Cooperative networks for the future wireless world

IEEE Communications Magazine • September 200470 0163-6804/04/$20.00 © 2004 IEEE

WIRELESS WORLD RESEARCH FORUM

INTRODUCTION

Beyond-third-generation (B3G) systems havebeen envisaged as an evolution and convergenceof mobile communications systems and IP tech-nologies to offer a multitude of services over avariety of access technologies. To fulfill thevision, it is necessary to understand the require-ments with respect to the support of heterogene-ity in network accesses, communication services,mobility, user devices, and so on. It is equallyimportant to promote the necessary research innetworking technology by providing a guidingframework of research areas and technical issueswith priority. The new architectures and tech-nologies will have to address the fundamentalassumptions and requirements that govern thedesign. All these are being tackled by the Coop-erative Network working group (CoNet) of theWireless World Research Forum (WWRF),which is working on a series of white papers out-lining B3G visions and roadmap, architecturalprinciples, research challenges, and candidateapproaches. In the remainder of this articleCoNet’s main objectives are outlined, while weexamine the most important elements of CoNet’sarchitectural principles. Additionally, we addressthe flamboyant research challenges in coopera-tive networks and try to give a glimpse of whatwe think would govern their design and encom-pass their futuristic form by introducing themain network components and technologies.This is mainly accomplished by focusing on three“hot concepts”: mobility management, multipleaccess, and moving networks.

Christos Politis, University of Surrey, U.K.; Toshikane Oda, Nippon Ericsson, Japan;

Sudhir Dixit, Nokia, USA; Andreas Schieder, Ericsson Eurolab GmbH, Germany;

Hong-Yon Lach, Motorola Labs, France; Michael I. Smirnov, Fraunhofer FOKUS, Germany;

Sami Uskela, Nokia, Finland; Rahim Tafazolli, University of Surrey, U.K.

ABSTRACT

Beyond-3G (B3G) systems have been envis-aged as an evolution and conv∂ergence ofmobile/wireless communication systems and IPtechnologies to offer a multitude of servicesover a variety of access technologies. To fulfillthe vision, it is necessary to understand therequirements with respect to the support ofheterogeneity in network accesses, communica-tion services, mobility, user devices, and so on.Besides, it is equally important to promote thenecessary research in networking technology byproviding a guiding framework of researchareas and technical issues with priority. Thenew architectures and technologies will have toaddress the fundamental assumptions andrequirements that govern the design. All theseissues are being tackled by the CooperativeNetwork working group (CoNet) of WWRF;the group is working on a series of white papersoutlining B3G visions and roadmap, architec-tural principles, research challenges, and candi-date approaches. This article outlines theCoNet concept, architectural principles, andguidelines for research into cooperative net-works assuming that the B3G systems will bebuilt over generic IP networking technologies.The article also presents the key research chal-lenges, research framework, and major networkcomponents and technologies. The key pointsare that the system should be layered ondemand, encourage reuse of independent mod-ularized functional blocks, support multiple ser-vices and service creation, ensure consistentend-to-end connectivity across different accesstechnologies, and cooperate in terms of net-work control, operations, and maintenance. Thearchitecture shall include the endpoints of com-munications as part of the communications sys-tem, and should provide a secure and trustedenvironment in which network functions areperformed; the network should self-organizedynamically. Additionally, the article presentsthe IST WSI architecture proposed to CoNet as

Cooperative Networks for theFuture Wireless World

a reference model along with some approachesto the outlined research challenges. Finally, thisstudy selects three important network compo-nents and technologies (i.e., mobility manage-ment, multiple access, and moving networks) inorder to provide with answers and possiblesolutions the research challenges presented inearlier sections.

Authorized licensed use limited to: University of Surrey. Downloaded on May 13,2010 at 16:26:30 UTC from IEEE Xplore. Restrictions apply.

IEEE Communications Magazine • September 2004 71

CONET OBJECTIVES

The CoNet concept will enable seamless com-munication activities on mobile devices operat-ing in networks composed of heterogeneoustechnologies by enhancing the secure networkingcapability of B3G systems in terminal devices aswell as in the infrastructure. It is envisaged thatB3G networking by 2010 will support [1]:• Coexistence and convergence of different

legacy (wireless, wired, broadcast, etc.) andnew networks, including moving and ad hocnetworks

• Evolutionary deployment of secure, extensi-ble, scalable, reconfigurable, and manage-able B3G systems

• Independent evolution of network technolo-gies and services

• Controlled service access by users anddevices anywhere anytime by different means

• Pervasive and seamless mobile multipartymultimedia communications and access toall kinds of services to minimize adverseimpact on user experience in the changingnetworking environment

• Value-added interface to upper-layer appli-cations to enable and enhance their loca-tion, context, and quality of service (QoS)awareness

• Interactions with access systems for optimalmobility management and better use ofscarce radio resources

• Unrestricted and innovative business modelsThe desire to seamlessly deliver application

services across heterogeneous access networks ina B3G system leads to an obvious conclusionthat B3G systems would best be built on IP net-working technologies. However, it is also obviousthat existing IP networking technologies do notprovide all the necessary features to enable aB3G system and must be extended.

Therefore, it is inappropriate to set a singleroadmap for technology development to achievethe vision, due to different concerns and priori-ties with respect to technology maturity, technicaldifficulties, market requirements, business mod-els, and more. There will be different paths pur-sued, learning and leveraging from each another.

The following are some possible goals B3Gresearch efforts could aim for to achieve thevision outlined above [1]:• A system in which a user can move seam-

lessly between cellular, wireless, and wiredaccess networks

• Interdomain administration and manage-ment support to allow seamless mobilitybetween administrative domains

• Support of multiparty multimedia (includingQoS-sensitive) sessions between networksand administrative domains

ARCHITECTURAL PRINCIPLESThis section describes architectural principlesand basic requirements commonly regarded asthe cornerstone for studying related technologiesand designing future IP-based cooperative sys-tems. These features mainly include functionalblocks, multiple services and their creation, con-nectivity issues, end-to-end concept, security,

control, operation and maintenance, and self-organization [2].

In CoNet, a layered approach is globally sup-ported. This means that the functionality of thesystem is grouped into distinct layers that formlogically separate subsystems. Any layered modelhas at least three layers: application, connectivi-ty, and access. The application layer can be fur-ther divided into service application and servicesupport sublayers. The connectivity layer can befurther divided into network control and trans-port (IP) sublayers. And the access layer maycontain several independent access networksthat can also function simultaneously.

The layers should have well defined inter-faces and be functionally independent of eachother. Such an approach is required to ensureeasy adaptation of heterogeneous access tech-nologies, related technology changes, and flexi-ble support for rapid service innovation.

Because of the need for cross-layer optimiza-tion (e.g., for power management, QoS support,etc), the layered approach in CoNet is consid-ered as a design principle rather than as a uni-versal design pattern, meaning that every B3Gsystem may potentially have its own layeredarchitecture.

INDEPENDENT FUNCTIONAL BLOCKS,MODULARIZATION, AND REUSE

The CoNet architecture should define each func-tion as an independent functional block. Thispromotes component-based modular architec-ture, where different building blocks can becombined as needed to realize complex systems.This in turn requires standardized definition ofsyntax and semantics of these interfaces. Forinstance, functionalities required to process androute user data, handle control signaling, anddeal with network management can be separatedinto different functional blocks in the user, con-trol, and management planes, respectively. Itshould be noted that each layer of the architec-ture has its own separate set of functions foreach plane; hence, nine different functionalblocks can be identified, as shown in Fig. 1.

Realization of the functionalities as indepen-dent building blocks allows the introduction ofnew functions when needed without changingthe whole architecture. The building blocksshould be reused, when applicable, with differentaccess technologies and realization of various

� Figure 1. Layers and planes.

Access layer

Applicationlayer

Connectivitylayer

Service application sublayer

Service support sublayer

Network control sublayer

Transport sublayer

Userplane

Controlplane

Managementplane


IEEE Communications Magazine • September 200472

services. Therefore, implementing the same orvery similar functionality multiple times shouldbe avoided.

COOPERATIVE CONNECTIVITYThe CoNet architecture should ensure connec-tivity between all the entities of a network in aconsistent manner across all access technologiesfor any service. This requires consistent supportfor device mobility, QoS, authentication, autho-rization, and accounting (AAA),1 and so on. Theconnectivity layer provides cooperation acrossvarious realizations of networks, called coopera-tive connectivity, and shall be independent of thevarious transport technologies used to link thenodes of the network together.

By separating access and transport the CoNetarchitecture makes it transparent to the commonand standardized transport infrastructure andhides the technology from the end user, whilefacilitating the most efficient usage of spectrumresources. The CoNet architecture shall supportboth session continuation (i.e., handover whenthe access technology used is changed) andsimultaneous usage of several access networks.Finally, the user should be able to seamlesslyroam across different access technologies andadministrative domains without any manual userintervention. This implies that the CoNet archi-tecture should also support connection of sub-scribers to private IP networks through NetworkAddress Translators (NATs) and Simple Traver-sal UDP through NATs (STUNs).

AN END-TO-END APPROACHThe CoNet architecture should include the end-points (i.e., terminals) as part of the communica-tion system, and support end-to-end (E2E)negotiation and fulfillment of QoS parameters,security settings, and so on. This E2E approachdoes not mandate that all functionality should belocated in the endpoints. On the contrary, thefunctionality can also be provided in hop-by-hop,and/or edge-to-edge manner subject to properand lawful termination of transport connections.The CoNet architecture should ensure that theinteroperability of (and communication between)heterogeneous endpoints is maintained.

SECURITYThe CoNet architecture should provide a secure-ly protected environment in which network ele-ments are deployed and interact. Besidesprotecting end users and networks from eachother through authentication and authorization,the components of the CoNet concept shouldalso be protected from any intrusion or mali-cious attack. The architecture should provide asecurity framework to enable protection of pri-vate information and data. Furthermore, theCoNet architecture should provide accountingcapabilities and further enhance the AAAparadigm. The architecture should be distributedand consist of security components able to com-municate through well defined interfaces. Under-lying mechanisms such as DIAMETER [3] wouldprovide the necessary secure communicationframework where multiple networks cooperate.

To accomplish these, we would be welladvised to rely on already existing security algo-

rithms that are stable, resilient, and well estab-lished. The creation of new mechanisms andprotocols is outside the scope of CoNet. Rather,its objective is to build the trust models andsecurity associations between the various compo-nents of the distributed CoNet architecture withscalable access control to distributed compo-nents and resources.

SELF-ORGANIZING CONETSThe CoNet architecture is expected to comprisea number of networks exhibiting different capa-bilities in terms of coverage, capacity, transmis-sion rates, transmission delay, and transmissioncost. Networks might be in a cooperative orcompetitive relationship with each another.

Part of the access domain in the wirelessworld may not be centrally organized in thefuture and may even provide infrastructurelessconnectivity. Nodes may come and go and beloosely associated with each other, formingalliances whenever and wherever. Such nodesthat could be part of the personal, the local, andeven the global sphere would temporarily coop-erate to provide connectivity in an ad hoc man-ner or share application resources. A networkthat is a member of the CoNet has to build rela-tions with other networks to provide expectedglobal connectivity and support the demandedaccess versatility. Relations between cooperativenetworks are supposed to be established dynami-cally. The creation of these dynamic relation-ships is a first aspect of self-organization.

The dynamic organization of relationsbetween networks and individual network ele-ments is supposed to result in a CoNet structurethat adapts its topology to meet the demands ofvarying traffic patterns and transmissiondemands. To ease administration and operation,the network nodes should mostly be self-config-uring, and resources should be distributed amongthem dynamically to cope with varying trafficvolumes and traffic characteristics.

THE ARCHITECTURAL FRAMEWORKAn architectural framework supporting theresearch work on CoNet should aid the structur-ing of research tasks and identify relevant fieldsof research, including a reference model2 toensure that research toward 4G is carried out ina harmonized and open-ended manner. TheInformation Society Technology (IST) projectWireless Strategic Initiative (WSI) has proposedits reference model to WWRF.

THE WSI REFERENCE MODELThe basic structure of the WSI reference modelcovers all aspects of the wireless world frombusiness models and user issues down to radiointerfaces. The reference model describes thegrand building blocks of the wireless world andhow they interact at reference points, and cap-tures user scenarios and different views. Themodel also describes what kind of elements canbe used to set up future wireless communicationsystems. It identifies how these elements caninteract, the functionality they have to provide,and how a system can be composed out of them.

Following the sphere model, developed by

1 Within the InternetEngineering Task Force(IETF) AAA combinesauthentication, authoriza-tion, and accounting;within CoNet this concepthas to be enhanced withbasic support for charging,roaming, auditing, anddigital rights management.

2 The development of theWWRF and its referencemodel was initiated by theEuropean IST projectWireless Strategic Initia-tive [4].

Besides protecting

end users and

networks from each

other through

authentication and

authorization, the

components of the

CoNet concept

should also be

protected from

any intrusion or

malicious attack.



the WSI think tank in 2000, the reference modelidentifies the building blocks of the wirelessfuture. The main achievement is the definitionof a communication element (a representation ofa certain device or node) that acts as a commu-nication entity in the different spheres; a largenumber of such elements may exist in the wire-less world. Depending on their current commu-nication context, they can be logically placed indifferent spheres (e.g., global, local, and person-al) reflecting the proximity of building blockswith respect to a user. Therefore, the personalsphere can be seen as some kind of body areanetwork, the local sphere serves as local net-working infrastructure, and the global sphere isresponsible for global connectivity.

The functionalities integrated in the commu-nication elements are provided by differentbuilding blocks, assuming that the referencemodel should separate content processing, con-trol, and management functions (see the sub-blocks inside a communication element) intotheir own E2E planes and subsystems. The build-ing blocks that form communication elementsare connected by vertical and horizontal refer-ence points. The vertical reference points defineinterfaces between the building blocks of thecommunication elements, and horizontal refer-ence points define interfaces between communi-cation elements that reside in different spheres.

RESEARCH CHALLENGESIn order to figure out features and characteris-tics the envisioned network shall have in differ-ent aspects, it is necessary to identify thesignificant requirements for the target networkin terms of service provisioning and system capa-bilities. Those requirements will give an integralpart of the guiding framework for investigationof design philosophies, principles and guidelinesfor reference models, network functional archi-tectures, and solutions for individual functionalcomponents.

Furthermore, it is vital to identify and under-stand technical components that require furtherresearch, taking into account the requirements forthe target network. We describe overall problemareas and generic technical issues to be addressedto provide guidance for defining the scope andspecific targets of respective ongoing and futureresearch activities/projects in the field of network-ing technologies to realize the CoNet vision [5].

MULTI-ACCESS CAPABILITYIn order to enable a system for multiple accessnetwork technologies, a number of differentissues need to be investigated. Here we discusssome of these issues.

The device needs to be provided with access-technology-independent capabilities (applicationprogramming interfaces, APIs, and protocols) todetect which different access networks are cur-rently available and learn their characteristics.Similarly, mobile terminals should be able toscan in a specific environment to discover candi-date available access networks and register somepolicy issues.

Furthermore, there are different ways toenable access for a user and/or device over multi-

ple access network technologies. Either the usercould log on to each access network separately,or the authentication and authorization mecha-nisms for the different access networks could beconnected to allow the user/device to movebetween different access networks without need-ing to log on multiple times. Preferably, the usershould not have to close down and/or restartapplications when moving between differentaccess networks. To enable this, we might need asolution for mobility management that is tied tothe logon procedures for the different access net-works. Concurrently, APIs toward the applica-tions should be designed so that an applicationcan communicate requirements for access selec-tion to the network or device, and also obtaininformation about the access network character-istics currently used. There is a need for a multi-access paging mechanism. It should be possibleto page a terminal with multiple accesses even ifsome of the interfaces are switched off. SinceB3G systems will potentially have a wireless linkin the E2E connection that is either inherentlyunreliable and/or deal with handovers, solutionswill be needed to support E2E session-level relia-bility, where a session goes into hibernation whendisconnection happens and is resumed when asignal reappears after a period of disconnection.

MOBILITY MANAGEMENTThe architectural changes in the networking infra-structure are mainly dictated by the increasing lev-els of heterogeneity in modern networks, theirinterconnectivity, and their transport capabilities.Single access shifts toward multiple simultaneouslyused access networks, also known as multihoming.Single address space shifts to multiple coexistentaddress spaces — IPv4, IPv6. Single device usageshifts toward multiple device usage: PDA, smartphones, laptops, digital cameras, and more, usedinterchangeably or at the same time for differentpurposes. Additionally, all networking entitiesbecome mobile: users changing devices, deviceschanging access networks, access networks allocat-ing network (IP) addresses dynamically and fromdifferent address spaces. Personal area networks,becoming reality in the immediate future, areinherently mobile. The use of vehicular and trans-port related networking will also increase withcompletely new applications, for example, in thetelemetric area, which put new requirements oncapabilities for mobility support. In essence,devices and gadgets without networking capabili-ties today are moving toward being networked asthey serve as sinks and sources of digitally format-ted information and content as well as runningapplications with networking needs.

Considering the above background, CoNetmobility management primarily consists of twomajor functional components: mobility routingmanagement, including handover, and locationmanagement, providing functionality of locationupdating and paging.

A large number of technical issues need to beaddressed in future research in the area ofmobility management in order to achieve therequired mobility routing capability in futurenetworks. The following lists some of the issues.• Mobility architecture (mobility support on a

number of levels, layers, and protocols)

The building

blocks that form

communication

elements are

connected by vertical

and horizontal

reference points.

The vertical reference

points define

interfaces between

the building blocks

of the communica-

tion elements, and

horizontal reference

points define

interfaces between

communication

elements that reside

in different spheres.



• Group mobility (formation and mobility ofnetworks, aggregation of data flows, groupmobility support on several levels, supportfor multicast mobility)

• Session continuity (seamless support withefficient, timely, and predictable behavior,harmonized with transport protocols, sup-porting context and/or state transfer)

• Naming and addressing, mobility triggering,proactive mobility support

MOVING NETWORKSA moving network (NEMO) consists of one ormore mobile routers (MRs) with a number ofdevices connected to them. These networks canchange their point of attachment to other net-works because the NEMO physically moves orchanges in topology.

NEMOs appear in different sizes, from smallgroups with a few devices attached to an MR upto NEMOs on ships and trains that might haveseveral hundred or more mobile nodes (MNs)attached to them. The administrative character-istics will differ between NEMOs as well,whether the MRs and MNs are administeredand owned by the same or different entities. Insmall NEMOs the MR and MNs are typicallyowned by the same entity, while in large NEMOsthe MR(s) and MNs are typically owned by dif-ferent entities. This affects, among other things,the level of trust between the different nodes.Different types of MNs have different require-ments on location updates and connections to

external networks (home operators) when theMNs are not actively participating in a session.A laptop might not require instantaneous loca-tion detection to a peer in another network,while a cell phone has to be reached immediate-ly when someone is calling it. This puts differentrequirements on what and how much updatetraffic will be sent through the MRs.

Here are some research areas that could beaddressed:• Addressing and scalability.• Speed: Some NEMOs may move at high

speeds, which puts requirements on hand-overs and location updates, as well as fre-quency of these handovers and locationupdates.

• Mobility management, nested mobility,access control, and security.

• Multihoming: A mobile node can beattached to multiple ISPs/operators.

• Aggregation of signaling and user data traf-fic between MRs and external networks.

NETWORK COMPONENTS ANDTECHNOLOGIES

The objective of this section is to provide severalpossible solutions for research challenges, suchas mobility management, multiple access IPQoS, and moving networks,3 that are identifiedin [5] and discussed in previous sections. Thepossible solutions should take into consideration

� Figure 2. The general architecture of IP2 MM.

Routing address to AR


Locationarea


IPha

IPha

Receiver(IPha)

Routingaddressallocation

Routingupdating

RAC

Locationupdating

Page

AR(receiver)

AR(sender)

IPha

IPha

Paging request

Routing cache retrieval

Location management Location manager

Advertise(location area)

Routing management

Routing manager

Transport network

Sender

3 These three research top-ics/challenges were pickeddue to their “great weight”in the global researchcommunity.



the requirements identified in [5] and the archi-tectural guidelines proposed in [2] and furtherdescribed here.

This section also provides evaluations of thepossible solutions and shows how these solutionscan be quantitatively effective with respect to therequirements and architectural guidelines.

MOBILITY MANAGEMENTOne mobility management concept is based onIP2, which is described in [6]. The solutionassumes a routing manager and a location man-ager to reside in the network control sublayercontrolling the access layer. Both managementfunctions treat routing information and locationinformation separately. The managers are trig-gered from an access layer to initiate access ofmobile terminals (MTs), and perform handoverand location updates and other functions. Themanagers are not to be directly accessed by theMTs. This enables the managers to be protectedfrom malicious attempts and may evolve inde-pendent of the access layer.

The routing manager creates, updates, anddeletes the routing cache in the transport sublay-er (i.e., in the routers) for the MT it is currentlymanaging. The location manager manages theMT location information and pages the MTwhen necessary. The routers forward packetsand perform address exchange procedures forpacket routing and rerouting. This enables MTsto communicate at any time without revealingtheir locations or movements to the peer termi-nal(s). In addition, this mechanism can also real-ize efficient utilization of radio frequencies(resources), MT power consumption, and use ofnetwork resources. Figure 2 shows a genericarchitecture of the proposed solution.

According to the vision of CoNet, the IP hostaddress (IPha) of an MT should not have to bechanged in order to maintain communication,even when the IP routing address (IPra) changesdue to its movement.

The IPha is used only to identify an MT, whilethe IPra is used only to transport packets withinnetworks and thus for location information. Onthe other hand, MTs are not aware of the IPra by

exchanging the IPha and IPra at the edge router(router between MT and network), and packetssent to MTs are not encapsulated. Therefore, theproposed routing mechanism can prevent thelocation information of the MT being disclosed tothe peer terminal. In addition, it can reduce thepacket header overhead caused by encapsulation.

Figure 3 shows the separation between IPhaand IPra. The access router (AR) of a peer ter-minal receives the packet of an MT (i.e., thedestination address of the packet, which is theIPha). It then changes the destination addressesfrom IPha to IPra according to the routingcache, where the IPha and IPra are mapped; thiswhole process is managed by the routing manag-er. Then the AR forwards the packet to the MTaccording to its routing table. The packets canbe routed to the AR to which the MT isattached, since the AR advertises the route inthe network prefix of the IPra. The AR to whichthe MT belongs receives the packet andexchanges the destination addresses according tothe routing cache from IPra to IPha. Then theAR forwards the packets to the MT with theIPha as the destination address.

If the AR of a peer terminal does not havethe routing cache for the MT IPha, the ARrequests the routing manager of the MT to setthe routing cache for address resolution.

When the MT moves from an AR to anotherAR, the routing manager is triggered by therouting access controller (RAC) in the accesslayer, as previously mentioned (Fig. 2). Then itupdates the routing cache in the AR of the peerterminal and creates a new one in the new ARto which the MT is attaching.

Before the update and creation steps, thenew AR allocates the MT with a new IPra andnotifies the routing manager of this. The accesslayer or routing manager triggers the new AR toallocate the IPra.

MULTI-ACCESS IP QOSIt is assumed that the E2E packet delivery ser-vice is provided at the IP layer. The quality ofthis delivery service determines the E2E QoSthat is available for the applications. From the

� Figure 3. The concept model of separation between IPha and IPra.

AR

AR

Peer terminal

Exchange(IPha → IPra)


Routing managerNetwork controlsublayer

AR

MT

MT

Moving

IPraIPha IPra

IPha

IPha

IPra

IPra


The managers are

triggered from an

access layer to

initiate access of

mobile terminals and

perform handover

and location updates

and other functions.

The managers are

not to be directly

accessed by the

mobile terminals.


https://www.researchgate.net/publication/3344179_IP-based_IMT_network_platform?el=1_x_8&enrichId=rgreq-ec860619-deb2-4fd3-b3db-bf912341ff43&enrichSource=Y292ZXJQYWdlOzMxOTkyMjM7QVM6MTUwNzU2MDg2Nzg0MDAyQDE0MTI5NTQ0NjA5NTg=


applications’ perspective, the key elements ofE2E QoS control are related to the followingaspects (Fig. 4):

•The mechanisms and interfaces that allowthe application (and application-level entities) tospecify the QoS requirements. For instance, theapplication may be associated with a configura-tion or management tool that allows the specifi-cation of default (qualitative or quantitative)QoS parameters assigned to the application.Alternatively, a configuration tool may providedefault QoS settings for a group of applications.

•The QoS requirements need to be trans-ported to each domain along the E2E path. Adomain may use this information to configuredomain-specific functions and resources suchthat the required QoS can be provided withinthe domain. When a new domain becomes partof the E2E connection (due to mobility or, e.g.,access selection) QoS related information mayhave to be transported to this new domain.

•Each domain may have specific mechanismsto support QoS requirements. Examples of suchmechanisms include per-flow admission controland resource reservation (e.g., in a cellular wire-less network), scheduling and prioritizationmechanisms (e.g., in a differentiated services IPdomain), or resource overdimensioning (e.g. in abest effort IP network).

The E2E QoS architecture of Fig. 5 builds onthe high-level view of Fig. 4 and takes into accountthe above three aspects of E2E QoS provisioning.A communication session typically begins withexchanging application and session layer informa-tion. Such information exchange allows theinvolved parties to negotiate service capabilitiesand application-specific attributes such as encod-ing formats and version numbers. The communi-cating parties may proceed by requesting anappropriate bearer service from the network. Thisphase requires that QoS information be distribut-ed to all domains of the E2E path. Specifically,the requirements on the E2E QoS informationdistribution mechanism should include:• Applicability over various QoS technologies.

• Clear separation from the control informa-tion being transported.

• Independence from the application.• Independence of the flow identification

(state association with a user flow and flowidentifier must be independent).

• Support for both uni- and bidirectionalreservations

• Efficient service re-establishment afterhandover. In particular, the distributionmechanism must be able to seamlessly inter-workwith other mobility and handover protocols.

• Admission control and resource reservationin domains along the E2E path must enablea domain to communicate the result of anadmission decision to involved parties.Once the QoS information is distributed

along the E2E path, each domain may configureits own respective mechanisms to provide thenecessary QoS for the user data flows.

MOVING NETWORKTo support moving network mobility, varioustechniques are proposed in the IETF NEMOWG [7]. However, all these proposals are basedon Mobile IP; thus, there are several open issues,which are the same as those of Mobile IP fromthe mobile carriers’ points of view regarding tri-angulation and route optimization techniques[8]. For instance, because of the packet over-head due to encapsulation, we cannot achieveroute optimization and location confidentiality atthe same time. In addition, these proposalsassume that the prefix inside a NEMO is con-stant even if handoff occurs. Therefore, it makesit possible to hide the mobility of a NEMO fromMTs; however, it causes more packet overheaddue to several encapsulations, or using the rout-ing header option to support route optimization,in comparison to host mobility.

As an alternative solution, four concepts areproposed: care-of prefix (CoP), aggregate router(AGR), concatenated routing management, and hier-archical address management to achieve the samequality as host mobility. However, location confi-dentiality and route optimization cannot be kept atthe same time, which is an open issue in MobileIP. In addition, the Mobility Management for IP-Based IMT Network Platform (IP2) is proposed asone of the possible approaches for the next-gener-ation mobile communications network beyond 3Gsystems [6, 8] to resolve this open issue.

According to the NEMO’s assumption, the IPpacket prefix in the NEMO is constant, and thepacket header size gets bigger than the host. Toresolve this problem, it is necessary to identify thecare-of address (CoA) of an MT in the NEMO aswell as the MT that can be anywhere in the corenetwork. The CoP enables this. More concisely,the CoP changes the prefix inside the NEMOevery time the NEMO moves to another accessrouter, and gets the correct topological address(CoA) of the MT in the NEMO. When the NEMOconnects to an AR, the AR assigns a new prefix,which shows that the NEMO has moved. Each MTin the NEMO uses this prefix to generate the CoA.In this way, an MT’s CoA can be identified fromanywhere in the core network. Furthermore, theallocated prefix is original for each NEMO inorder to avoid generating duplicate CoAs even

� Figure 4. A high-level view of a multi-access multidomain scenario.

QoS requirementscome from the

application

QoS requirementsare propagated toeach domain by IP

QoS signaling

IP QoS signalling

Each domain hasdomain- specific

mechanisms to supportthe QoS requirements

L2

L2L2

L2

L3

IP

APPL APPL APPL

IP

APPLAPPLAPPL



https://www.researchgate.net/publication/225201155_Route_Optimization_for_Mobile_IP?el=1_x_8&enrichId=rgreq-ec860619-deb2-4fd3-b3db-bf912341ff43&enrichSource=Y292ZXJQYWdlOzMxOTkyMjM7QVM6MTUwNzU2MDg2Nzg0MDAyQDE0MTI5NTQ0NjA5NTg=



when handoff occurs. In the CoP, all CoAs of allMTs in the NEMO are changed when handoffoccurs, so the number of binding update signals tothe home agents (of all MTs and all core net-works), in order to realize route optimization, aresent according to the characteristics of the NEMO,as shown in Fig. 6. The aggregate router and con-catenated routing management can solve this prob-lem and are described in the following sections.

AGGREGATE ROUTERThe aggregate router (AGR) handles the mobilitymanagement of a NEMO. This is a router likethe modified mobile anchor point (MAP) in Hier-archical Mobile IPv6 (HMIPv6) [9]. In addition,the AGR aggregates the update signals (U-Plane)to the ARs of all CNs. This reduces the numberof binding update signals toward CNs with whichthe NEMO is communicating. Furthermore, thepackets destined to the NEMO are always sentvia the AGR. Therefore, AGR should be opti-mally located, considering route optimization foreach communication path. Moreover, if the Mov-ing Network moves too far, we should relocate toa different AGR. In addition, there are trade-offrelationships between route optimization, handoffdelay, and AGR relocation frequency. Forinstance, if the AGR is located near the NEMO,it may achieve route optimization and reducedhandoff delay, but AGR relocation may occur fre-quently. On the other hand, if located far fromthe core network, an AGR cannot achieve routeoptimization and reduced handoff delay; but inthis case, AGR relocation rarely occurs.

CONCATENATED ROUTING MANAGEMENT(CONCATENATED HOME AGENTS)

In this concept, each home agent (HA) onlyholds information of an MT’s whereabouts/attachments (e.g., with mobile router 1, MR1),

while the MR1 HA maintains all the bindinginformation for home address (HoA) and CoA.In this way, we can realize NEMO handoffs onlyby updating the MR1 HA (Fig. 7).

HIERARCHICAL ADDRESS MANAGEMENTWhen the NEMO moves, it is necessary toinform the AGR and MR1 HA of the updatedCoA, since these are changed. Therefore, thedata volume of binding update signals is verylarge. The proposed address management con-cept resolves this issue. Concretely, in this con-cept the CoA of each MT in the NEMO consistsof shared information and individual informa-tion. Shared information indicates the locationof the NEMO, which changes whenever a hand-off occurs. On the other hand, individual infor-mation provides the location of each MT in theNEMO and does not change even if handoffoccurs. The CoP enables this address manage-

� Figure 5. Setting up an end-to-end IP QoS bearer by means of a QoS bearer specification mechanism.

IP QoScontroller

Local L2/L3QoS

mechanism

IP QoScontroller

IP QoScontroller

LocalL2/L3 QoS

mechanismLocal L2/L3QoS

mechanism Domain 2 Domain N

User data

Candidate accesses in domain 1

1 Session layer bearer service negotiation

IP QoS signaling - NSISIPQoS

controller

APPL

User terminal A

User IP

API

IP QoScontroller

LocalL2/L3 QoS

mechanism

IPQos

controller

APPL

User terminal Bor server

User IP

API2

3

� Figure 6. Characteristics of moving networks.

MT MT

NCPF

MT



ment because the AR delegates the individualprefix to the NEMO, in order to avoid generat-ing duplicate CoAs. In more detail, MR1, AGR,and MR1 HA manage the binding informationas mentioned above, so we can realize a handoffof the NEMO by updating only the shared (com-mon) information (Fig. 8).

Summing up, hierarchical address manage-ment including an aggregate router and concate-nated routing management makes it possible tosupport frequent handoffs.

IP2 ROUTING FOR THE MOVING NETWORKIP2 resolves open issues in Mobile IP, as men-tioned earlier [8]. Therefore, by applying thefour key technologies for IP2 routing manage-ment as discussed in the previous subsections,we can get much better results.

The IP2 architecture is as shown in Fig. 9,where the aggregate router and concatenatedrouting management are introduced in the net-work control sublayer. In this proposed architec-ture, we can categorize the two routingmechanisms by their location confidentiality (i.e.,the location privacy information). Here, one caseis the gateway, with IPha and IPra, being theMR. In this case, the MR deals with IPra, whichindicates the location of the MTs; thus, we can-not keep the location information confidentialfrom the MR. Therefore, it may be easier if theMR is owned by carriers/providers. Another caseis that the gateway with IPha and IPra is an AR.In this case, the MR does not deal with IPra, sowe can keep the location information confiden-

tial from the MR and MTs. Therefore, it is pos-sible that users can own the MR (Fig. 10).

CONCLUSIONSThis article outlines the CoNet concept, archi-tectural principles, and guidelines for researchinto cooperative networks assuming that B3Gsystems will be built over generic IP networkingtechnologies. The article has also presented thekey research challenges, research framework,and major network components and technolo-gies. The key points are that the system shouldbe layered on demand, encourage reuse of inde-pendent modularized functional blocks, supportmultiple services and service creation, shouldhave consistent end-to-end connectivity acrossdifferent access technologies, and cooperate interms of network control, operations, and main-tenance. The architecture shall include the end-points of the communications as part of thecommunication system, should provide a secureand trusted environment in which network func-tions are performed, and should self-organizedynamically. Additionally, the article has pre-sented the IST WSI architecture proposed toCoNet as a reference model along with someapproaches to the outlined research challenges.

Finally, this study has selected three impor-tant network components and technologies (i.e.,mobility management, multiple access IP QoS,and moving networks) in order to provideanswers and possible solutions to the researchchallenges presented in earlier sections.

ACKNOWLEDGMENTSThe authors would like to acknowledge the con-tributions of their colleagues from the CoNetSIG in WWRF Working Group 3.

REFERENCES[1] H. Abramowicz et al., “The Wireless World Initiative: A

Framework for Research on Systems Beyond 3G,” ISTMobile and Wireless Commun. Summit 2004, Lyon,France, June 27–30, 2004.

[2] S. Dixit et al., “Cooperative Network Architectural Prin-ciples, Version 1.07,” WWRF, 10th mtg., Oslo, Norway,June 10–11, 2004.

[3] DIAMETER Base Protocol, IETF draft, http://search.ietf.org/internet-drafts/draft-ietf-aaa-diameter-11.txt

[4] IST-1999-12300, “WSI: Wireless Strategic Initiative,”http://www.ist-wsi.org

[5] T. Oda et al., “Co-operative Network Research Chal-lenges, Version 1.0,” WWRF, 10th mtg., Oslo, Norway,June 10–11, 2004.

[6] H. Yumiba, K. Imai, and M. Yabusaki, “IP-based IMTNetwork Platform,” IEEE Pers. Commun., vol.8, no.5,Oct. 2001, pp. 18–23.

[7] NEMO WG, http://www.nal.motlabs.com/nemo/[8] C. Perkins and D. Johnson, “Route Optimization in

Mobile IP,” draft-ietf-mobileip-optim-11.txt, Sept. 2001,work in progress.

[9] H. Soliman et al., “Hierarchical Mobile IPv6 MobilityManagement (HMIPv6),” draft-ietf-mobileip-hmipv6-07.txt, Oct. 2002.

BIOGRAPHIESCHRISTOS POLITIS [M] ([email protected]) received an engi-neering degree from the Technological University of Athens,Greece, in 1996, an M.Sc. in mobile and satellite communi-cations from the University of Surrey (UniS), United King-dom, in 1999, and his Ph.D. in mobile networking from theCentre for Communication Systems Research (CCSR) at UniSin 2004. Past positions include telecommunications engineerwith INTRACOM SA, Athens, IT engineer at AMSAT, and

� Figure 7. AGR and concatenated HAs.

MN1->CoA_mn1MN2->CoA_mn2

->MR1

MR1's HA MN1's HA

AR1MR1 AR3

AR4

CN2

CN1

MN2

MN1

MR1MN2

MN1AR2

->MR1

MN2's HA

Handover

AGR

� Figure 8. Hierarchical address management.

HoA CoA

MT1

MT2

Number 1

Number 2

Individual

Movingnetwork 1

Common










wireless communicationss engineer at Hellenic Air ForceGeneral Staff. He is currently a senior researcher at CCSR,UniS. He has been involved with several IST (FP5) projectsand is currently working in the IST MAGNET IntegratedProject. He is a patent holder, and has published morethan 30 papers at international journals and conferences.His research interests include wireless LANs, mobility man-agement, security and QoS frameworks for IP-based infras-tructures, and personal area networks.

TOSHIKANE ODA ([email protected]) has been withNippon Ericsson K.K., Tokyo, Japan, since 2000, where he iscurrently manager of the Market Support and New Technol-ogy Department. He is vice-chairman of the Working Groupon IP-Based IMT Platform in TTC. He worked in R&D on net-work design and services in KDD, and was also involved instandardization, especially in International Telecommunica-tion Union — Telecommunication Standardization SectorStudy Group 2 (ITU-T SG2). He received his M.E. and Ph.D incommunication engineering from Osaka University, and anM.S. in computer science from Columbia University.

SUDHIR DIXIT ([email protected]) has been a researchfellow at Nokia Research Center, Burlington, Massachusetts,since November 2003, and concentrates on research onnext-generation networks. Prior to this, he was a seniorresearch manager from 1996 to October 2003, and man-aged the Mobile Internet Performance group and its earlierincarnations, specializing in pervasive communications, con-tent networking, and optical networking. Before joiningNokia, he held various engineering and management posi-tions at other major companies, including NYNEX (now Ver-izon), GTE, Motorola, Wang, Harris, and STL (now NortelEurope Labs). These are on wireless IP, IP over WDM, andcontent networking in the mobile Internet. He received aB.E. degree from MANIT, Bhopal, India, an M.E. degreefrom BITS, Pilani, India, and a Ph.D. degree from the Univer-sity of Strathclyde, Glasgow, Scotland, all in electrical engi-neering. He also received an M.B.A. degree from FloridaInstitute of Technology, Melbourne.

ANDREAS SCHIEDER ([email protected]) receivedhis M.Sc. degree and Ph.D. in electrical engineering fromthe Technical University of Aachen in 1995 and 2003,respectively. During his occupational career within EricssonResearch he was involved in research on UMTS and GPRSperformance evaluation. He concentrated on the supportof real-time applications and actively contributed to thestandardization of the 3G radio access network GERAN. Heis a WG vice chair of the WWRF.

HONG-YON LACH ([email protected]) received aB.S. degree in computer science, with a minor in economics,from the American University of Paris, France, in 1987. Hehas extensive experience in R&D of OSI and IP communica-tion protocols and applications. In particular, he was a majorarchitect of and contributor to ETSI Project HIPERLAN (Type1) to design and specify the MAC and CAC services and pro-tocols, and has been contributing to IETF. He has also active-ly participated in various EU projects in the ESPRIT, ACTS,and FP5 programs. In May 1997 he joined Motorola Labs —Paris, France. Currently, he is manager of the Edge Network-ing Research Laboratory (ENRL), which researches, develops,and demonstrates innovative technologies and software toenable IP-based mobile multiparty multimedia networking,especially in the context of B3G systems. He also helps leadMotorola Labs’ overall Internet research activities, both tech-nically and strategically, in Motorola Labs Global Centre forNetworks and Systems Research.

MICHAEL I. SMIRNOV ([email protected]) gradu-ated with honors from St. Petersburg Electrotechnical Uni-versity in 1977, got his Ph.D. (computer science) in 1984,and a degree of associate professor in 1989; in 1992 hewas appointed deputy director (research) of IMICS ResearchInstitute in St. Petersburg, Russia. Since 1994 he is withGMD (now Fraunhofer) FOKUS (Research Institute for OpenCommunication Systems) in Berlin as a researcher and pro-ject manager (1994-96), head of competence centre(1999–2001), Direktorium member (since 2001) responsiblefor Internet research, standardization activities, and links toacademia. He has published over 50 technical papers inRussian, three in German, 38 in English; and has co-editedsix books, two journals (special issues), and five IETF drafts.His contributions include routing research, performanceanalysis, implementation of IP QoS multicast over ATM;distributed computing algorithms and protocols; systemsresearch in networking; QoS; scalability analysis, advanced

Internet services, policy-based control and programmabili-ty, and autonomic communication. He is serving the net-working community as COST263 chair (1999–2003);E-NEXT PMC member, as chair of international conferences(1986, 1992, 1995, 2000, 2001-2004); as EU Commissionexpert (high-performance networking, future and emergingtechnologies). Since 1997 he is an adjunct professor at TUBerlin in advanced Internet services.

SAMI USKELA ([email protected]) received his M.Sc.degree in electrical engineering from Helsinki University ofTechnology in 1999. He has been with Nokia since 1997.He has been involved in research on network architecturesand service platforms ever since. He has been activelyinvolved in standardization of the 3G systems: during 1998he contributed to network architecture and service plat-form standardization in TTC (Japan) and during 1999–2000to all-IP standardization in 3GPP. He has also been active inWWRF; he served as editor of WWRF white paper on coop-erative networks during 2002–2003 and contributed active-ly to the discussions in the forum. Since May 2001 he hasbeen responsible for B3G network research within NokiaNetworks, currently working as senior research manager.He has published several articles in international confer-ences and journals, and holds over 20 patents and patentapplications covering different areas of network architec-tures and service platforms.

RAHIM TAFAZOLLI ([email protected]) is the head of theMobile Communications Research Group in CCSR, UniS. Hisresearch activities are on optimization techniques formobile multimedia networks, mainly on advanced resourcemanagement, mobility management, and media accesscontrol. He has published more than 300 research papersin refereed journals and international conferences, and asan invited speaker. He currently has more than 15 patentsin the field of mobile communications. He is advisor andconsultant to a number of mobile companies. He is thefounder and past chairman of the International Conferenceon 3G Mobile Technologies. Regulations on InformationTechnology and Telecommunications, a member of theWWRF Vision Committee, and past Chairman of New Tech-nologies group of WWRF, and academic coordinator of theU.K. Mobile Virtual Centre of Excellence.

� Figure 9. IP2 architecture for a moving network.

MN1 -> IPra1MN2 -> IPra2 -> MR1

MR1

AR1MR1 AR3

AR4

CN2

CN1

MT2

MT1

MR1MT2

MT1AR2

IP2 core network

Handoff

MT1-> MR1

MT2

AGR

� Figure 10. Categories of routing mechanisms.

IPha IPhaIPra

IPha IPhaIPra

Network control sublayer

IP2 core network


Cooperative networks for the future wireless world

Documents

Transcript of Cooperative networks for the future wireless world