Completing the convergence puzzle: a survey and a roadmap

IEEE Wireless Communications • June 200986 1536-1284/09/$25.00 © 2009 IEEE

AC C E P T E D F R O M OP E N CALL

DJAMAL-EDDINE MEDDOUR, USMAN JAVAID, AND NICOLAS BIHANNIC, ORANGE LABSTINKU RASHEED, CREATE-NET RESEARCH CENTER

RAOUF BOUTABA, WATERLOO UNIVERSITY

COMPLETING THE CONVERGENCE PUZZLE:A SURVEY AND A ROADMAP

INTRODUCTION

The proliferation of fixed and mobile accesstechnologies and communication devices havehighly enlarged the choice for network operatorsand service providers to offer a wide variety ofservices. This context is illustrated with accessnetworks such as the Universal Mobile Telecom-munications System (UMTS), WiMAX, WiFi,digital video broadcasting (DVB), and heteroge-neous mobile terminals (e.g., smartphone, PDA),

supporting enhanced integration of applications.Currently, with the massive interest in thedeployment of these mobile broadband wirelesstechnologies, the integration and convergence ofthese networks and technologies is becoming notonly possible, but also a necessity to provide sev-eral value-added services to consumers at afford-able prices. In the past few months, convergence(eventually with seamless mobility) has morethan ever been at the center of fixed and mobileoperators’ attention throughout the world. Froma technical perspective, it is considered to be thenext big step in the evolution of telecommunica-tion networks. For instance, wireless technolo-gies such as WiFi offer high data rates at lowcost but do not guarantee seamless coverage,especially with high mobility. Bluetooth technol-ogy supports low data rates compared to thehotspot technologies, but saves in the powerconsumption required for wireless access. Incontrast, cellular networks such as Global Sys-tem for Mobile Communications/General PacketRadio Service (GSM/GPRS) and UMTS providewide area coverage and support high mobility ata higher cost (when assessing the bandwidthcost). From marketing perspectives, thanks to itstechnical edge, convergence allows the operatorto enlarge its portfolio and attract more cus-tomers with new and aggressive offers as thesame services are likely to be provided over allexisting networks (fixed and mobile).

In such a diverse environment, the concept ofbeing always connected becomes always bestconnected [1, 2]. This refers to being connectedin the best possible way by exploiting the hetero-geneity offered by the access networks in orderto experience a large variety of network services,particularly in the event of user mobility (access-ing services using various terminals). Moreover,end-user devices are increasingly equipped withmultiple interfaces enabling access to differentwireless networks subject to network availability,device characteristics, and the applications used,all of which introduce the need for networkinteroperability in this heterogeneous environ-ment. Also, the tremendous growth of connectedwireless devices has augmented the endless com-

BTS-MIH

AP2-MIH

AP1-MIH

Accein

ABSTRACTConvergence has more than ever been a cen-

tral issue for fixed and mobile operators through-out the world and is considered to be the nextbig step in the evolution of telecommunicationnetworks. Convergence opens new marketopportunities and competition among networkoperators and above all offers enhanced userexperience. Multimode handsets and the prolif-eration of terminals and access technologies aregenerating increasing demands for solutions thatenable convergence, seamless handover, andtransparent service delivery across heteroge-neous access networks. Different strategies areavailable for operators, depending on the ser-vices they intend to deliver to their customers,from basic commercial convergence limited tounified billing for Fixed/Mobile/Internet up toin-dept network convergence covering new appli-cations and services.

This article surveys different technologieswhich offer seamless handover and convergedaccess to mobile voice, video, and data services.It provides first the different network partsinvolved in defining the operator global conver-gence strategy and then surveys different tech-nologies which achieve this step-by-stepconvergence. We present the main features ofthese technologies and discuss their limitationsand potentials to enable convergence in hetero-geneous networks. We also provide a personalstance as to the emergence of these technologiesand our vision towards the long term convergedtelecommunication networks.

The authors surveydifferent technologieswhich offer seamlesshandover and converged access tomobile voice, video,and data services.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 86

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 2009 87

petition for scarce wireless resources and hassignificantly exposed the challenges for heteroge-neous network resource management. It there-fore consists, for the operator, of selecting themost efficient network for ongoing user sessionswith regard to network cost efficiency and userexperience efficiency.

In this increasingly heterogeneous networkingarchitecture, integration and convergence can beachieved in different ways by integrating tech-nologies at different levels and ensuring efficientroaming solutions [3]. From basic commercialconvergence (unified billing for fixed/mobile/Internet) or basic service convergence (a serviceprovider allowing service delivery across differ-ent access networks without the need to supportmobility management), to network convergence(transparent service delivery when changingaccess, with service continuity and seamlessmobility), different strategies can be adopted byoperators, depending on the services they wantto deliver to their customers. In this article weprovide a comprehensive survey of each of thesetechnologies highlighting their design goals,architectures, and protocols. A comparison studyillustrating their differences, advantages, andlimitations is also presented. We conclude bypresenting our personal views on the evolutionof the discussed technologies toward true andubiquitous convergence of these heterogeneousnetworks (also named seamless convergence inthis article).

SEAMLESS CONVERGENCE: A LONG-TERM VISION

The beyond third generation (B3G) network is amultitier hierarchical system that supports IP-based mobile multiparty multimedia servicesover heterogeneous wired and wireless networks.Convergence is the key word used to refer tothose networks that offer unified solutions whereheterogeneous components are seamlessly inte-grated and interoperate to provide service con-formity and assurance to customers. While theresearch community is interested in the integra-tion and interoperation of B3G heterogeneousnetworks at various scales, in this section, wefocus our discussion on convergence aspectswhich are relevant in a service provider network.

Seamless network architectures can be rough-ly classified as those ensuring either the mobilityof users accessing services with various termi-nals, or the mobility of the terminals accessingservices across access networks. An example ofmobility for the first class (mobility of usersaccessing services with various terminals) is toallow a user to transfer, when at home, part orall of a set of media in an ongoing session towardanother terminal with more appropriate capabili-ties. Each type of convergence is coupled withthe services that can be supported with differentimpacts on devices and network infrastructures.For instance, seamless mobility is expected forvoice services, whereas some resuming or book-marking features without seamless mobility canbe satisfying for streaming services. The hetero-geneity convergence of concern to the operatoris illustrated hereafter through an enforcement

of convergence at four network levels (the homenetwork, access network, core network, andapplication server levels). The long-term conver-gence for an integrated operator (offering bothfixed and mobile services) will be built by inte-grating these four convergence streams within itsglobal strategy on network and service evolution.

HOME NETWORK CONVERGENCEHome network convergence can be defined asthe capability to break the silo approach where aterminal is dedicated to the use of a given ser-vice. Home network convergence allows the fol-lowing benefits:• The user can access different types of services

from the same terminal. An example is theability to handle a call with a handset and beable to display on this same handset contentretrieved from another device located in thehome network.

• A service is available on more than one hand-set. For instance, the video-on-demand (VoD)service is not only displayed on the user’s tele-vision screen but can also be viewed on a PCor mobile handset.

• Diverse communication technologies areexpected within the home network sphere.Most prominent are WiFi, power line commu-nication (PLC), and Gigabit Ethernet (GbE).Hence, the home network is a convergencearena where the devices are able to communi-cate with each other and also with the serviceplatforms. As a complement to these tech-nologies, we must mention middleware oppor-tunities such as universal plug and play(UPnP) and its associated certification alliance(Digital Living Network Alliance [DLNA])that favors handset capability to interoperate,thereby offering simplicity in the user experi-ence (automatic device discovery, automaticdevice interoperability).Home network convergence is mainly built

around the introduction of a home gateway. Thisequipment provides IP connectivity to devicesfor local data exchange and interconnection toservice platforms. Some application capabilitiescan be added to routing capabilities, such as theintroduction of a Session Initiation Protocol(SIP) feature that complements services handledin the core network by an IP multimedia subsys-tem (IMS) infrastructure (such new capabilitiesare management of local device registration orsimultaneous service notification on devices).

ACCESS NETWORK CONVERGENCEAccess network convergence can either beachieved with convergence focusing on the trans-port layer and/or convergence involving the ser-vice control layer on the access network (theservice control layer corresponds to networkmechanisms in charge of delivering the servicelike call establishment for conversational ser-vices).

The first one is mainly driven by the reduc-tion in operational expenditure (OPEX) costs.An example is to aggregate mobile access nodesinto a backhaul network shared with a fixedaccess network, or more generally the use of ashared infrastructure for heterogeneous accesssolutions.

Home network convergence is

mainly built aroundthe introduction of a

Home Gateway. This equipment

provides IP connectivity to

devices for local dataexchange and

interconnection toservice platforms.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 87

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 200988

Convergence at the service control layerallows access to the same service irrespective ofthe access network infrastructure.

CORE NETWORK CONVERGENCECore network convergence addresses conver-gence in the core network and is typically associ-ated with the definition of a common frameworkable to handle any service invocations irrespec-tive of the access network. The most relevantexample is the specification of the IMS bothspecified in the 3G Partnership Program (3GPP),and endorsed by Telecommunications and Inter-net Converged Services and Protocols forAdvanced Networks (TISPAN) of the EuropeanTelecommunications Standards Institute (ETSI)and the Next Generation Networks Global Stan-dard Initiative (NGN GSI) of the InternationalTelecommunications Union (ITU) [4]. TISPANand NGN GSI specify access to the IMS plat-form installed in the core network from the fixedbroadband access network, whereas 3GPP dealswith access to IMS from a mobile access net-work.

The IMS is a core network infrastructure tocontrol user sessions for the following services:• Conversational services with multimedia com-

ponents such as voice and video• Real-time data-oriented services such as

instant messaging and presence• Audio-visual services, in the scope of specifica-

tions at ETSI (with TISPAN) and ITU (withIPTV GSI)Some of the benefits expected by the opera-

tor deploying an IMS infrastructure are:• To use a common functional infrastructure for

services control while being as much accessnetwork agnostic as possible. A strong advan-tage in time to market (TTM) performance isalso expected with efficient integration of ser-vices once the IMS infrastructure is deployed.

• Enhanced mechanisms to reserve bandwidthon the user data path as negotiated duringsession establishment between end-user hand-sets. This allows the operator to better controlresources, especially significant in the mobiledomain for packet switching (PS) services.

• Service triggering toward application servers inaccordance with user service profiles.

• Solution for public switched telephone net-work (PSTN) renewal and expectations ofOPEX/capital expenditure (CAPEX) reduc-tions.The implementation of an IMS infrastructure

has impacts not limited to the core network butextended to the whole operator network: newcapabilities in the terminal to support the SIPprofile, updates of mobile gateways (like GGSN)to support a new interface for resource control,new application servers on an IMS to either han-dle SIP-based service logic or interwork withlegacy service platforms (e.g., CAMEL-based forsome mobile services), and finally on informa-tion systems (IS) for service provisioning.

APPLICATION SERVER LEVEL CONVERGENCEConvergence at the service platform level (alsocalled the application server level) can take twodirections. First, it can be coupled with the intro-duction of a common control infrastructure such

as IMS in order to address heterogeneous net-works (fixed and/or mobile) from the same ser-vice platform with the capability to offerdifferentiated quality of service (QoS) to endusers. This case can largely meet carrier gradestrategies. The second direction is to consider itas a standalone convergence, as discussed below,and this strategy is supported by Web players.

Standalone convergence allows service pro-viders to benefit from generalized IP connectivi-ty of terminals (fixed and mobile) to offer theirservices. This model is based on the Internetmodel with best effort QoS, unlike IMS, whichallows the operator to set policy on QoS andcharge for services accordingly. Note that thisconvergence proposed by the service providers(who do not generally own the network) can alsobe implemented in a decentralized way, alsoreferred to as peer to peer (P2P). In the latterthe user accesses his/her services (voice, IM, orcontent sharing) in a simple way: this decentral-ized application convergence only requires rely-ing on the IP connectivity of the terminal andrunning the application on the terminal. Areduced number of centralized nodes need to beoperated like an AAA server to control useraccess to paid services, interconnection of gate-ways to extend the service legibility to non-IPenvironments like the PSTN.

This long-term convergence built by the inte-grated operator (offering fixed and mobile ser-vices) is the integration of these four streamswithin its global strategy. Additional technolo-gies fasten this seamless convergence betweenfixed and mobile environments. Such solutionsare pointed out in the following section.

EXISTING SOLUTIONS TOWARDSEAMLESS CONVERGENCE

The past few years have witnessed tremendousgrowth in the number of wireless hotspots basedon WiFi. Today, a large number of users accessthe Internet through wireless local area networks(WLANs) in a variety of places and environ-ments, including their homes, offices, and publicplaces. WLANs have emerged as a promisingnetworking platform that offers high data ratesto mobile users at low network deployment cost.Anyone can simply plug a WLAN access pointto the Internet and make it available to wirelessusers to enjoy connectivity. Normally WLAN-based hotspots are deployed in areas with highuser density and high bandwidth demands (e.g.,in a town center). In contrast, base stations(BSs) in UMTS offer larger cells, and with inter-BS links, the UMTS provides nearly ubiquitousworldwide coverage. The interconnection ofUMTS and WLAN networks provides an eco-nomical solution to offload some traffic fromlicensed to unlicensed spectrum technologies.Moreover, UMTS/WLAN cooperation providesan interesting blend, where the user can leveragethe global coverage of UMTS and high data ratesupport of WLAN. The user can directly benefitfrom this cooperation through offers where theuser may gain some data exchanges from Webportals for free when under WiFi coverage.

To this end, 3GPP Unlicensed Mobile Access

This Long Term convergence built bythe integrated operator is the integration of thesefour streams withinits global strategy.Additional technologies fastenthis seamless convergencebetween fixed and mobile environments.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 88

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 2009 89

(UMA) and 3GPP Interworking-WLAN (I-WLAN) technologies offer a first step of conver-gence managed at the access network level.

Meanwhile, the integration of heterogeneousnetworks is supported in the core network byarchitecture like IMS [4] for user session controland potentially enriched with service platforms(e.g., Voice Call Continuity [VCC] or Multime-dia Session Continuity [MMSC]) for internet-work mobility.

On the other hand, the media-independenthandover (MIH) entity of IEEE 802.21 [5] is aflexible framework that does not intend to pro-vide a standalone solution for fixed-mobile con-vergence (FMC), but rather assists theintertechnology handover decision and interop-erability in coordination with other mechanisms.

In this section we discuss these complemen-tary technologies by highlighting their respectivedesign goals, architectures, and protocols.

UMAUMA technology is designed to enable FMC inan access network. It is currently endorsed bythe 3GPP [6] under the name generic aaccessnetwork (GAN) (we use UMA and GAN inter-changeably in the rest of this article).

A major feature of GAN is to offer call conti-nuity from a GAN-capable terminal between alocal area network (ultra wideband [UWB] or802.11) terminating at the fixed access and theGSM infrastructure. Data services are also sup-ported, but are limited in throughput since inter-connection to the packet-switched core network(PSCN) is performed using the 3GPP-definedGb interface. A recent evolution of GAN enrich-es user experience for data services with theenhanced GAN (EGAN) specifications (TR43.902). The Gb interface is updated by the Gninterface to allow the enhanced GAN controller(GANC) entity to interconnect directly with theGPRS gateway service node (GGSN) entity.This evolution aims to reduce latency and over-head for PS services. No change on the circuit-switched domain is required. More precisely,UMA is today an available technology alreadydeployed by certain operators like Orange withits Unik1 offer.

In the GAN architecture an IPSec tunnel isestablished on the up interface between theGAN terminal and the GANC. This flow tunnel-ing is a strong security requirement that allowsconveying both signaling and user data flows(GSM/GPRS signaling and user plane flows are

piggybacked into GAN-specific protocols andthe IPSec tunnel) over an access network(named the generic IP access network) that isnot supposed to be under the control of themobile operator. The newly defined GANC enti-ty reuses the already 3GPP-defined Gb and Ainterfaces to interconnect to the PSCN and cir-cuit-switched core network, respectively. Notethat the administration, authorization, andaccounting (AAA) server is used to authenticatethe GAN terminal when it sets up the securetunnel. Figure 1 presents the architecture ofGAN and its positioning with respect to theGSM/GPRS architecture.

I-WLAN3GPP is developing interworking solutionsbetween 3G and WLAN networks under theauspices of I-WLAN aiming to realize UMTS/WLAN integration [7, 8]. (I-WLAN is a 3GPPstandard that intends to define an interworkingarchitecture between a WLAN access networkand the 3GPP core network.)

I-WLAN Architecture — In the 3GPP Release 6specifications that aim at providing access tomobile operator services from a WLAN accessnetwork (AN), I-WLAN introduces three maincomponents to achieve 3G/WLAN convergence :a wireless access gateway (WAG), a packet datagateway (PDG), and an AAA server, as shownin Fig. 2. The user equipment (UE) is typicallydual-mode-capable: under WLAN coverage, it iscapable of connecting to the WLAN AN usingWiFi (as an example of radio technology) beforeattachment to the I-WLAN infrastructure, andwhen outside WLAN coverage, it can connect tothe UMTS operator network. Data coming fromUE through fixed ANs (generic IP access net-work in Fig. 2) are aggregated at the WAG,which is further connected to the PDG. In theroaming case, the visited WAG is also able toroute packets toward the home domain of theoperator to which the user has subscribed. ThePDG in the I-WLAN architecture works as agateway toward either external packet data net-works (PDNs) or the operator service infra-structure, as shown in Fig. 2. The PDG alsointeracts with the AAA server to perform ser-vice-level AAA functions.

When entering into the coverage area of aWLAN AN, the UE triggers its attachment pro-cedure with the I-WLAN infrastructure; thus, anIPSec tunnel is established between the UE and

1 Orange Unik™(http://unik.orange.fr)

A major feature ofGAN is to offer call

continuity from aGAN capable

terminal between alocal area networkterminating at the

fixed access and theGSM infrastructure.

Data services arealso supported but

are limited inthroughput.

�� Figure 1. GAN architecture and functional components.

GGSNESC/RNC

Up

AP

GbA

SGSN

MSC

AAAGeneric IPaccess network

GANC

HLR

PSTN

OtherPLMN

Internet External PDN

BOUTABA LAYOUT 6/4/09 3:48 PM Page 89

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 200990

the PDG. Packet-switched (PS) domain signalingand user plane data are carried into this securetunnel over a Wu interface.

I-WLAN Protocols — The protocol stack betweenthe WLAN UE and the PDG is depicted in Fig.3. The protocol layers introduced in the I-WLANstack are:• Remote IP layer: The remote IP layer is used

by the WLAN UE to communicate with theexternal PDN. The PDG routes the remote IPpackets without modifying them.

• Tunneling layer: The tunneling layer consistsof a tunneling header (IPSec), which allowsend-to-end tunneling between WLAN UE anda PDG. It is used to encapsulate remote IPlayer packets. The tunneling header containsthe information which is further required bythe PDG and the UE to decrypt the IP pack-ets.

• Transport IP layer: The transport IP layer isused by the intermediate entities/networks andWLAN AN in order to transport the remoteIP layer packets encapsulated into the IPSectunnel.

I-WLAN Evolution — On the core network side, anew work item called System Architecture Evo-lution (SAE) was defined. In this evolved UMTSarchitecture it is expected that IP-based serviceswill be provided through various access tech-nologies. A mechanism to support seamlessmobility between heterogeneous access net-works is needed for future network evolution.

To this end, I-WLAN is included in the SAE toensure a smooth migration path from the R6 I-WLAN work to a generic multi-access solution.Thus, mobility is under study for I-WLAN inrelease 8 and is based on Mobile IP (MIP)unlike the legacy GPRS systems with GTP-based mobility in the core network (UMA alsouses this GTP mobility since interconnecting tothe GPRS core network). The mobile device isrequested to integrate a DSMIPv6 stack thatdialogs with a home agent entity (located in thecore network) that is in charge of routing userdata flows toward the ANk where the user isattached to.

MIH: IEEE 802.21The IEEE 802.21 Working Group (WG)defines MIH in order to offer seamless con-vergence across heterogeneous networks [5].The MIH defines a framework to supportinformation exchange that facilitates mobilitydecisions, as well as a set of functional com-ponents to execute those decisions. The MIHshields link-layer heterogeneity and providesa unified interface to upper-layer applica-tions in order to support transparent servicecontinuity. The handover scenarios consid-ered in 802.21 WG include wired as well aswireless technologies — the complete IEEE802 group of technologies and 3GPP/3GPP2AN standards.

The MIH framework provides methods andprocedures to gather useful information fromthe mobile terminal and network infrastructure

�� Figure 2. I-WLAN R6 architecture and functional components.

Gol

dcl

ient

s

WA.4 WA.3 WA.1 PB.1 PB.4 PB.4P-/S-CSCFA

DomainA

I-/S-CSCFBIOMA + PSA IOMB + PSB

DomainB

SUBSCRIBEfrom otherwatchers

PUBLISHfrom otherpresentities

(1) SUBSCRIBE

(13) OK

(15’) OK msgs

(15’) OK msgs

(2) 200 OK

(2) SUBSCRIBE (3) SUBSCRIBE

(6) 200 OK

(12’) OK

(12”) OK

(11’) Common NOTIFY(one per presently)

(11”) Batched NOTIFY(one per watcher)

(7) 200 OK

(4) SUBSCRIBE(5) 200 OK

(9) PUBLISH(10) OK

(13’) NOTIFYmsgs

(14) OK

(16’) OK

(13”) aggregatedNOTIFY

(16”) OK

(11) NOTIFY(12) NOTIFY

(14’) NOTIFY

(14’) NOTIFY

PrioritizedNOTIFY schedule

Watchers Authorization andNOTIFY construction

Watcher Authorization, NOTIFY aggregation, NOTIFY construction

Silv

ercl

ient

sC

oppe

rcl

ient

s

When entering intothe coverage area ofWLAN AN, the UEtriggers its attachment procedure with the I-WLAN infrastructureand thus an IP Security (IPSec)tunnel is establishedbetween the UE and the PDG.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 90

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 2009 91

in order to facilitate handover between hetero-geneous access networks. MIH provides networkdiscovery procedures that help the mobile termi-nal determine available networks in its currentneighborhood. A mobile terminal selects themost appropriate network with the help of thegathered information such as link type and qual-ity, application class, network policy, user pro-file, and power constraints.

MIH Architecture — MIH function (MIHF) lies atthe heart of the MIH architecture and providesan intermediary or a unified interface betweenthe lower-layer heterogeneous ANs and higher-layer components.

MIH provides generic access-technology-independent primitives called service accesspoints (SAPs). SAPs are application program-ming interfaces (APIs) through which the MIHFcan communicate with the upper and lower layerentities.

The MIHF facilitates three services: media-independent event service (MIES), media-inde-pendent command service (MICS), andmedia-independent information service (MIIS).As highlighted in Fig. 4, those services areresponsible for:• Signaling state changes at lower layers• Control by higher layers

• Provision of information regarding the neigh-boring networks and their capabilities, respec-tively

MIH Functional Components — In the following wedescribe the MIH functional components ingreater detail.

MIES — It notifies the upper layers of the occur-rence of lower-layer events (triggers) in order tooptimize handover performance. The events canbe local (i.e., within a mobile client) or remote(i.e., sent by the network component). The eventmodel follows the notification/subscription prin-ciple. Since events are advisory and not manda-tory, registration to a specific event is needed foran entity to be notified whenever such an eventoccurs.

MIES events may be broadly classified intotwo categories: link events and MIH events.Both link and MIH events typically flow from alower layer to a higher layer. Link events aredefined as events originating from event sourceentities below the MIH function and typicallyterminate at the MIH function. Within theMIHF, link events may be further propagated,with or without additional processing, to upperlayer entities that have registered for the specificevent. Events that are propagated by the MIH to

�� Figure 3. I-WLAN R6 protocol stack.

Remote IP

Tunnelinglayer

TransportIP

L2/L1

Tunnelinglayer

Remote IP

TransportIP

L2/L1

L2/L1Transport

IP

L2/L1

TransportIP

L2/L1

UE WLAN AN

TransportIP

L2/L1

TransportIP

L2/L1

WAG PDG

Analogous to MIES,MICS can also bedivided into two

categories i.e., MIH commands and

link commands. Both of these typesof command follow

the same principle asdescribed for MIES.

�� Figure 4. MIH architecture and functional components.

MIH events MIH commandsInformation

service

Upper layers (L3 and above)

Media independent handover function (MIHF)

SIP MIPv4 MIPv6 HIP L3MP

Lower layers (L2 and below)

802.3 802.11 802.16 3GPPWCDMA

3GPP2CDMA2000

Link events Link commandsInformation

service

BOUTABA LAYOUT 6/4/09 3:48 PM Page 91

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

the upper layers are defined as MIH events.Some of the common events include LinkGoing Down, L2 Handover Imminent, andLink Parameters Change. Upon receptionof a certain event, the upper layer makes use ofthe command service to react to the change inthe network state.

MICS — MICS refers to commands sent fromhigher layers to lower layers in the MIH frame-work. MIH commands are used to subscribe tocertain information from the lower layers suchas gathering information about the status of con-nected links, as well as execute higher-layermobility and connectivity decisions at the lowerlayers. Similar to MIH events, commands canalso be local or remote.

Analogous to MIES, MICS can also be divid-ed into two categories: MIH commands and linkcommands. Both of these command types followthe same principle described for MIES. Some ofthe common commands include MIH Poll ,MIH Scan , MIH Configure , and MIHSwitch.

MIIS — It is used by a mobile node or networkentity to discover and obtain information aboutneighboring networks. The purpose of the infor-mation service is to acquire a global view of theheterogeneous networks to facilitate seamlesshandover across those networks. For instance,when a mobile node is about to move out of thecoverage of the current network, it queries thenetwork (MIIS) about the available neighboringnetworks in order to optimize the handover pro-cess. MIIS provides access to both static anddynamic information. The static information mayinclude names and providers of the neighboringnetworks. Examples of dynamic informationinclude channel information, MAC address, andsecurity information. MIIS stores the informa-tion in a standardized format such as ASN.1 orXML.

LIMITATIONS AND POTENTIALS

The objective of this section is to provide a com-prehensive comparison between convergencetechniques by illustrating their advantages andlimitations.

UMA presents a relevant solution in the

short and mid terms for voice service over a cir-cuit-switched core network with GSM-like per-formance for voice handover. However, it isworth noting that UMA has many limitations,notably the significant overhead due to the com-plexity of the protocol stack at the terminal(with upper layers of GSM/GPRS protocol stackover new UMA protocol stack for circuit-switched [CS] and PS domains, respectively).UMA also suffers from poor performance inpacket mode, and evolution toward EGAN(enhanced GAN) is considered to overcome thisweakness. Some mechanisms for QoS handlinglike DSCP marking can be implemented by oper-ators to manage QoS in the home network andANs (up to the mobile core network domain),but the use of IPsec tunnel sets constraints onmarking rules, especially when GAN is a bearerfor multiservices with different QoS needs.Finally, UMA also requires the use of a new ter-minal for the end user to access FMC servicesbuilt over this technology.

I-WLAN is an alternative solution to UMAaiming at integration of UMTS core networksand WLAN. It is designed for packet-based ser-vices and is targeting a more long term evolutionthan UMA. Indeed, enhancements with 3GPPRelease 8 (R8) specifications on I-WLAN areintended to support mobility and have beenreleased with the first stable SAE specification.

This mobility based on MIP can be seen as afirst step in mobility management. Indeed, itallows data transfer either between I-WLANaccesses or from I-WLAN access toward a cellu-lar network (and conversely) for services likeWeb browsing or content streaming. For servicescontrolled by IMS (like voice over IP in I-WLAN), more advanced use cases for mobilitywill be expected and be delivered with additionalfeatures available from IMS as explained here-after.

Indeed, a weakness of I-WLAN is the lack ofsupport for circuit-based services. Consequently,the operator is required to couple I-WLANdeployment with an IMS infrastructure with aVCC enabler to support call handover withlegacy CS networks (typically GSM) and anMMSC enabler to manage session transfer formultimedia services. 3GPP has gathered in R8framework these two former concepts within theIMS service continuity (ISC) concept [9]. Some

IEEE Wireless Communications • June 200992

�� Figure 5. MIH potential integration with the current network architecture.

Mobilitymanager

Converged corenetwork

Correspondentnode network

BTS-MIH

AP2-MIH

AP1-MIH

Access networkinterface

I-WLAN is an alternative solutionto UMA aiming atintegration of UMTScore networks andWLAN. It is designedfor packet-based services and is targeting a morelong term evolutionthan UMA.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 92

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 2009 93

issues are still outstanding, like the dynamicprovisioning of operator policies within themobile handset.

Like GAN, I-WLAN suffers from certain lim-itations on QoS handling due to an always acti-vated IPSec tunnel that hides the nature of eachencrypted data packet along with the fixed accessnetwork. Some additional mechanisms exist tocounter this limitation by propagating the DSCPvalue (used for flow prioritization by IP routers)out of encrypted headers of IPSec packets. Butthis always activated IPSec tunnel feature is notoptimized, especially when the same operatorcontrols the fixed access network and the mobilenetwork, since in this case encrypting flows isnot justified.

Unlike UMA, I-WLAN appears less con-straining on handset implementation since somevendors propose to download I-WLAN clientsoftware on their smart phones to become I-WLAN-capable. However, this update may suf-fer from a lack of integration that is notconvenient for high end-user experience.

MIH is an interesting approach for interop-erability with respect to long-term networkarchitecture evolution. It provides mechanismsfor improving handover decisions and mobilitypolicy (e.g., what kind of service the user isallowed, to which AN the user is allowed toattach, what QoS is expected) for services sup-ported over both PS and CS networks. Mobilityhandling can be triggered and enforced byeither the network or the terminal. The majorpoint to highlight here is that the mobility pro-

cedure remains under the control of the net-work operator, whereas the interaction with theservice platform is considered out of the scopeof the standard.

MIH, as a generalized framework for mobili-ty, can benefit the mobility policy in a 3GPParchitecture context. A potential architecture forintegration of MIH with the current networkarchitecture is shown in Fig. 5. Indeed, someMIH functions can provide complementaryinformation to that sent by the mobile terminalfor handling user mobility. In particular, the3GPP standardization committee has put forthseveral requirements related to non-3GPP mobil-ity support, and the committee recognizes thatthe network-based mobility scheme needs someform of media independent mobility signaling tocoordinate the access change between networks.From this perspective, some MIH functions,such as those related to the generation of MIHevents, can be inserted into some access pointscomposing the convergence network of the oper-ator. This location of MIH functions within thenetwork acts as a complement to integratingMIH functions within devices. For example, thehome gateway (as defined previously) can pro-vide the mobility manager located in the con-verged core network complementary informationnot available at the terminal side and necessaryfor optimized mobility management. The avail-able bandwidth on the radio interface used bythe considered mobile is an example of suchcomplementary information. The evolution ofthe home network architecture is indeed raising

�� Table 1. Comparison of technologies for seamless convergence.

GAN/UMA I-WLAN 802.21/MIH

Standardization body 3GPP 3GPP IEEE

Mobilityperformance

++: Applied to circuit-switched and packet-switched services++: Integration effortreduced (in-built mobilitysolution)

+: Under study (MIP-based) andapplicable to packet-switched servicesonly–: Need for integration effort with IMSinfrastructure for enhanced mobilityuse cases (referring to ISC [9])

++: Under study and applicable topacket-switched services and circuit-switched. Interworking with broadcastnetworks is under investigation.

Security

++: EAP-based userauthentication (EAP SIM,EAP AKA), use of IPSec ontransport plane

++: EAP-based user authentication(EAP SIM, EAP AKA), use of IPSecon transport plane

+: EAP protocol between the mobilenode and its point of attachment (e.g.,MIH-AP in Fig. 5), and AAA protocol tothe authentication server.

Network complexityfor deployment

++: Network impactsreduced: limited numberof new nodes, legacyprotocols (GSM,GPRS)largely reused

++: Network impacts reduced:limited number of new nodes

+: The impact on the network varieswith regards to the approach used (ter-minal-oriented, network-oriented, orhybrid). Nevertheless, the impact on thenetwork is limited even in the network-centric approach.

Handset impact – –: New handsetrequired

–: Software update possible for openOS but potential issue on quality ofintegration

–: Software update required at differentnetworking stacks (3GPP, 802.x, …) inboth terminal-centric and hybridapproaches. Limited impact in thenetwork-centric deployment.

Billing +: Supported +: Supported – –: Out of scope

Legend: ++: Strong advantage; +: Advantage; –: Drawback; – –: Strong drawback

BOUTABA LAYOUT 6/4/09 3:48 PM Page 93

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 200994

new requirements, and a global view of the stateof resources in the home network is needed forbetter mobility management.

A number of issues are still to be addressedin order to integrate MIH capabilities with exist-ing 3GPP mechanisms.

The first issue concerns the methods for con-veying reports of MIH events from the accesspoints up to the mobility manager. In the con-text of previous 3GPP technologies (GAN and I-WLAN), strong security constraints are set tointerconnect with the mobile infrastructure. Twopossible approaches are considered: accessthrough the secure gateway (SEGW) or a directinterconnection with the mobility manager. Thefirst approach is similar to the security require-ment set for a terminal with an IPSec tunnelestablishment coupled with terminal authentica-tion. The second approach is more appropriatein a trusted environment involving the mobileoperator and the AN operator.

The second issue concerns the correlation ofthe reports of MIH events with reports generat-ed by the mobile terminal. This issue is associat-ed with the location management in order tocorrelate MIH event reports from a given accesspoint with the reports of terminals attached tothis same access point.

A third issue is related to the actor modeland associated regulation statements. The actormodel stands for identifying all operators thatparticipate in the service delivery. It encompass-es, for instance, the access network provider, ser-vice control provider, and content provider inthe case of audiovisual services. An example ofregulation constraints is given hereafter.

Today, WiFi access points in home networksare generally integrated in the home gateway,which is the equipment under the control ofoperator. Moreover, the WiFi access point ispresently transparent regarding mobility policyas it is not involved in triggering events formobility management. A potential regulatory

issue can be raised when allowing interconnec-tion of fixed operator controlled equipment (thehome gateway) with the mobile operator infra-structure to convey MIH information in order toimprove mobility performance. In this lattercase, these MIH-based enhanced mechanismsmust be reproducible to third parties that wishto deliver mobility solutions. These third partiesmust be able to benefit from these same MIH-based inputs to enforce mobility decisions whentheir customers are attached to this WiFi accesspoint.

Table 1 provides a comparison of the variousconvergence technologies discussed in this arti-cle.

A possible strategy for a mobile operator todefine its mobility architecture can be:•The assumption made on the trigger for I-WLANdeployment as technology renewing UMA/GANis the capability of I-WLAN to support usermobility. This requirement for mobility is moreconstraining than with UMA/GAN, especiallyfor voice services since it has to be correlatedwith the introduction of IMS to control VoIPand multimedia user sessions. For this purposeQoS handling for VoIP services over mobile net-works remains a major challenge today.Note that I-WLAN rollout can be hastened byoperators when no mobility feature is required.•The mobility between the GSM and I-WLANsystems will be performed first with VCC forvoice services only, and then be enriched withMMSC for enhanced services coupling voice anddata. An enhanced use case for a user with anongoing multimedia session with voice and videocomponents on his/her handset and enteringhis/her home network is: the voice componentcan be maintained on his handset, whereas thevideo component is split and displayed on theTV screen.

Figure 6 presents a possible timeframe forthe adoption of FMC solutions by network oper-ators. From the figure, MIP mobility is suited for

�� Figure 6. Timeframe proposal for fixed/mobile convergence solutions deployment.

UMA/GAN for voice and data services

UMA/GAN

Availabletechnology

Possibletechnologyrollout

Enhanced GAN

I-WLAN withoutmobility

I-WLAN with MIP-based mobility

MIH standard

Time frame

Enhanced 3GPPmobility featureswith VCC, MMSC

First I-WLAN deployment with basicmobility features

Enhanced mobility coupling differenttechnologies from 3GPP or IEEE (longterm vision)

Seamless convergence of heterogeneousaccess networks isessential in today’stelecommunicationsystems. Accordingly,operators can provide telecommuni-cation services without worryingabout user’s location,access technology, or device.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 94

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 2009 95

data services like Web browsing or contentstreaming, VCC mobility for voice services con-trolled by IMS between CS and PS networks,and MMSC mobility for multimedia servicescontrolled by IMS.

The enhanced mobility step begins when theoperator takes advantage of hybrid-like solutionsthat blend the benefits of each technology (e.g.,those standardized by 3GPP or IEEE) to buildmobility architecture with enhanced policy deci-sions or enhanced mobility triggers (referring toMIH events).

CONCLUSION

We present an overview of network convergencecoupled with the services and their levels of inte-gration as well as possible impacts on the net-work architecture. A comprehensive survey ofdifferent standardized seamless convergencesolutions is discussed with an in-depth observa-tion of their potential benefits and limitations.The ultimate goal of this study was to assess dif-ferent mid- to long-term architectural scenariosfor convergence of heterogeneous access net-works. Particular emphasis was on terminalseamless mobility between different access net-works, ensuring continuity of service even for themost stringent types of applications.

Seamless convergence of heterogeneousaccess networks is essential in today’s telecom-munication systems. Accordingly, operators canprovide telecommunication services without wor-rying about user location, access technology, ordevice. In fact, seamless convergence avoids theproblems of maintaining multiple networks,which obviously creates interoperability issuesand complicates maintenance, support, andupgrading. In addition, it provides the opportu-nity for fixed-only operators to defend their busi-ness against the current trend towardmobile/nomadic services, and also enables inte-grated operators (offering both fixed and mobileservices) to avoid developing separate facilitiesfor each type of network. In contrast, the seam-less convergence of heterogeneous access tech-nologies enables network operators to offerenhanced services and better user experience. Itis also worth mentioning that seamless conver-gence was discussed within several industry dom-inated projects, for instance, Daidalos2 andAmbient,3 among others.

To complete the convergence puzzle, iftoday’s approaches are mainly focused on inte-grating WLANs with UMA and I-WLAN as ashort- and mid-term solution, focus in the dis-tant future might be to integrate cellular andnon-cellular technologies. MIH coupled withmobility protocols at the higher layers (DSMIPv6on the handset or Proxy MIP on the network forhandset without MIP capabilities) and in coordi-nation with IMS throughout VCC or MMSCapplication servers could indeed be the future ofseamless convergence enabled network architec-tures.

The enhanced defined mobility correspondingto the long-term vision could be structuredaround hybrid-like solutions that both blendadvantages of different technologies specified bydifferent standardization bodies (e.g., 3GPP,

IEEE) and also are not limited to the scope ofspecification for each technology. As an instanceof this latter case, coupling access to a mobilenetwork (through UMA or I-WLAN technolo-gies) with a simultaneous openness to the homenetwork permits new usage for the end userwhen entering the domestic area. Connectivity tothe home network gives the user access to his/herfixed driven services (e.g., local contentexchange, local media transfer within an ongoingsession), and connectivity to the mobile networkoffers reachability whatever the user’s mobilitystatus. These examples enforce the promise ofbeing not only always connected, but always bestconnected.

REFERENCES[1] U. Javaid, T. Rasheed, and D. E. Meddour, “Cooperation

in 4G Systems — A Service-Oriented Perspective,”Wksp. Wireless Mesh Sensor Net. (with IEEE VTC-Spring), Dublin, Ireland, Apr. 2007.

[2] J. McNair and F. Zhu, “Vertical Handoffs in Fourth-Gen-eration Multinetwork Environments,” IEEE WirelessCommun., June 2004, pp.8–15.

[3] European Commission ICT Report for Brussels RoundTable, “Telecoms in Europe 2015,” Feb. 2007.

[4] G. Camarillo and M. Garcia-Martin, The 3G IP Multime-dia Subsystem (IMS): Merging the Internet and the Cel-lular Worlds, Wiley, 2006.

[5] L. Vollero and F. Cacace, “Managing Mobility andAdaptation in Upcoming 802.21 enabled Devices,”Proc. 4th Int’l. Wksp. Wireless Mobile Applications Ser-vices WLAN Hotspots (with MOBICOM 2006), Los Ange-les, CA, Sept. 2006.

[6] S. Parkvall, “Long-term 3G Evolution — Radio Access,”Ericsson res. rep.

[7] 3GPP TS 22.234, “Requirements on 3GPP System toWLAN Interworking (Release 6),” 2005–2006.

[8] 3GPP TS 22.234, “Requirements on 3GPP System toWLAN Interworking (Release 7),” 2006.

[9] 3GPP TS 23.237, “IP Multimedia Subsystem (IMS) Ser-vice Continuity,” 2008.

ADDITIONAL READING[1] I. F. Akylidiz et al., “A Survey of Mobility Management in

Next-Generation All-IP-Based Wireless Systems,” IEEE Wire-less Commun., vol. 11, no. 4, Aug. 2004, pp. 16–28.

[2] E. Gustafsson and A. Jonsson, “Always Best Connect-ed,” IEEE Wireless Commun., Feb. 2003, pp. 49–55.

[3] IETF RFC 3775, “Mobility Support in IPv6.”

BIOGRAPHIESDJAMAL-EDDINE MEDDOUR [M] received his computer engi-neering degree with honors from Institut National d'Infor-matique, Algiers, Algeria, in 2000, his Master's degree incomputer science from the University of Versailles, France,in 2001 and his Ph.D. in Computer Science from Universityof Paris VI in 2004. He is currently a senior researcher withOrange Labs (formerly France Telecom R&D), Lannion,France. His main research activities concern resource man-agement for wireless mesh networking, quality of service inhome networks, and multimedia delivery, management,and interoperability in new-generation wireless networks.He is very active in research communities as a guest editorand TPC member for numerous international conferencesand journals. He is the co-author of several internationalarticles and book chapters, and holds many patents. He isan active member of the IEEE 802.16j and 802.16m groups.

NICOLAS BIHANNIC received his Engineer degree in telecom-munication from the Institut Supérieur de l'Electronique etdu Numerique in 2000. He worked as a consultant on therange 3G access network rollout within Orange Supportand Consulting, Paris, France. In 2004 he joined OrangeLabs to work on fixed and mobile convergence architec-tures in Lannion, France. His research interests focus on thedevelopment of new converged service architecture in thehome network area.

TINKU RASHEED received his B.E. degree in electronics andcommunications from Kerala University, India; his M.S.degree in telecommunications from Aston University, Birm-

2 IST Daidalos project;http://www.ist-daidalos.org

3 IST Ambient Networkproject; http://www.ambi-ent-networks.org/

The enhanceddefined mobility

corresponding to theLong Term Vision

could be structuredaround hybrid-likesolutions that are

both blending advan-tages of different

technologies specified by different

Standardization bodies and also notlimited to the scope

of specification foreach technology.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 95

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.

IEEE Wireless Communications • June 200996

ingham, United Kingdom; and his Ph. D. degree in com-puter science from the University of Paris-Sud XI, Orsay,France, in 2002, 2004, and 2007, respectively. He was withOrange Labs in France from May 2003 until December2006, where he was extensively involved in the develop-ment of routing protocols and QoS provisioning schemesfor large-scale heterogeneous mobile networks, and con-tributed to several national and internal projects. Currently,he is a senior research staff member of Create-Net’s Perva-sive Research Group, Trento, Italy, where he is involved inseveral industry and European funded projects. He has sixgranted patents and is the author or co-author of over 30journal and conference publications. His research interestsinclude computer networks and protocols, wireless net-working, large-scale heterogeneous systems, performanceevaluation, and implementation.

USMAN JAVAID is serving as a new technologies and innova-tion specialist at Vodafone Group, Newbury, United King-dom. He is responsible for technical assessment of newtechnologies, which create new revenue opportunities forVodafone and significantly reduce the cost of offering ser-vices. He is also leading the global precommercial trials ofthe fourth generation of mobile networks called LTE, in

close collaboration with Verizon Wireless and China Mobile.Prior to Vodafone, he served Orange, France Telecom Groupas a research engineer, and was actively involved in Euro-pean and French research projects. He earned his Ph.D.degree in the field of mobile wireless networks from theUniversity of Bordeaux, France, and his Master’s degree intelecommunications from the University of Paris.

RAOUF BOUTABA [M'93-SM'01] ([email protected])received M.Sc. and Ph.D. degrees in computer science fromthe University Pierre & Marie Curie, Paris, France, in 1990and 1994, respectively. He is currently a professor of com-puter science at the University of Waterloo, Ontario, Cana-da. His research interests include network, resource, andservice management in wired and wireless networks. He isthe founder and Editor-in-Chief of IEEE Transactions onNetwork and Service Management, and is on the editorialboards of several other journals. He is currently a distin-guished lecturer of the IEEE Communications Society, chair-man of the IEEE Technical Committee on InformationInfrastructure, and Director of ComSoc Conference Publica-tions. He has received several best paper awards and otherrecognitions including the Premier’s Research ExcellenceAward.

BOUTABA LAYOUT 6/4/09 3:48 PM Page 96

Authorized licensed use limited to: National Chung Cheng University. Downloaded on November 29, 2009 at 13:39 from IEEE Xplore. Restrictions apply.