PAVING THE ROAD TO SYSTEMS BEYOND 3G - THE IST MIND PROJECT Dave Wisely1 and Enric Mitjana2 1
BT Labs, Adastral Park, Martlesham Heath, Ipswich IP5 3RE, United Kingdom Phone: ++ (44) 1473 – 643848 Fax: ++ (44) 1473 – 646885 Email:
[email protected] 2 Siemens AG, ICM N PG SP RC PR, Hofmannstrasse 51, D-81359 Munich, Germany Phone: +49 89 722 46118 Fax: +49 89 722 44958 Email:
[email protected] Abstract - In systems beyond 3G mobile access will be considerably more diverse than is currently the case with highly integrated cellular technologies. Complemented with new access technologies, such as Wireless LANs, future mobile systems will offer a different mix of bandwidth, cost and coverage. There will also be extensions to traditional wireless access networks – both by operators with wireless routers and by users forming ad-hoc networks that attach at the fringe of the access network. Creating services and delivering multimedia over these networks is a very real technical challenge. In this paper the overall approach of the IST MIND project to tackling these issues is described. Further the rationale for an IP-based architecture and how it is being extended to encompass new access concepts of wireless and mobile routers is detailed. Finally the paper looks at how we are enhancing WLAN Systems to give greatly increased performance for systems beyond 3G. I. INTRODUCTION Many European operators have, or plan to have in the very near future, operational 2.5G systems such as GPRS. This is seen as an intermediate step in the introduction of 3G coverage – giving users higher rate data transfer (average around 64kbit/s), per packet charging and an always on capability. The first 3G deployments offer higher data rates, on demand bandwidth and better QoS options [1]. At the same time we are seeing a tremendous interest in Wireless LANs. Individuals are using them to extend their DSL connections; organisations are using them to allow flexible access to their Intranets and operators are planning to offer broadband Internet and Corporate access from hot spots like cafes and railway stations . The advent of the complement of traditional mobile networks via the use of WLANs and other, non cellular, technologies is leading to the concept of systems beyond 3G[2]. There is not yet an industry consensus on what “systems beyond 3G” will look like but ideas and concepts often cited include [3] [4] [5]: • Transition to an all-IP network • Ad-hoc subnetworks as part of future networks • Addition of micro-cellular technologies, such as Wireless LANs (WLANs)
0-7803-7589-0/02/$17.00 ©2002 IEEE
• •
•
Software Radio allowing a single terminal to connect to multiple access technologies Multi-homed terminals capable of vertical handover between different access technologies (e.g., cellular to WLAN) Fixed-mobile convergence whereby users access the same services and applications but these are adapted to their current terminal and/or location
There are some very key issues that need to be addressed before such systems beyond 3G become a reality. At the very top is the business case and value chain, which will contain additional players relating to the new scenarios that these technologies will give rise to. In an ad-hoc fringe users will only forward packets in return for something else – either monetary payment or forwarding of their own packets at a later time. Users will have a greater choice of access technologies, offering different bandwidths, QoS, costs and so on. However, delivering services, seamlessly, over these different access mechanisms will require much greater functionality in the end terminals, more complicated security technologies and support for personal mobility. The new IP networks that support WLAN coverage will need QoS and mobility functionality – for hand-overs and real-time service support. That functionality will need to interface with the ad-hoc network fringe where QoS and mobility is no longer under the direct control of the operator. Finally, WLANs do not currently have the range or capabilities (eg maximum speed, QoS performance..) to support all the situations where users will want access to them. Enhancements are needed to extend this performance in many areas. The MIND project is addressing these concepts and working towards a complete architectural (and service) solution for systems beyond 3G. In this paper we describe our thinking in a top-down manner – through the scenarios, network design, WLAN enhancements to practical trials. II. THE MIND PROJECT The MIND (Mobile IP based Network Developments) [6] project is partially funded by the European Commission in the frame of the Information Society Technologies (IST) Programme and brings together major players in the mobile
PIMRC 2002
Stephanie and her children drive from the suburbs to the town centre and on the way set up an ad-hoc network to play a game. During a traffic jam the network grows and several vehicles connect using Hiperlan/2 links. As the jam clears the network partitions but, since the game is very exciting, the various players decide to switch it to UMTS. Stephanie’s terminal is used as a gateway, via its UMTS connection, to a game server. Stephanie’s children connect to Stephanie’s terminal via bluetooth links. The connection to the game server is actually free but playing requires the children to use pre-pay vouchers. This scenario is shown in figure 2. Multihoming devices: shortrange and longrange Game Server
Billing
AAL2 Switching Cloud
Vertical Handover
AAA
SIP Proxy Server
e (U rang Long
MTS
.)
Longrange (UMTS, GPRS, etc.)
Internet
S pe hor rfo tra rm ng an e h c e ig lin h k
Player
h hig k ge lin an c e rtr an o m Sh rfor pe high ange Shortr
Shortrange high performance link
k ce lin rman perfo
hp hig
Player
rtra nge
, etc.)
Simple device: no multihoming and only shortrange
S, GPRS
MMR / Multihoming device
erfo rma nc e
link
Mobile Communication Network
Longra nge (UMT
MIND is scheduled to run from June 2001 to November 2002. The overall aim is to facilitate the rapid creation of broadband multimedia services and applications that are fully supported and customised when accessed by future mobile users from a wide range of wireless access technologies. MIND is the follow up of the IST project BRAIN (Broadband Radio Access for IP based Networks) [7] and will extend, as well as practically demonstrate, BRAIN concepts.
Player etc RS, , GP
S ho
domain working on a vision of “systems beyond 3G”. The project partners of MIND are drawn from three key areas: • Manufacturers: Ericsson Radio Systems AB (Sweden), Nokia Corporation (Finland), Siemens AG (Germany), Sony International (Europe) GmbH (Germany) and Infineon Technologies (Germany), • Network operators: British Telecommunications plc (UK), France Télécom S.A. (France), NTT DoCoMo, Inc. (Japan) and T-Systems Nova GmbH (Germany), and • SME and academia domain: Agora Systems S.A. (Spain), Universidad Politécnica de Madrid (Spain) and King‘s College London (UK).
This link is splitting apart when the two cars are no longer in proximity
Player
SP Server Farm
IGSN
Gateway
IP backbone DiffServ IP - macro mobility management
UMTS
Mobile User
Global Internet
Fig. 2.- Cooperative gaming from the leisure time scenario
Bluetooth Hiperlan/802.11
ADSL
IP-based local mobility ISSL QoS BRAIN Access Routers
MIND Wireless Routers
Hiperlan/802.11
MIND Mobile Routers
Hiperlan
Fig. 1.- The MIND architecture Figure 1 shows the MIND overall architecture. The technical approach takes, as a starting point, the concept of an IP core accessed by a variety of technologies with the radio access network being extended by an ad-hoc fringe. III. THE MIND SCENARIOS The project follows a top down approach starting with the definition of scenarios that need to be supported by future wireless networks. These scenarios then allow the project to identify the involved elements and players, and assign functionalities and to add any constraints upon them. The top down analysis permits the project to derive clear technical requirements to support this user centric vision and to describe the flows of services and money within them. As an example, the scenario called “Leisure Time” is described here:
The project also investigates new business models for “systems beyond 3G” and identifies the roles of the new players in the value chain. Novel architectures for flexible and dynamic service provision are also proposed within the work. IV. THE MIND TRIALS As part of MIND, a number of testbeds to demonstrate some of the key concepts developed within the predecessor BRAIN project are being built. These are best explained with respect to the BRAIN ENd Terminal Architecture (BRENTA) [8]. The BRAIN terminal and network are underpinned by two powerful interfaces, the IP2W (IP to Wireless) and the ESI (Extended Socket Interface). The IP2W serves to allow a common IP access network, with IP level mobility and QoS mechanisms, to make use of any underlying functionality that the particular layer 2 access technology can provide. The ESI enables application/session layer access to QoS-enhanced transport services, howsoever they are provided, via a generic interface. Figure 3 presents BRENTA and the afore mentioned interfaces.
Application Layer
Type A Session Legacy Applications Layer Type B Use SIP/RSVP ESI Interface Type C Use BRAIN Component API Type D IP2W Interface Use full BRAIN API
Type A
Type B
Type C Type D BROKER
BRAIN QoS BROKER GUI
•
MIND Mobile Routers (MMRs) – Nodes that act as routers through which mobile nodes may be connected to the network. They may be stationary but are considered lightweight and owned by end users meaning that they can also be mobile.
BRAIN High-level API
BRAIN Component Level API Session Layer Protocols(SIP, H323, RSVP..) QoS and Mobility Support
Transport Layer IP Layer with QoS and Mobility Support Link Layer with QoS MAC PHY
Fig.3.- BRAIN ENd Terminal Architecture (BRENTA) In the MIND trials aspects of the BRENTA application layer QoS control (the type A application with separate QoS control in Figure 3) will be tested. This will allow configuration of terminal interfaces and access to QoSenabled IP networks built using ISSLL (Integrated Services over Specific Link Layers – in this case over DiffServ). In addition mobility support to these IP networks will be added, with HMIP (Hierarchical Mobile IP) being compared with the BRAIN Candidate Mobility Protocol (BCMP) [9]. As much as possible of the trials will be conducted with IPv6 to allow assessment of the new mobility features afforded by this technology. The MIND trials will research how UMTS to WLAN vertical handover can be accomplished, concentrating on the network layer handover, often called the “ no coupling” approach. The project will study the packet delays and losses encountered, and look at techniques, such as Mobile IP, to keep sessions alive during such vertical handovers. Finally, the practical trials will examine how effective the enhancements to the Hiperlan/2 physical layer proposed in BRAIN are at increasing spectral efficiency. The testbeds are distributed all over Europe, from Helsinki to Madrid and from Berlin to Ipswich. It is envisaged that the experimentation software will be exchanged between testbeds and a high capacity IPv6 connection made between them. V. ROUTING, QoS AND MULTI-HOMING IN THE MIND NETWORK Within the MIND project the key goals of the network layer research are to consider how ad-hoc nodes can attach to an existing IP access network, how these ad-hoc nodes can conspire to deliver IP packets on anything other than a best effort basis and how users can obtain a secure, end-to-end, service under such circumstances. The MIND network extends the original BRAIN all-IP access network [10] by the addition of: • Access Network Wireless Routers (ANWRs) – routers that are: owned by the access network provider, mains-powered and portable but changing position only slowly. They lack the full capabilities of normal access routers and have only wireless interfaces.
The first consideration of the network research was to develop a technical scenario that, at the network level, encompassed the three high level MIND usage scenarios mentioned above. The resulting network scenario is shown in Figure 4. BMG
ANWR BAR ANWR BAR
ANWR MN
MMR
MN MMR
MMR
MMR
MN – Mobile Node MMR – MIND Mobile Router BAR – BRAIN Access Router BMG – BRAIN Mobility Gateway ANWR – Access Network Wireless Router
MN
Fig. 4.- MIND IP Access Network From this scenario the project has identified a number of key architectural issues: • Optimum routing, QoS and security protocols to use in the ad-hoc extension. • Multihop routing: where should it be carried out, at layer 3 or layer 2? • Handling of the untrusted parts of the network. • Handling of multicast within the ad-hoc extension. • The support and interface that the fixed network offers to the ad-hoc extension. The work done within the predecessor project BRAIN to provide micro-mobility and QoS [11] in the radio access network is being extended and completed to support these new topologies. Traditional Mobile Ad-hoc NETwork (MANET) protocol development has focussed on routing in stand alone networks. The MIND scenarios are more concerned with the ad-hoc network as an attachment to the fixed access network. Accordingly, there are certain requirements that are placed on the ad-hoc network such as the support for idle-mode terminals and paging, to reduce signalling and conserve battery power. Handovers are particularly challenging – for example, when the ad-hoc network connects to the fixed network via a single “ pivot” node, do
all the ad-hoc nodes have to hand-over to a new access router when the pivot node does? How is multicasting supported in the ad-hoc fringe? The project has identified two different technical options: firstly where multicast effectively ends at the fixed network boundary and, secondly, where the first hop multicast router is within the fringe and there are different mulicast protocols running in the ad-hoc and fixed network parts. Finally QoS support relying on nodes not under the direct control of the operator and user concerned will always be a challenge. The components of a QoS solution have been identified: signalling, admission control, priority queuing at the routers, policy enforcement, metering and traffic shaping. Currently a QoS solution that offers these functions within the ad-hoc fringe and that is robust, scalable and efficient and interfaces with our IP access network design is being worked out. As an example of the work of MIND we have analysed 20 MANET routing protocols for consideration as components of a routing solution in the ad-hoc fringe of our architecture. Fig.5 shows a classification of these protocols. Reactive
TORA
extended to a mobile connected to two different access networks – possibly because they offer different characteristics such as cost and reliability. Within the project the implications of these scenarios for mobility, QoS and service delivery are being worked out. VI. HIPERLAN/2 – A CANDIDATE BROADBAND AIR INTERFACE MIND has chosen the WLAN standard Hiperlan/2 as candidate for its broadband radio interface. Other interfaces such as IEEE802.11 and Bluetooth can be adapted using the same methodology, but MIND has opted to specifically enhance Hiperlan/2 for new networking scenarios and take those enhancements to standardisation (ETSI BRAN [13]). IP Enhancement to support IP2W Interface
IP2W Interface
CL Control SAP
CL User SAP
Convergence Layer (CL) Service Specific Convergence Sublayer (SSCS)
C ontrol Plane
User Plane
C ontrol Plane
User Plane
DSDV Common Part (CP)
DSR
AODV
ZRP
Hybrid
Common Part Convergence Sublayer (CPCS) Transparent to control messages
Segmentation And Reassembly (SAR)
DLC Control SAP
Data Link Control (DLC) layer
FSR
DLC User SAP
C ontrol Plane
User Plane
H/2 Convergence Layer(s) Proactive
OLSR
TBRPF-FT
Poorer Better Source’ s knowledge about how to get to destination
Fig. 5 Classification of ad-hoc routing protocols From this analysis the following was concluded: • Mesh protocols tend to operate better than treebased ones • Routing loops severely limit performance • On-demand approaches are better than pro-active ones • There is a need for link repair mechanisms • QoS with standard protocols is hard to accomplish. MIND has also extended the traditional IETF concept of multi-homing to include new scenarios such as a terminal connected to two different radio access networks – possibly UMTS for uplink and WLAN for downlink, say. The possibility that terminals might connect to the access network through two access routers (from the ad-hoc fringe) has also been considered. How QoS might be improved by bicasting packets from some point in the access network is being looked at. This concept of multihoming can be further
ATM Cell based
1394 Ethernet IP Packet based
3G Interworking
Fig.6 Overview of the convergence layers supported by Hiperlan/2 and detail of the IP In order to fulfil the requirements on QoS, mobility, and IP support dictated by upper layers, the MIND air interface includes a series of enhancements with respect to the reference Hiperlan/2 system at the convergence layer as well as at the DLC and the physical layer. Specifically the MIND-enhanced Hiperlan/2 system supports: • Efficient transport of IP packets for all multimedia applications • A QoS service to the IP network layer • Network layer mobility management protocols (e.g., by providing paging) • Handover of users to other BRAIN Access Routers (horizontal handover) as well as to non-BRAIN networks (vertical handover) with minimum delay and/or loss of packets • Unicast, multicast and broadcast services • A transparent service to the IP layer
In particular MIND has detailed a service specific convergence sub-layer for IP to support these functions, see Figure 6. MIND is also looking at the layer 1 and layer 2 issues that the new scenarios bring, including vehicle to vehicle (mentioned in the described “ Leisure Time” usage scenario) and meshed scenarios (point to point and point to multipoint). This requires addressing topics such as: • Multihop with centralised/distributed controller (MAC, RLC, Scheduling) • Radio propagation models (e.g., outdoor) • Multipoint-to-multipoint MAC protocol • Addressing, link adaptation and QoS scheduling 0
Packet Error rate
10
ETSI Channel C; burst length = 1.476 ms; BPSK 1/2
-1
10
-2
10
-5
3 km/h 20 km/h 30 km/h 40 km/h 0
5
C / N [dB]
10
15
20
Fig. 7.- Hiperlan/2 performance degradation for increasing speeds Since basic Hiperlan/2 quickly degrades as mobile speed rises (see Figure 7) MIND is looking at techniques to increase the maximum speed performance. VII. CONCLUSION The MIND project aims to ease the creation and provision of broadband services and applications that are fully supported and customised when accessed by future mobile users from a wide range of wireless access technologies. To achieve this goal, novel ad-hoc topologies for connecting user terminals to IP based radio access networks are being addressed. Among other aspects, the research considers mobility, QoS, multi-homing, multicast and security issues. Those topics are analysed from the service, network, air interface and terminal perspective. Furthermore, selected key concepts are validated and evaluated in the frame of trial activities. Relevant results have already been submitted to international standardisation bodies and further contributions are foreseen as the project progresses.
ACKNOWLEDGEMENTS This work has been performed in the framework of the IST project IST-2000-28584 MIND, which is partly funded by the European Union. The authors would like to acknowledge the contributions of their colleagues from Siemens AG, British Telecommunications PLC, Agora Systems S.A., Ericsson Radio Systems AB, France Télécom S.A., King' s College London, Nokia Corporation, NTT DoCoMo Inc, Sony International (Europe) GmbH, T-Systems Nova GmbH, University of Madrid, and Infineon Technologies AG. REFERENCES [1] “ Isle of Man: The future has started” available at www.siemens-mobile.com/pages/ isleofman/index.htm [2] Wisely D.R, Eardley P. and Burness L. – “ IP for 3G” , John Wiley, April 2002, ISBN 0471 48697 3. [3] P. Chaudhury, W. Mohr and S. Onoe. “ The 3GPP Proposal for IMT-2000” . IEEE Communications Magazine, Vol. 37. No. 12, December 1999, pp. 72 – 81. [4] ITU-R TG 8/1: Detailed specifications of the radio interfaces of IMT-2000. Document 8-1/TEMP/ 275-E, 18th meeting of Task Group 8/1, October 25 to November 5, 1999, Helsinki, Finland. [5] UMTS-Forum. “ IMT-2000 Licensing Conditions & Status” . www.umts-forum.org/licensing.html [6] IST-2000-28584 MIND www.ist-mind.org/ [7] IST-1999-10050 BRAIN www.ist-brain.org/ [8] A. Kassler, L. Burness, P. Khengar, E. Kovacs, D. Mandato, J. Manner, G. Neureiter, T. Robles, H. Velayos. “ BRENTA - Supporting Mobility and Quality of Service for Adaptable Multimedia Communication” . Proceedings of the IST Mobile Communications Summit 2000, Galway, Ireland, 1-4 October 2000, pp 403 – 408. Available at www.ist-brain.org/ [9] C. Keszei, N. Georganopoulos, Z. Turanyi, A. Valko. “ Evaluation of the BRAIN Candidate Mobility Management Protocol“ . IST Global Summit 2001 Barcelona, September 2001. Available at www.istmind.org/ [10] BRAIN deliverable D2.2 “ BRAIN architecture specifications and models, BRAIN functionality and protocol specification” . Available at www.istbrain.org/ [11] C. Keszei, J. Manner, Z. Turányi, A. Valko. “ Mobility Management and QoS in BRAIN Access Networks” . 1st International BRAIN Workshop, London, November 2000. Available at www.ist-brain.org/ [12] ETSI BRAN www.etsi.fr/bran/
