz band

in 1974, and AMPS served 40 000 mobile customers in 1976 [66] Other sources indicate later launch dates though, for instance 1979 for a pre-operation AMPS system in Chicago [67] See also Reference [68], which provides a detailed history of cellular communications.

At the time of writing, the major force driving the growth in cellular communicationsare still such 2G systems, which were first introduced in Europe, Japan, and the US, andare now in operation worldwide.

Initially, these systems operated only at 900 MHz (800 MHz in the US and Japan),with up-banded versions at 1800 MHz (1900 MHz in the US, 1500 MHz in Japan)coming soon after These up-banded systems are aimed primarily at people movingaround in cities at pedestrian speeds with hand-held telephones They are referred to

as Personal Communications Networks (PCN) in Britain and Personal CommunicationsSystems (PCS) in the US, to distinguish them from the ‘classical’ cellular systems oper-ating below 1 GHz

One of these 2G systems needs to be singled out, GSM, a TDMA-based systemwith optional slow frequency hopping The acronym stood initially for ‘Groupe Sp´ecialMobile’, but fortunately lends itself conveniently to ‘Global System for Mobile Commu-nications’ With a subscriber number close to 500 millions in early 2001, a footprintcovering virtually every angle of the world, and a share of the digital cellular marketclose to 70% in February 2001 (according to figures published by the GSM Associ-ation [70]), it truly deserves this name Introduced in early 1992, already by the end

of 1993, GSM networks had been launched in more than 10 European countries, andoutside Europe for instance in Hong Kong and Australia From 1994, GSM graduallyconquered the markets in the remaining European countries and large parts of the rest ofthe world

The pan-European standardisation effort leading to GSM was initiated by the Conf´erence

Europ´eenne des Administrations des Postes et des T´el´ecommunications (CEPT) in 1982

with the formation of the Groupe Sp´ecial Mobile (GSM) [3] In 1988, the EuropeanTelecommunications Standards Institute (ETSI) was founded, which was from then onresponsible for the evolution of the GSM standards [10] ETSI is still formally respon-sible for the GSM standards, but the technical work relating to system evolution is nowcarried out within the framework of the Third Generation Partnership Project (3GPP), abody created in 1998 to standardise UMTS The GSM evolution work was added in theyear 2000 ETSI is a 3GPP member along with other standardisation bodies and interestgroups from around the world In the context of cellular systems, ETSI’s role is now

to transform technical specifications created in 3GPP into regional standards for Europe,whether this is for evolved GSM or for UMTS

While GSM is currently the uncontested standard for 2G digital cellular communications

in Europe, there is no clear dominance of a single digital standard in most of the rest

of the world In the US, for instance, there are essentially two types of 2G cellular

systems operating at 800 MHz, which are incompatible with each other The first is aTDMA system called North American Digital Cellular or Digital AMPS (D-AMPS), andsometimes simply referred to as TDMA, according to the basic multiple access scheme

it is based on This can create confusion, since there are other TDMA-based cellularsystems, notably GSM The second system, which was launched later, is cdmaOne, thefirst operational CDMA system [71] The relevant air-interface specifications are the so-called interim standards IS-136 (for D-AMPS) and IS-95 (for cdmaOne)

The fragmentation observed in the US was accentuated when the Federal nications Commission (FCC) sold frequencies in the 1900 MHz band for PCS withoutmandating the technology to be used On top of up-banded D-AMPS and cdmaOnesystems, this allowed a 1900 MHz version of GSM to enter the US market Similar

Commu-developments could also be observed in the rest of the Americas, although Brazil, ratherthan selling spectrum at 1900 MHz, decided in the year 2000 to auction 1800 MHzspectrum for PCS This decision favours GSM because, at least at the time of writing,1800-MHz versions of D-AMPS and cdmaOne equipment were not available.

The first and most popular Japanese 2G digital standard is Personal Digital Cellular(PDC) It was later complemented by the Personal Handyphone System (PHS), somehow

a hermaphrodite between mobile and cordless system, which caters only for low mobility,but is popular for certain applications owing to relatively high data-rates (32 kbit/s,later enhanced to 64 kbit/s) Both standards are TDMA-based and have not seen widedeployment outside Japan PDC has received some attention outside Japan, though, owing

to the tremendous success enjoyed by the ‘i-mode’ service since its launch in February

1999 This is a service similar to WAP (wireless application protocol) services, but (atleast at the time of writing) rather more popular It enables access to some form of Internetthrough mobile handsets, and runs on top of the packet-overlay added to PDC, referred

to as PDC-P In the late 1990s, a cdmaOne system was launched in Japan CdmaOnesystems enjoy considerable success also elsewhere in the Far East, most notably in SouthKorea, where an IS-95 derivative dominates the 2G market

2.3 First 3G Systems

2.3.1 Requirements for 3G

Already before the launch of 2G systems, the research community started to think aboutrequirements for a new, third generation of mobile communication systems and aboutpossible technological solutions to meet them (see for instance References [1,7,12,72,73]).Before 3GPP was established, ETSI was one of the major players regarding the standardis-ation of 3G systems It called its 3G representative Universal Mobile TelecommunicationsSystem (UMTS) and established a number of requirements, according to which such asystem should be designed Those of interest here are the ones relating to the air interface

or radio access, the so-called UMTS Terrestrial Radio Access (UTRA), which are listed inReference [74] This list appears to capture most of the 3G requirements stated in the liter-ature of the early 1990s, when 3G emerged as a mainstream research topic It is thereforesummarised in the following The UTRA requirements pertain to four different cate-gories, namely to bearer capabilities, operational requirements, efficient spectrum usage,and finally complexity and cost

2.3.1.1 Bearer Capabilities

(1) UMTS has to deliver services with bit rates up to 2 Mbit/s indoors, at least

384 kbit/s in suburban outdoor and at least 144 kbit/s in rural outdoor ments

environ-(2) UMTS should be flexible in terms of service provision, in particular:

— negotiation of bearer service attributes should be possible;

— parallel bearer services to enable service mix should be possible;

— circuit-switched and packet oriented bearers should be provided;

— variable-bit-rate real-time capabilities should be provided; and

— scheduling (and pre-emption) of bearers according to priority should bepossible.

(3) UMTS has to provide services with a wide range of bit-rates in a variety of differentenvironments with bit error rates (BER) as low as 10−7 for certain services

To deliver optimum performance in all of these, a very flexible air interface isrequired

(4) UMTS should provide seamless handover between cells of one operator, possiblyeven between cells of different operators Seamless mean in this context, that thehandover must not be noticeable to the user

2.3.1.2 Operational Requirements

(5) Compatibility with services from the following existing core networks must beprovided: ATM bearer services, GSM services, Internet Protocol (IP) based services,and ISDN services

(6) Automatic radio resource planning should be provided, if such planning is required

2.3.1.3 Efficient Spectrum Usage

(7) The air interface should make efficient use of the radio spectrum for typical mixtures

of different bearer services

(8) Given the asymmetric UMTS frequency allocation and the likely overall trafficasymmetry (due to Web browsing, for instance), the air interface should supportoperation in unpaired frequency bands or, according to Reference [74], ‘variabledivision of radio resource between uplink and downlink resources from a commonpool [must be provided]’

(9) UMTS should allow multiple operators to use the band allocated to UMTS withoutcoordination (this includes public, private, and residential operators)

(10) UMTS should allow flexible use of various cell types and relations between cells(e.g indoor cells, hierarchical cells) within a geographical area without undue waste

of radio resources

2.3.1.4 Complexity and Cost

(11) Handportable and PCMCIA-card-sized UMTS terminals should be viable in terms

of size, weight, operating time, range, effective radiated power and cost

(12) The development and equipment cost should be kept at a reasonable level, takinginto account the cost of cell sites, the associated network connections, signallingload and traffic overhead

Goodman stated in the early 1990s a vision for 3G as to ‘create a single networkinfrastructure, that will make it possible for all people to transfer economically any kind

of information between any desired locations’, a statement which appears to cover theessence of the above listed requirements He added that ‘a unified wireless access [willreplace] the diverse and incompatible 2G networks with a single means of wireless access

to advanced information services’ [1] The latter is something which may well fail tomaterialise, as will be pointed out later on

Two questions, which will be discussed in the following, arise here:

(1) To what extent can evolved 2G systems meet these requirements?

(2) Do currently specified 3G systems meet these requirements?

2.3.2 Evolution of 2G Systems towards 3G

With the large subscriber base and the considerable investment in 2G infrastructure inmind, it cannot come as a surprise that operators of such systems are keen to protectthis investment Assuming that there is a market need for 3G systems meeting the aboverequirements list (one would assume that operators having paid billions of dollars for 3Glicenses must have made this assumption), there appear to be two ways to achieve this:

• Standardise 3G in a manner so that at least part of the 2G network infrastructurecan be reused In the case of GSM and UMTS, this has materialised to some extent.Certain GSM core network nodes can potentially be reused for UMTS Also, theUMTS handover requirements state that handover to 2G systems, e.g GSM, should bepossible Furthermore, it was at least attempted to choose design parameters for UTRAwhich ease implementation of dual-mode GSM/UMTS handsets; dual-mode operation

is expected to be a standard feature of most UMTS handsets Correspondingly, it ispossible to deploy UMTS gradually in a GSM system, where in a first phase onlyselected sites are equipped with UMTS base stations, while universal coverage isprovided by GSM

• Evolve capabilities of 2G systems to meet 3G requirements, for instance throughenhancements to the air interface

Given the importance of GSM and the large number of advanced features which havebeen or are still being standardised for this system, it will be discussed briefly to whatextent such an evolved GSM system may fulfil 3G requirements from an air-interfaceperspective The main air-interface related enhancements to GSM, which are alreadystandardised (as discussed in detail in Chapter 4), are:

• higher data-rates for circuit-switched services through aggregation of several time-slotsper TDMA frame with High Speed Circuit-Switched Data (HSCSD) [75];

• efficient support of non-real-time packet-data traffic with the General Packet RadioService (GPRS), which entails enhancements to both the air interface [54] and thenetwork [76];

• higher data-rates on individual GSM physical channels through use of higher ordermodulation schemes within the existing carrier bandwidth of 200 kHz, referred to

as Enhanced Data Rates for Global Evolution (EDGE) ‘Plain GSM’, HSCSD, andGPRS can then exploit these higher data-rates (see Reference [77])

Ignoring network constraints, with HSCSD, GSM could in theory offer user data-rates

of at most 8× 14.4 kbit/s = 115.2 kbit/s for circuit-switched data traffic With GPRS,

a data-rate of up to 8× 21.4 kbit/s = 171.2 kbit/s can be provided for packet traffic

for the case where no forward error correction coding is used [54] Neither do thesedata-rates meet the UMTS requirements, nor can they be achieved in an economicalmanner, since aggregation of eight time-slots in a TDMA frame would result in ratherpower-thirsty handsets and raise other issues such as how to dissipate the additional heatbeing generated A detailed discussion on realistic data-rates is provided in Chapter 4 Byapplying EDGE to GPRS, without error coding, the data-rates per slot can be increased

to 59.2 kbit/s, hence with eight time-slots up to 473.6 kbit/s could be achieved, whichexceeds the UMTS requirements for all environments but indoors Note though that apartfrom requiring time-slot aggregation, due to lack of error protection, such throughputlevels can only be achieved at very high carrier-to-interference ratios, which has obviouslyrepercussions on cell planning and capacity With mobile terminals capable of aggregatingeight time-slots, the suburban requirements of 384 kbit/s could be met while allowing formoderate error protection, which would reduce the required CIR a bit

GSM was initially designed with specific tele-services in mind, each of them beingindividually standardised GPRS and the general bearer service introduced mainly, butnot only for HSCSD [78] provide increased flexibility, as do other standardisation workitems related to services2 However, the constraints of the GSM air interface will notallow the second and third requirement listed above to be satisfied fully For instance,GPRS was not designed for real-time packet-data services and in its early versions, it

is only suitable for real-time services with severe restrictions (if at all) Proper real-timecapabilities are being added to GPRS, but provision of high and variable bit-rate real-timeservice will remain restricted

Automatic resource planning is normally not provided, although dynamic channelassignment (DCA) and dynamic resource assignment (DRA) could be used to ease theplanning process Alternatively, a combination of slow frequency hopping with fractionalloading may allow the deployment of a one site/three sector (1/3) reuse-pattern for carriersnot carrying broadcast channels, eliminating the need for frequency planning for thesecarriers [79,80] A system scenario for GSM with two hierarchical layers is described inReference [81], where 1/3 reuse is applied to hopping channels of macrocells, 1/3 or even1/1 reuse to those of microcells, and carriers with broadcast channels for microcells areplanned adaptively The GPRS COMPACT mode [82], a stand-alone data-only solution(i.e without support of GSM circuit-switched services) relies also on a 1/3 reuse Suchmatters are discussed in detail in Chapter 4

An evolved GSM system will fail to meet 3G requirements on two more counts.Firstly, asymmetric frequency allocation is not possible (it is possible to provide asym-metric services, but the total resources managed by base stations are always symmetric).Secondly, for public operation, GSM certainly does not allow multiple operators to use

the total allocated band for GSM without co-ordination (i.e without reserving for each

operator a dedicated part of the available spectrum) However, if the respective ment (number (9) in the list) is interpreted in this manner, it will also not be met byUMTS Residential (or private) use without co-ordination is a different matter; a GSM-based cordless telephony system is a part of the GSM evolution story As far as efficientuse of the radio spectrum is concerned, refer to the discussion of multiple access schemes

require-2 Tele-services are fully specified end-to-end services providing the complete capability, including terminal equipment functions, for communication between users Bearer services, by contrast, provide only the capability for the transmission of signals between user–network interfaces; they provide bearers used by tele-services.

in Chapter 3 for comments on spectral efficiency of TDMA used in GSM, and CDMA,the main alternative.

In conclusion, 2G systems continue to evolve and the boundary between advanced 2Gand ‘true’ 3G systems is increasingly being blurred GSM, for instance, with its wealth

of implemented and imminent features, is now rightly referred to as an advanced 2Gsystem (or alternatively, a generation 2.5 system [72]), and may, in certain manifestations,become an integral part of 3G systems However, as stand-alone systems, 2.5G systemswill struggle to meet all 3G requirements In particular (at least in the case of an evolvedGSM system), they will only to a limited extent be able to provide efficient support ofhigh and variable bit-rate multimedia services

2.3.3 Worldwide 3G Standardisation Efforts

These high and variable bit-rate multimedia services are, at least from today’s perspective,exactly those services that may create market demand for new systems This is one reasonwhy the mobile communication communities in Europe, the Far East, and the US specified

‘true’ 3G systems There are other reasons, as well For instance, handset and infrastructuremanufacturers must naturally be interested in deployment of new systems

Japan was pushing particularly hard for 3G and leads on 3G deployment, mainly because

of overcrowded 2G systems, but likely also motivated by a wish to break out of a nological isolation in which its industry found itself with PDC and PHS

tech-The International Telecommunications Union (ITU), which refers to 3G systems aseither Future Public Land Mobile Telecommunications System (FPLMTS) or, more handy,International Mobile Telecommunications 2000 (IMT-2000), initially had the intention ofcontrolling the 3G standardisation process in a manner such that a single system wouldemerge This would have allowed, for the first time, worldwide roaming with a singlehandset, as envisaged by Goodman in Reference [1] The idea was that the differentregions of the world would submit system proposals capable of meeting a given set ofrequirements The proposal best meeting these requirements would then be selected or,

if this was not possible, an attempt would be made to merge different proposals into asingle one in a consensus building phase

While all bodies standardising such systems actually submitted their proposals to theITU in the middle of 1998 [83], it became clear that the ITU would not be in a posi-tion to enforce a unified system Instead, it would essentially have to approve all viableproposals meeting the core ITU 3G requirements, which the regional bodies intend toimplement To complicate matters, while ETSI in Europe and the Association of RadioIndustries and Businesses (ARIB) in Japan put in place their own procedures to selectone such proposal, there were no concerted efforts in the US towards 3G standardisa-tion As with 2G systems, it was believed that the marketplace should choose a system,resulting in several 3G proposals from the US For these reasons, ITU then advocated

‘a “family of systems” concept, defined as a federation of systems providing IMT-2000

service capabilities to users of all family members in a global roaming offering’ [83] Thelatter entailed efforts during the consensus building phase at least to enable worldwideroaming, for instance by facilitating the implementation of multi-mode terminals, giventhat a complete harmonisation seemed unachievable

Eventually, two main camps (with various sub-streams) formed The first one is united

in the original Third Generation Partnership Project (3GPP) mentioned previously, dealing

with the standardisation of UMTS and the evolution of GSM, the second one in a similarstructure named 3GPP2, dealing with cdma2000, an evolution of cdmaOne The systemsbeing developed by these two organisations are based on different core network standards.Moderately successful air-interface harmonisation efforts have taken place in the frame-work of an ‘operator harmonisation group’, but although similar air-interface technologiesare considered for UMTS and cdma2000, the systems remain essentially incompatible.Other harmonisation efforts in the same framework resulted in the introduction of ‘hooks’and extensions in the relevant standards, allowing the 3GPP air interface to be deployed

on a 3GPP2 network infrastructure and vice versa, as agreed in the second quarter of

1999 What relevance this option will have in practise remains to be seen

2.3.4 The Third Generation Partnership Project (3GPP)

In Europe, several radio interface proposals were seriously considered within ETSI forUMTS Five concept groups were set up in ETSI SMG2 during 1997, classifying thedifferent proposals according to the basic multiple access schemes employed These werewideband CDMA (WCDMA), wideband TDMA, hybrid TDMA/CDMA (referred to asTD/CDMA), orthogonal frequency-division multiplexing (OFDM) with a TDMA elementfor multiple access, and opportunity driven multiple access (ODMA) Strictly speaking,ODMA is not a basic multiple access scheme, but rather a technique that transformsmobile terminals into relay stations Signals from a terminal far away from a base stationcan be relayed by other terminals nearer to it (potentially over multiple hops) to improvecoverage and lower transmission power, thereby reducing interference and thus increasingspectral efficiency

The strongest contenders were WCDMA and TD/CDMA, while ODMA was eventuallysuggested as an option on top of whatever basic multiple access scheme would be chosen.After a heated debate, it became evident that a unanimous decision for only one proposalwas not possible While a majority preferred WCDMA, it was appreciated that TD/CDMAwould lend itself better to time-division duplexing suitable for operation in unpairedfrequency bands As a compromise choice, rather than picking both frequency-divisionduplex (FDD) and time-division duplex (TDD) mode of one of the two proposals, it wasdecided to choose the WCDMA FDD mode together with the TD/CDMA TDD mode.This is reflected in the ETSI candidate submission [84] to ITU In the following, thesetwo modes are referred to as UTRA FDD and UTRA TDD respectively

Japan, in its own selection process, was also considering various proposals and finallyopted for a WCDMA based system [85] very similar to the ETSI WCDMA mode SinceJapanese companies had also contributed to the WCDMA concept in ETSI and were eager

to avoid finding themselves in similar technological isolation as with PDC, it was onlynatural that they decided to join forces with ETSI This led eventually to the creation of3GPP in 1998, which took over detailed standardisation of the UTRA FDD and TDD

modes from ETSI (with UTRA now standing for universal terrestrial radio access rather

than UMTS terrestrial radio access) Apart from ETSI and Japanese bodies, 3GPP wasalso joined by some of the US and the South Korean WCDMA proponents Furtherinformation on this topic can be found in Reference [86]

The core network to be used for the 3GPP system is an evolved GSM core network As

a result, work on specifications dealing with protocols and network components common

to GSM and UMTS was transferred from ETSI to 3GPP in 1998 3GPP can therefore be

viewed as providing a 3G evolution route mainly for GSM operators, with Japanese PDCoperators also adopting this route to 3G.

The most important D-AMPS operators decided also to join forces with the GSM camp,albeit in a somewhat different manner They decided to enhance their 2G digital voicecapability with a GPRS-based high-speed packet-data capability To meet the ITU data-raterequirements for 3G (which are more relaxed than the respective UMTS requirements),GPRS would have to be enhanced through EDGE, leading to enhanced GPRS (EGPRS)

This is why EDGE now stands for enhanced data-rates for global rather than GSM

evolution, as it did earlier Initially, it was planned that EGPRS would be closely coupled

to D-AMPS, for instance by enabling D-AMPS signalling messages to be directed todual-mode D-AMPS/EGPRS terminals via EGPRS Recently, however, one key D-AMPSoperator has reconsidered its evolution route to 3G and decided to deploy full GSM/GPRSfirst, with a view to introduce UMTS later on In such a scenario, EGPRS may still bedeployed as an enhancement to the existing GPRS infrastructure, but D-AMPS would berelegated to a more or less ‘stand-alone’ 2G technology

GPRS and first release EGPRS radio access networks need to be connected to a ‘plainGSM’ core network Since further releases can be connected to the evolved GSM corenetwork used for UMTS, most of the remaining standardisation work related to GSMevolution was transferred in the year 2000 from ETSI to 3GPP as well

A further addition to 3GPP took place in late 1999, namely that of a Chinese partnerwith its own air-interface technology submitted earlier to ITU as an IMT-2000 candidate,namely a UTRA TDD derivative This is not included in the first release of the UMTSstandards (namely release 1999), but added as another UTRA mode later on

2.3.5 The Universal Mobile Telecommunications System (UMTS)

A main architectural design principle of the universal mobile telecommunications system(UMTS) is the split of the fixed UMTS infrastructure into core network (CN) and accessnetwork (AN) domains An additional design principle is the logical split of the globalarchitecture into a so-called ‘access stratum’, containing equipment and functionalityspecific to the access technique (e.g radio-related functionality), and ‘non-access strata’,

as shown in Figure 2.2 The access stratum includes protocols between the mobile terminaland the access network, and between the access network and the serving core network.While the former support the transfer of detailed radio-related information, the latterare independent of the specific radio structure of the access network This is important, itmeans that the CN should not be affected by the choice of radio transmission technologies

in the access network, such that new types of access networks can be defined as and whenrequired and attached to the existing core network

The only suitable access network type defined in release 1999 specifications is theUMTS or Universal Terrestrial Radio Access Network (UTRAN), consisting of a set ofradio network subsystems These in turn are composed of a Radio Network Controller(RNC) and a number of base stations, the latter (for lack of agreement) somewhat oddly

termed Node B in UMTS The radio technologies featured by UTRAN release 1999 are

the UTRA TDD and the UTRA FDD mode The CN consists of a ‘circuit-switcheddomain’ or CS-domain, which is composed of Mobile services Switching Centres (MSC)very similar to those already used in GSM, and a ‘packet-switched’ or PS-domain, which

Non-access strata

Access stratum

Mobile station

Uu (radio) interface Iu interface

Figure 2.2 Basic logical UMTS architecture

is an evolution of the GPRS core network Accordingly, there are two variants of the Iuinterface between AN and CN shown in Figure 2.2, namely Iu-CS and Iu-PS

The fundamental UMTS service principle is to standardise service capabilities ratherthan the services themselves, which helps achieving flexibility in service provision With

an appropriate set of service capabilities, users, service providers, and network operatorsshould be in a position to define services themselves according to their specific needs [87].WCDMA was investigated as one of the two final multiple access modes in the ACTSFRAMES project (alongside TD/CDMA) WCDMA built also on concepts evaluatedduring the RACE Codit project [12] Furthermore, during the concept development phasetaking place in ETSI SMG2, features of the Japanese WCDMA proposal [85] were alsoadopted This led to the final concept adopted as a basis for the UTRA FDD mode [88],which was further enhanced during detailed standardisation work in 3GPP, for instancewith features of some of the American WCDMA submissions to ITU

UTRA FDD makes use of direct-sequence wideband CDMA operating at a basic carrierspacing of 5 MHz (which can be reduced down to 4.4 MHz, if required) In order toprovide high data-rate services in an efficient manner (e.g with some degree of trunking

or multiplexing efficiency) and to benefit as much as possible from frequency diversity,the carrier bandwidth should be as wide as possible, preferably even wider than 5 MHz

On the other hand, the spectrum situation did not allow for more, in fact, even 5 MHzcarriers will limit deployment flexibility considerably, as will be discussed below.Initially, a chip-rate of 4.096 Mchip/s was envisaged, and 16 time-slots were supposed

to fit into a 10 ms frame Harmonisation efforts with cdma2000 lead to a reduced chip-rate

of 3.84 Mchip/s and the elimination of one slot per frame, such that now 15 slots of 0.666 ms length form a frame of 10 ms On dedicated channels, a link betweenmobile station and base station is continuously maintained at some minimal bit-rate Thepurpose of the time-slots is therefore not to provide a TDMA feature, but to structure thetransmitted data, for instance to specify the periodicity of power control commands andother overhead related to the physical layer In UTRA TDD, by contrast, these are ‘proper’time-slots (in the TDMA sense) A good and readily available overview of WCDMA isprovided in Reference [89]

time-The TD/CDMA proposal [90] adopted as a basis for the UTRA TDD mode is to alarge extent based on a scheme proposed in Reference [13], and was further investigated

as mode 1 of the two final multiple access modes considered in the European ACTS

FRAMES research project [91] (mode 2 being WCDMA) It employs a combination ofCDMA and TDMA as basic multiple access methods The total number of time-slotsper frame can be shared flexibly (alas, with certain deployment constraints) betweenuplink and downlink, thereby enabling asymmetric use of the total spectrum resources.Initially, the suggestion was to apply the GSM time-slot/frame structure and to adopt

an integer multiple of the 200 kHz used in GSM for carrier spacing This would (atleast in theory) have allowed TD/CDMA to be deployed on a per-time-slot basis in anexisting GSM system, provided that six or eight carriers are allocated to a single GSMcell Such an approach could be viewed as an evolutionary approach similar to thosedescribed in Subsection 2.3.2, but just a little bit more radical However, having beenadopted in conjunction with WCDMA as the other UTRA mode, the main worry was thenharmonisation between these two modes As a result, the TD/CDMA scheme adoptedsome fundamental design parameters such as chip-rate, carrier spacing and slot/framestructure from the WCDMA scheme [84] First, the harmonised parameter set was based

on a chip-rate of 4.096 Mchip/s, but following harmonisation efforts with cdma2000,also TD/CDMA ended up with a chip-rate of 3.84 Mchip/s and 15 time-slots per 10 msframe

2.3.6 The Spectrum Situation for UMTS

Before being able to discuss whether we can expect UMTS to meet the requirementslisted earlier, we need to examine the UMTS spectrum situation, as this has a quiteconsiderable impact on this discussion The initial spectrum identified for 3G systems issituated around 2 GHz It consists of 2× 60 MHz of paired spectrum set aside for 3G

in Europe, Japan, South Korea and other parts of the world, but not in the US, where apart of this spectrum was auctioned by the federal government for PCS This paired bandwith two equally sized portions for the uplink and the downlink is for instance suitablefor UTRA FDD In addition to that, in Europe, 20 MHz of unpaired band was set asidefor public operation of 3G systems, and another up to 15 MHz of unpaired band for

‘license-exempt’ or ‘self co-ordinated’ use, e.g for residential telephony, the latter not ofinterest in the following Unpaired band is not suitable for UTRA FDD, but can be used

by the UTRA TDD mode

In most countries having licensed this 3G spectrum, the available bandwidth is carved upbetween four to six licensees (i.e operators), so that with few exceptions, the best an indi-vidual operator could get was 2× 15 MHz of paired band plus 5 MHz of unpaired band

In the UK, for instance, there are five licensees, resulting in differently composed trum packages, as shown in Figure 2.3 Three of these packages consist of 2× 10 MHzplus 5 MHz, one of 2× 15 MHz and the biggest one, set aside for a new entrant (i.e anoperator not owning a 2G network in the UK), of 2× 15 MHz plus 5 MHz of spectrum.Similarly, in Germany, where six licensees share the spectrum, each is assigned only

spec-2× 10 MHz of paired band, and four of them an additional 5 MHz of unpaired band.This is rather miserable, particularly when taking into account that some GSM 1800 oper-ators hold licenses for 2× 30 MHz of paired band In the year 2000, more spectrum wasidentified for 3G, which will eventually be made available to operators, although it is up

to individual countries to either improve the spectrum situation of existing 3G licenseholders or assign this spectrum to new licensees