Heterogeneous Services and Architectures forNext-Generation Wireless Networks

Simone Frattasil, Student Member, IEEE, Rasmus L. Olsen1, Mauro De Sanctis2,Frank H.P. Fitzek1, Member, IEEE, and Ramjee Prasad1, SeniorMember, IEEE

'Center for TeleInFrastruktur (CTIF), Aalborg University, Niels Jernes Vej 12, 9220 Aalborg, Denmark{sfIrlolff|prasad}@kom. aau. dk

2Dpt. of Electronic Engineering, University of Rome "Tor Vergata", Via del Politecnico 1, 00133, Rome, Italymauro.de.sanctis@uniroma2.it

Abstract- Even though it is named as the successor of theprevious cellular generations, the Fourth Generation of WirelessMobile Communication Systems (4G) is predominantly presentedas a convergence platform that will provide clear advantages interms of coverage, bandwidth, energy consumption and a varietyof heterogeneous services, ranging from pop-up advertisementsto location-based and interactive or on-demand ones (so calledIP datacasting). Based on the above vision of 4G, in this paper wepropose innovative services and architectures that may achieveenhanced security, energy efficiency, and just-for-you and just-in-time content delivery. Furthermore, since the state-of-the-arttechnologies do not appropriately cover the needs required for4G, a context sensitive service discovery that can operate in awireless as well as in a wired network environment, ranging fromlocal to global distances, is described and applied to the case ofthe proposed services.

Index Terms- 4G, Cellular networks, Context management,Context sensitive service discovery, Convergence, DAB, DVB,Heterogeneous services, Next-Generation Networks, Short-rangenetworks, WLAN, WPAN.

I. INTRODUCTIONAt the turn of the last century, telecommunications' fo-

cus changed considerably: from traditional wired telephony-oriented services and infrastructures to data-based services;from homogeneous to heterogeneous networks; from non-intelligent devices to smart handhelds, personal digital as-sistants, and mobile computers. To enable the growth in theseexpanding markets, we need to think about their future, theirtechnologies, and their markets' convergence. A key questionwhen we discuss the evolution of hardware and softwaretechnology in the telecommunications and computing fieldis: What are the targets of mobile Next-Generation Networks(NGNs)? One target could be computing anytime, anywhere,and another could be personalized mobile communications -that is, ubiquitous computing [1]. The primary goal of ubiqui-tous computing is to embed many small and highly specializeddevices within the everyday environment, so that they operateseamlessly and become transparent to the person using them,either offline or online. Ubiquitous computing products aimto be everywhere (for example, by being portable); small; andaware of their environments, users, and contexts. Productsand devices embodying these characteristics will provide aphysical entity with complete freedom of movement andfreedom of interaction.0-7803-9206-X/05/$20.00 ©2005 IEEE

213

Following the paradigm of generational changes, it wasoriginally expected that the Fourth Generation of WirelessMobile Communication Systems (4G) would follow sequen-tially after the Third Generation (3G) and emerge between2010 and 2015 as an ultra-high speed broadband wirelessnetwork [2]. In Asia, for example, the Japanese operator NTTDoCoMo introduced the concept of Mobile multimedia; Any-time, anywhere, anyone; Global mobility support; Integratedwireless solution; and Customized personal service (MAGIC)for defining 4G [3], which mainly concentrates on publicsystems and envisions 4G as the extension of 3G cellularservice. This view is hence referred to as the linear 4G visionand, in essence, focuses on a future 4G network that willgenerally have a cellular structure and will provide very highdata rates (exceeding 100 Mbit/s).

Although it is named as the successor of the previousgenerations, 4G is not limited to cellular systems and thus hasnot to be exclusively understood as a linear extension of 3G[4]. In Europe, for example, the European Commission (EC)envisions that 4G will ensure seamless service provisioningacross a multitude of wireless systems and networks, fromprivate to public, from indoor to wide area, and provide anoptimum delivery via the most appropriate (i.e., efficient)network available. From the service point of view, it foreseesthat 4G will be mainly focused on personalized services [5].This view is referred to as the concurrent 4G vision andemphasizes the heterogeneity and integration of networks andnew service infrastructures, rather than increased bandwidth'per se'. Therefore, along with this view, a large variety ofheterogeneous services, ranging from pop-up advertisementsto location-based and interactive or on-demand ones (so calledIP datacasting), will be ultimately available to the user [4].

Based on the above vision of 4G, in this paper we pro-pose innovative services and architectures that may achieveenhanced security, energy efficiency, and just-for-you and just-in-time content delivery. Furthermore, since the state-of-the-art technologies [6]-[9] do not appropriately cover the needsrequired for NGNs, a context sensitive service discovery thatcan operate in a wireless as well as in a wired network envi-ronment, ranging from local to global distances, is described[10] and applied to the case of the proposed services.The rest of the paper is structured as follows: Section

II describes the new services and architectures; Section III

illustrates the context sensitive service discovery and its ap-plication. Finally, the concluding remarks are given in SectionIV.

II. PROPOSED SERVICES AND ARCHITECTURESA. Enhanced Services

In Figure 1 a possible scenario for configuration setup isshown. In particular, the two sketches presented illustrate twodifferent interactions very familiar with the Wireless PersonalArea Network (WPAN) environment:

. Person-to-Person (P2P) communication. If two WirelessStations (WSs), such as WS1 and WS2 in Figure 1,want to intercommunicate using their short-range com-munication capabilities (solid lines for data transmission),in order to enable connection set up, the Base Station(BS) may perform the initial configuration (dashed lines),including authentication and security (exchanging validkeys).

. Person-to-Machine (P2M) communication. Whenever acustomer is interested in using a machine (e.g., a printer,a screen, etc.), which is controlled by the BS, as forWS3 in Figure 1, the cellular network can provide theconfiguration of the customer's terminal (WS4), such asconfiguration of drivers, etc. A high security can be alsoguaranteed by distributing security keys.

The proposed service brings the following major advan-tages: 1) It offers a trusted authentication that can be used asa Public Key Infrastructure (PKI) and, consequently, assuresa higher security respect to the current one in short-rangead-hoc networks (e.g., Bluetooth-based); 2) It makes use ofunlicensed bands, thus the Service Providers (SPs) can reducethe expenses for the service provision and, at the same time,charge the peer-to-peer communication.

B. Cooperative Services1) Resource Sharing: One of the key challenges in wireless

personal communication is the efficient exploitation of theenergy stored by the batteries of portable devices, namelythe energy efficiency. It is worth noting that the design ofa low power transmitter is different from the design of anenergy efficient transmitter. When the final purpose is toincrease the battery lifetime, the most suitable performancemetric is the energy efficiency. The latter can be achievedover different protocol layers exploiting low power supply andavoiding the need for frequent batteries recharge [11] [12].However, if we start thinking about cooperative scenarios,novel exploitation methods of the resources, able to decreasethe energy consumption, arise. Personal communication sys-tems are characterized by radio-related resources and device-related resources, where the former are transmission powerand bandwidth, and the latter are complexity and energystored by the batteries. In cooperative scenarios, besides thebandwidth sharing, different devices owned by the same ordifferent users can share their device-related resources, thusdecreasing the energy consumption of the overall system.An example of cooperative service is the one depicted in

Figure 2, where one specific terminal WS1 in combination

with the BS gives network access to the other terminals inshort-range (WS2, WS3, and WS4), even if they are notequipped with the appropriate technology to connect directlyto the BS'. WS1 hence acts as a bridge/router for the other ter-minals, offering secure access. In general, the master/gatewaycan be chosen in terms of hardware capabilities (e.g., the mostpowerful terminal within the group) and it cannot act selfishly.As a rule, we suppose that in order to belong to a group,each terminal has to previously agree to behave according toa specific cooperative protocol. However, the group is set uponly in case of a benefit for all its components. Therefore, wecan foresee that the cooperation with the surrounding terminalswill permit to the master/gateway to obtain, for example, acertain service at a lower price, because this will be sharedamong all the users in the group. WS2, WS3 and WS4 canhence connect to the Intemet with the advantage of drasticallydecreasing their power consumption due to the use of a powersaving technology, at the price of a slight increase of the powerconsumption for WS,.

2) Synchronization: Assuming a certain number of ter-minals in the range of the same BS and served by thesame unicast video flow2, due to channel conditions (maybeeven for different codecs, etc.) they will all have differ-ent/unsynchronized play-out times. For a group of userslocated close together, this can lead to undesired situations:if they are watching a football match, for example, the audiomay be heard with many echoes from the neighbors, or someinformation, such as a goal, may arrive earlier to some ofthe users in the group. To synchronize the terminals, twodifferent approaches are possible: 1) A global synchronizationfor the whole cell; 2) A local synchronization among relatedternminals. Since the first option may lead to huge play-outvalues as all the terminals have to tune their play-processto the worst terminal in the cell, we advocate the usage oflocal synchronization. The latter imposes the tuning to theworst terminal only in the context of the same group; thus, thebehaviour of one group does not interfere with the behaviourof another one.

Local synchronization can be achieved by using short-rangecommunications within the same group. The latter will beformed either by detecting whether neighboring terminals areusing the same stream or by receiving this information fromthe cellular system. For instance, two WSs detect that theybelong to the same group and want to synchronize, theydecide on a master in a timely fashion and the synchronizationprocess can start [13] [14]. The same approach is also usablein case multi-terminals receive the same multicast flow. Infact, due to different play-out values or different time pointswhen a streaming has been started, the possibility to beunsynchronized is still valid. As an example, in Figure 3 weassume that the BS transmits a sequence of video frames(violet color) with the first packet three times larger than

Just the possession of the Subscriber Identity Module (SIM) card mayallow each user to be charged separately by the BS, without any extra effortfrom WS1.

2The use of unicast seems reasonable if different video formats, encodingstandards, SPs, etc. are being used, even though the terminals are receivingthe same content.

214

.S ' \i '-

....A ".....

- - --'-'- -- -- -- -- -- -- - - -- -- -- -- --

Fig. 1. Configuration of Communication Pairs for P2P and P2M.

the following ones, which all have the same size3 and aretransmitted in 40 ms each. We further hypothesize that ClientI (Cl) starts buffering packets for a buffer value equal to thesize of four small packets. Therefore, after 160 ms, he willstart playing out the first video frames. The next ones willbe instead played out every 40 ms. In case that now Client 2(C2) decides to watch the same video content, when Cl playsout the first frame, C2 will have buffered the same amount ofdata: from Packet 3 up to Packet 6. After buffering Packet 6, hewill start also the play-out with Packet 3. At this time, C I willplay-out Packet 5 and thus the two streams are asynchronousby two packets, leading to a delay of 80 ms. Larger play-outvalues lead to a larger gap between the streams.

13101b*XCl =OSUtph.acl,2 TOUcA2 1 mSa2

J-u-rnFig. 3. Asynchronous Delivery of a Multicast Video Streaming.

Although we referred to audio and video, the proposedservice infrastructure can be applied in a more general fashion;for instance, gaming is another concrete example where itcould be helpful.

C. IP Datacasting ServicesWith this kind of services we propose to start up IP

datacasting delivery allowing the download of multimedia

31n Variable Bit Rate (VBR) encoded video all frames would be of differentsize, but for illustration purpose we have chosen the given packet sizes.

Fig. 2. Gateway Functionality.

content, such as video clips, MP3s, etc., onto the user terminalwithout the awareness of the user himself. This means thatthe device will filter all the incoming information accordingto its technical features (e.g., screen size, storage memory,etc.) and the user's profile (the latter can be updated by theuser whenever he wants, in order to select only the neededinformation). When the file has been completely stored, twokinds of interactions may be possible:

. Device-to-User (D2U) interaction. A message informiingthe user of the availability of a certain content ready to beplayed is popped-up. The advantage of this option relieson the fact that the unpleasant information is immediatelydeleted from the temporary folder of the user terminal,granting space for other incoming content.

* User-to-Device (U2D) interaction. In a certain moment,the user himself will ask the network for a specific contentalready present in his terninal. Therefore, the SP willsupply him with the decryption keys4, charging for therequested service.

Examples of the various methods to deliver IP datacastingcontent according to the proposed approach are listed below:

* Digital Audio/Video Broadcasting (DAB/DVB) delivery.Since DAB/DVB technologies both offer possibilities tobroadcast or multicast free-to-air content and applicationsintegrated with a cellular back-channel, it is natural toapply the proposed approach. In particular, the implicitinteractivity, the high security level, and the billing sys-tem associated to the cellular network are exploited.

* WPAN/WLAN delivery. Since WPAN/WLAN technolo-gies both offer high bandwidth connectivity in hot spotareas, this feature can be exploited to deliver and uploadIP datacasting content onto the user terminal. In this case,as the time spent for a user passing through a hot spotis very short, we have to highlight the importance of theauthentication process performed by the cellular networkin order to avoid a quite laborious procedure usuallycarried out by the local Access Point (AP). Furthermore,

4For increased security the decryption keys may be dynamically changedfor each content, thus avoiding any unauthorized use by the user himself.

215

in this delivery scenario the cellular network may bealso in charge of the further task of transferring all theavailable information onto the hot spot server. Since thenumber of users may increase from time to time, thiswill avoid the continuous transfer of the possible neededcontent from the SP network onto the hot spot server. Inthis way, the overall procedure is much more efficient interms of time and amount of transferred information, thusmaking very fruitful even a short time passage in a hotspot.Cellular delivery. Even though in the previous bullets wehave referred to specific convergence scenarios, the sameapproach may be also applied in case the user is only inthe range of a BS. In this situation the cellular networkis using its resources in order to deliver the content. Inparticular, the idea is to exploit the periods in which thenetwork is not very congested (e.g., during the night).

Examples of services based on the proposed approach,where each delivery method may be used indiscriminately, arethe following:

. If we specifically refer to the delivery of content soonavailable, such as new video clips, sport highlights,new movie trailers, etc., the user may book the servicein advance in order to have the content immediatelyavailable on his terminal. Moreover, the sooner he willapply to receive such kind of content the lower priceshe will experience. Nevertheless, some information canalso be charged using a per-time subscription, whichimplies that the SP will provide the decryption keys fora series of services that the user is allowed to receive fora certain negotiated period (e.g., subscription for news,sport videos, etc.). The user profile will be then upgradedand the user terminal will mainly receive the pre-paidcontent in the exact moment in which they are available.Furthermore, the network will not be overload by theauthentication process.

. Supposing that we are using the Multiple DescriptionCoding (MDC), the multimedia content will be dividedin N descriptors (each descriptor contains a partial in-formation regarding the content itself) and transmittedon the air interface by the cellular BS. Since the userterminal could differ in characteristics such as screensize and/or processing power, the devices with a lowprocessing power and a small screen size (e.g., wearableand mobile) will use only M<<N descriptors, whereasthe ones with high processing power and large screensize (e.g., Personal Digital Assistant (PDA), Web Tablet,and laptop) will need M<=N descriptors. The number ofdescriptors received by the user terminal will be set to L,with L<M, and the user will be able to exploit the serviceat a reduced quality for free. After having a look at thecontent, if he will be really eager to watch it at a highquality, he will receive the remaining descriptors, thusbeing charged for the service. In particular, the delivery ofthe remaining descriptors may be performed through theexploitation of all the networks available (see Figure 4).

* The SP can decide to send free of charge partial decryp-

tion keys to selected users for promotional purposes. Thepartial decryption keys will open the content (e.g., anMP3, a video, a book) only for a part of it. The user willhence enjoy the first chapter of a book or the first part ofa video or an MP3 and then decide to buy it or not.

Fig. 4. IP Datacasting Delivery Scenario.

The SPs will take advantage from the proposed architecture(see Figure 5) as they will charge their customers for the newservices. In particular, both high data rate and low data ratecontent with frequent updating (e.g., news, cinemas, transportschedules, horoscopes, sport results, etc.) may be delivered.Therefore, even a terminal with low on-board memory mayexploit these services. On the other hand, the customers willexperience a zero-time download with a clear advantage interms of time effectiveness and Quality of Service (QoS). Avideo clip, for example, already being stored onto the userterminal may be played at the maximum allowed resolution,thus avoiding any lowering of the performance due to thestreaming.

III. CONTEXT SENSITIVE SERVICE DISCOVERYTo achieve true context sensitive service discovery is crucial

to realize the personalization and user centricity of the forth-coming generation. What is the benefit of high bandwidth,seamless connectivity, high security, easy sharing of resources,etc., if the user is being annoyed by unavailable, far away,expensive services that do not really fit his current needs?If a successful context sensitive service discovery can beperformed, then the user's experience of the system as an intel-ligent entity can be achieved. Furthermore, the psychologicaleffect that the system 'knows what I mean' and consequentlyreacts would be of fundamental importance.Two main entities are defined in the model of context sensi-

tive service discovery, a Context Manager (CM) and a ServiceManager (SM). The CM manages local context information,i.e., file access to user profile, interacts and gets measure-ments from surrounding sensors and/or sensor networks, andfinally monitors the resources of the overall network. TheSM performs service discovery according to a hierarchicalapproach: starting from the smallest Radio Domain (e.g., a

216

g.nd: Ai Ceilulat Air bowf6c; AA AuzhaialoAb hslato

Accoiznt AS: Appicateon Sewt CCN: Ce&dAr Con Nust D,DlgM Smsd=Wft Air b*eot O& NOW Bmdctlng;DC"M:D%aW Bwa*aain CQ. NeahW LR, L*" Rargq Air 1ntec.

Fig. 5. Service Network Architecture for IP Datacasting Delivery.

piconet) forwards its request to the world outside the PAN, i.e.,to the global level, where it will interact with other ContextSensitive Service Discovery (CSSD) nodes by using whateverconnection is available. As the CM will be a registered service,context information will be able to use service discovery toalso discover context in foreign domains. Within this approach,issues about scalability and security are inherited from theused service discovery protocol. In the SM entity, a contextsensitive engine is interacting with the CM. The purpose of thisengine is to rank services discovered according to local andforeign service context information. This requires additionaloverhead of information to be exchanged between service andcustomer, but the gain is a more intelligent behaviour of theservice discovery.

Context sensitive service discovery has many useful aspectsconcerning the proposed services:

. In case of the enhanced services described in Section II-

A, let us imagine that the user in Figure 1 has found100 printers in a network. He will not be able to knowwhich one is the most suitable for him. One might bebusy, another one too far away, and so on. Ranking theprinters according to the preferences of the user (e.g.,how much he is willing to pay, how far away the printeris, how long is the queue, how many colors are available,etc.) is just one of the many challenges lying ahead forresearch in context aware service discovery.

. In case of the synchronization services described inSection II-B.2, CSSD could be used to locate an appro-

priate video or streaming server, based on the requests,preferences (e.g., language) or location of the user.

. In case of the IP datacasting services described in SectionII-C, CSSD could be used to find the appropriate service,

whether it should use a DAB/DVB, WPAN/WLAN orCellular delivery. Here the choice could be weightedamong cost of usage, current QoS parameters for thedifferent network services, user preferences, etc.

IV. CONCLUSIONThe research of new and rich services has to be addressed

with the aim at driving the development of technologicalsolutions for the forthcoming generation of wireless mobilecommunication systems. Based on the concurrent 4G vision,in this paper we have proposed innovative services and archi-tectures that may achieve enhanced security, energy efficientcommunications, and just-for-you and just-in-time contentdelivery. Furthermore, since the state-of-the-art technologiesdo not appropriately cover the needs required for 4G, a contextsensitive service discovery that can operate in a wireless aswell as in a wired network environment, ranging from localto global distances, has been described and applied to the caseof the proposed services.

V. ACKNOWLEDGEMENTS

This work has been supported by Samsung Electronics, Co.,LTD, Korea.

REFERENCES

[1 IJ.E Huber, "Mobile Next-Generation Networks", Multimedia, IEEE, vol.11, no. 1, pp. 72-83, Jan-March, 2004.

[2] E. Bohlin, S. Lindmark, J. Bjrkdahl, A. Weber, B. Wingert, P. Ballon,"The Future of Mobile Communications in the EU: Assessing thePotential of 4G", ESTO Publications, February, 2004.

[3] K. Murota, NTr DoCoMo, "Mobile Communications Trends in Japanand DoCoMo's Activities Towards 21st Century", in Proceedings of the4th ACTS Mobile Communications Summit (AMOS), Sorrento, Italy,June 8-11, 1999.

[4] S. Frattasi, H. Fathi, F. Fitzek, Kiho Chung, R. Prasad, "4G: A User-Centric System", Mobile e-Conference (Me), August, 2004.

[5] J.M. Pereira, "Fourth Generation: Now, it is Personal", in Proceedingsof the I Ith Intemational Symposium on Personal, Indoor and MobileRadio Communications (PIMRC), IEEE, London, UK, September 18-21, 2000.

[6] E. Guttman, C. Perkins, J. Veizades, M. Day, "Service Location Protocol,Version 2", Intemet Engineering Task Force (IETF), RFC 2608, June,1999 (http://www.rfc-editororglrfc/rfc2608.txt).

[7] Salutation Consortium, "Salutation Architecture Specification (Part-l),Version 2.0c", June, 1999, (http://www.salutation.org).

[8] Sun Microsystems Inc., "Jini Network Technology"(http://wwws.sun.com/software/jini).

[9] UPnP Forum, "Universal Plug'n'Play" (http://www.upnp.org).[10] R.L. Olsen, H. Murakemi, H.P. Schwefel, R. Prasad, "User Centric

Service Discovery in Personal Networks", in Proceedings of the 7thIntemational Symposium on Wireless Personal Multimedia Communi-cations (WPMC), Abano Terme, Italy, September 12-15, 2004.

[11] C.E. Jones, K.M. Sivalingam, P. Agrawal, J.C. Chen, "A Survey ofEnergy Efficient Network Protocols for Wireless Networks", WirelessNetworks, Association for Computing Machinery (ACM), vol. 7, no. 4,pp. 343-358, August, 2001.

[12] W. Stark, H. Wang, A. Worthen, S. Lafortune, D. Teneketzis, "Low-Energy Wireless Communication Network Design", Wireless Commu-nications, IEEE, vol. 9, no. 4, pp. 60-72, August, 2002.

[13] P. Popovski, F.H.P. Fitzek, H. Yomo, T. Madsen, R. Prasad, "MAC-layer Approach for Cluster-Based Aggregation in Sensor Networks",Intemational Workshop on Wireless Ad-hoc Networks (IWWAN), June,2004.

[14] P. Popovski, T. Kozlova, L. Gavrilovska, R. Prasad, "Device Discoveryin Short-Range Wireless Ad Hoc Networks", in Proceedings of the 5thIntemational Symposium on Wireless Personal Multimedia Communi-cations (WPMC), Honolulu, Hawaii, October 27-30, 2002.

217