Infrastructure for Networking Research

Infrastructure for Networking Research- Rationale

The UK academic network research community is - and is internationally recognised as - world class. Yet it does not have access to the high performance network infrastructure on which to conduct realistic-scale experiments required for the most crucial of today's and tomorrow's network challenges. Given the increasing dominance of ‘Information Networking’ - with data now exceeding voice on the main telecommunications networks both in the USA and the UK and growing at a super-exponential rate - this is a serious impediment to progress. And this is a research area where the importance of experiments cannot be overstated; we are charting new waters and ‘learning by doing’ is necessarily the order of the day! For this it is critically important that the networking research community in the UK has access to high performance network infrastructure of appropriate scale and connectivity on which to observe behaviour, measure performance and trial solutions to new networking paradigms. It must be possible to load the network to the point of overload - to 'break' it - so that recovery and performance implications can be assessed. This cannot be effected in a small scale laboratory simulation, nor on a 'service' network such as SuperJanet where the requirement of the network operator (UKERNA) to service the basic communication needs of 'normal' users must inevitably seriously compromise networking research agendas.

It is this serious shortfall of key research infrastructure provision that prompts this proposal. It comes from leading 'networks research' universities operating in concert with leading industrial organisations, proposing the provision under JIF of ‘Infrastructure for Networking Research’ (INR). This will put in place in the UK a much needed world class facility, complementary to that already available to researchers in the USA as the Next Generation Internet (NGI) infrastructure initiative. It will enable the relevant researchers in the UK to continue their activities at the pinnacle of 'networking-in-the-large' research and it will position the UK very favourably for full participation in European Framework V initiatives as these develop down-stream. This is especially the case given the strong emphasis on partnership with the relevant national industries such as are engaged here.

The recent Next Generation Internet Principal Investigators' meeting in Washington gave an excellent overview of the subjects to be tackled in the US programmes in this area. It was also indicated that strong collaboration, and even possibly interconnection, with similar programs elsewhere would be welcomed. The UK networking community was represented at the meeting by Professor Peter Kirstein (UCL).

The rest of this document is divided into two sections. Firstly, we present planned and possible collaborations using the facility. Secondly, we list some of the track record for initial set of unversities who are supporting this bid. Both sections are listed alphabetically by University (department).

Continuing and Future Research

Particular strengths of this proposal include:

The role of the UK’s major communications provider, BT, supporting this proposal, and a number of their equipment providers.

The role of UKERNA in providing access to the service network, and possible introduction of real traffic from it onto the research facility.

Strong UK participation in Internet research and standards organisations (compared with the rest of the Eu).

The convergence role of the telecom research in the various EE departments, and their interaction with the Internet and Multimedia researchers in EE and CS departments across the consortium.

The relative youth of the lead researchers at most sites.

Here we present research that is proposed or planned at each university with emphasis for opportunities for collaboration.

Bristol

Bristol University is a member of the Digital Broadcast VCE as is UCL. The Mobile VCE core research programme is being proposed for a further 3 years, including further work on the topics above in the mobile domain. Applications have also been submitted to the LINK Broadcast programme (one involving the Digital VCE) concerning research in the impact of convergence of the major service domains, of heterogeneity of information structure, content and distribution, and in the scalability of mass interaction.

Edinburgh

The Division of Informatics at the University of Edinburgh is very interested in the concept of a UK Infrastructure for Networking Research (INR). The University has been centrally involved in UK academic efforts on providing high-speed networking services, and the proposed network would continue this work in the context of networking available for research, rather than continuous service. Various current research projects, at the moment restricted to LANs and the local MAN, would benefit from the scaling-up possible in a UK context. The INR might also be used for validation of current modelling and simulation work.

The Division, through Dr Gordon Brebner, is currently developing research activity in the area of active networking. The INR would be a perfect experimental vehicle for practical work in this area, particularly the use of the "active protocol" concept that has been developed. More widely, the INR would underpin a general programme of work being formulated in the area of "global computation", which spans theory and practice, and includes the development of novel communication mechanisms. This links well with work at Sussex.

Essex

Future research includes the development of client-side procedures for recognising and predicting traffic-contention and packet-loss in high bandwidth isochronous UDP transmissions over IP WANs. There are at least two outcomes from this work:

verifyable parameters for use in theoretical models for network traffic management;

practical feedback signals for use by adaptable codecs and adaptable routers.

Design of video codecs for better immunity against loss

Generation of video traffic at any rate

Use of a video distortion meter to measure the QoS, as close as possible to what the human observer sees

Monitoring the QoS in an ATM type network using OAM cells

These topics link well with those at Lancaster, UCL and Loughborough.

Glasgow

The industrially funded work looking at congestion control, and the economic resource modelling work being carried out within the end-system in the Pegasus II project both suggest that there is significant research potential in new transport level protocols which take account of both network resources and end system resources. Such protocols would rely on significantly different (economic related) network monitoring and resource prioritisation principles within the network itself. Further research in this area could only be carried out with collaborators over a realistically scaled network such as the one proposed. This links well to pricing work at UCL, as well as potential onward collaborations with Cambridge.

Lancaster

With the proposed infrastructure in place, Lancaster's activities in developing and evaluating QoS support mechanisms would be considerably enhanced. It has long been an aim to work in cooperation with other institutions (such as UCL) and research labs (BT Labs in particular), and this would become enabled. Trials on the IPv6 hardware and software, and on the recent REDO RSVP mechanisms, developed at Lancaster would be conducted.

First, connectivity would be vastly improved: being geographically isolated, Lancaster has never had high-capacity and experimental links to its partner sites elsewhere. The proposed JIF link would, through its connection to London, enable further links to the South-East of England (to BT Labs, in particular) as well as to other sites in Europe and even to North America. Lancaster University has on several occasions been invited to join research partnerships in Europe and more recently by US partners to conduct networking experiments, but has never been able to take a full part because of its limited connectivity.

Second, the local lab-based experiments (on the campus network) can be extended to long-distance investigations. For example, BT Labs along with several other research institutions have installed the Lancaster mobile IPv6 video servers: there would be significant benefits gained both by those partners and by Lancaster in being able to inter-connect using the proposed JIF infrastructure. Researchers at Lancaster would be able to plan and submit project bids based on realistic, long distance experiments, for example to test the utility and performance of the lightweight Internet signalling mechanism of REDO RSVP that has been developed and tested by means of simulation up till now.

Imperial College

ALPINE (Application Level Programming Inter-Network Environment) is a new collaboration proposal with University College, Lancaster University, Sussex University and University of Technology, Sydney submitted to BT, to investigate the use of the Darwin Architecture Description Language (ADL) to specify the ALPINE components required for a particular application and to investigate Policy Based Management of Proxylets and Protocol Stacks.

Policy Based Management of Programmable Networks is a proposal in submission concerning the need to be able to program network components to adapt to application requirements for quality of service, specialised application dependent routing, to increase efficiency, to support mobility and sophisticated management functionality. There are a number of different approaches to providing programmability all of which are extremely powerful and can potentially damage the network, so there is a need for clear specification of authorisation policies i.e., who is permitted to access programmable network resources or services. Obligation policies are event triggered rules which can perform actions on network components and so provide a high-level means of 'programming' these components. This proposal will investigate the use of policy as a means of programming the flexibility needed within the network to support network aware applications and application aware networks. This links well with work at Lancaster and UCL.

Lougborough

A national research backbone would be highly attractive to CS, because it would mean that we would be able to experiment with much more bandwidth intensive applications than are currently practical on the JANET network. In order to be good network citizens, it's necessary for us to throttle back our demand for bandwidth to avoid saturating the site's Internet uplink and thus impacting on production IP traffic. This relates closely to UKERNA interests. Services plan to work closely with the departments of Computer Science and Electronic and Electrical Engineering to make the INR available in parallel to the established network infrastructure. In the longer term we would anticipate working with the local cable company, Diamond Cable (which provides much of the dark fibre for EMMAN), to provide INR access for other EMMAN sites - the University of Nottingham, Nottingham Trent University, the University of Derby, and De Montfort University.

Loughborough EE intend extend our understanding of network performance monitoring issues in a number of areas, including Intelligent Processing and Collection of Network Performance information. This work would provide higher levels of interpretation of network performance data leading to better understanding of the problems effecting high performance network operation; understanding the relationships between network performance and human performance when using future applications running on high performance networks such as Virtual Reality systems; understanding the relationships between low level network performance and mid-level protocol operation. Extending monitoring concepts to enable new activities such as fraudulent use of networks to be detected.

This research requires extensive access to high performance networks in order that real, advanced applications can be monitored, and experience of the operational performance of high performance networks can be studied economically and in such a manner that experimental tools will not disturb conventional users. This relates very closely to work at Essex (CS and EE), UCL (CS and EE) and Lancaster and Imperial.

RAL

Increasingly, there is a need to be able to conduct exploitation trials of emerging application technology requiring network support for continuous media without interference from other traffic which can destroy the trial. This is particularly the case in trials for mission-critical technology, as for example in the case of expert guidance or training during medical surgery. RAL proposes level-2, leased-line access to the research networking infrastructure, which would be shared with UKERNA. There are two areas for potential early exploitation of joint research network infrastructure by RAL.

Tele-surgery

Work is in progress (with University of Buckingham) in designing a compression system for use with a head camera worn by an operating surgeon, to allow a remote consultant to offer advice or training during the course of an operation. Wide area trials of this will be needed, possibly as a preliminary to international trials.

High-performance multimedia delivery

Work in INK resulted in a cheap PC-based prototype video server. A proper trial of this technology was conducted in the local area by installing in a hotel for some months and enabling guests to use it. This was possible because only low-bandwidth, best-effort, wide-area access was required to support email and web browsing. Exploitation of video-server technology in other application contexts will require the ability to conduct trials in the wide area.

A continued need is foreseen for high-performance infrastructure to support development and exploitation of pilot applications. Such applications are expected to be tailored more frequently for specific end-user communities. Two areas are identified: interactive, conversational systems in which remote guidance or training in practical techniques is involved; and rapid browsing and delivery of high-quality stored video, potentially to multiple subscribers. This links perfectly with work at Lancaster.

Surrey

Surrey are involved in several Framework V bids with UCL. The research interests of the group cover, but are not limited to, the following areas:

Traffic characterisation for broadband multimedia services

Broadband access technologies

Integration and co-existence of IP and ATM

New-generation Internet, Integrated and Differentiated Services

Network and Service Control and Management using IN, TMN and TINA principles

Open Network Control, Programmable and Active networks

Mobile Intelligent Agents in Telecommunications Systems and Services

Network security

Sussex

The ALPINE project with BT, UCL, Lancaster and Essex. In this, we will be using the SafetyNet language to allow application control over elements within the network, and providing abstractions within the language to match the behaviour of real network elements which provide more complex services than simple best effort packet delivery, such as controlled QoS. A proposal to the EPSRC tdevelop the SafetyNet language further, providing byte code representation and enhancements for security, and prototype implementations to demonstrate performance.In the all the SafetyNet work, we need to engineer the switching elements so that the packet queues are controlled by the SafetyNet

runtime. These experiments are impractical if the networks also carry service traffic, since the service traffic cannot risk disruption.

UCL

UCL is one of the leading university research centres in networking in the world. UCL has recently been awarded support (totalling over £5m over 8 years) to establish a new ‘Engineering Doctorate Centre’ in Communications. This provides support building up over 3-4 years for an additional 40 Research Engineers in communications at UCL. Added to the existing base of postgraduate and post-doctorate research and project staff this represents a formidable human resource with which to carry out research work on the proposed network infrastructure. Some illustrative examples of appropriate research areas are as follows:

network, system and resource Management

dark fibre: testing new approaches to high capacity networking exploiting synergy between optical packet and DWDM schemes in field conditions

long distance delivery of mm-wave radio signals for tetherless remote access with combination of fibre and radio based access featuring heavily

study and practical evaluation restoration strategies in heterogeneous networks involving e.g. IP, ATM, SDH and wavelengths

General system demonstration of optical office area networking and provision of interconnects into the routers/switches of the main test-bed facility.

Existing project work will be extended to investigate both distributed and centralised management in the WAN context. Experimentation would be expected to consider such issues as the affects of timing and performance under 'real' load. Prospective FW5 projects concerning policy and IP routing would use the infrastructure for in depth studies and verification. The 'active' (i.e. open programmable) management platforms currently under development for LEARNet and EPSRC projects could be distributed through the prospective network to facilitate experimentation with new emergent traffic management algorithms. In this context it is of critical importance to have a network of a realistic size (both in terms of distance for timing purposes; and component complexity for processing overheads) in order to understand fully issues of resource management. Resources in this context concerns capacities for such things as: transmission, processing and switching / routing / queuing. With data and MM oriented applications, the statistics of the demands on these resources may be expected to be very non-standard, although this may vary between core and access points. Only with a full size, advanced, captive network can these issues be studied.

Vertical Integration

With a fully captive DWDM/SDH/ATM/IP network, the BT/CASE project studies of: traffic engineering (planning, load balancing, restoration) can be deployed for full validation under load. With possibilities of (simulated) mobility, experimental exploration of studies under way concerning UMTS intra-PLMN backbone network (with and without flow/aggregation) can equally be performed.

UKC

Much of our work would benefit from the opportunity to set up wider scale collaborative projects. It would have clear benefits for our monitoring work and for other communications-oriented projects which are synergistic with it.

Theoretical work on group support and multicast routing at UKC (Dr A.G.Waters) has been under way for a number of years. The logical development of this work is to include practical tests of the application of multicast technologies in the support of a number of application areas. The availability of an open research infrastructure would make possible the performance of larger scale tests outside the local area.The recent move of Dr D.H.Shrimpton from Oxford Brookes to Kent has introduced an active research thread concerned with the support of interactive digital television. Practical test of this work also requires access to an internally monitored, quality of service managed research testbed. Finally, UKC monitors the performance of the JANET network on behalf of the funding bodies. This activity is currently limited to monitoring from user sites, because the operators are understandably reluctant to have probes attached within the network, because of the possible operational impact. However, there is a pressing need for more strategic information about traffic patterns and user behaviour. The ability to demonstrate the effectiveness and resilience of new monitoring techniques within an experimental testbed would increase the confidence of network operators generally, and help to establish a broader range of field trial opportunities within the UK. This clearly links wel lwith the work at Loughborough EE.

UKERNA

UKERNA proposes to participate (by contributing 0.25 SY per year to manage its interconnection to the proposed infrastructure) in order to:

gain early involvement and operational experience in network development leading to potential new technologies and services;

prepare for the early introduction of world leading network services in support of national competitiveness.

share (when possible with acceptable risk) operational traffic with the new infrastructure to provide realistic workloads;

use the JANET managed bandwidth service to enable the wider involvement of the UK research community in novel application development.;

assist in interconnection of the Research Network Infrastructure with Internet2;

It is proposed that UKERNA share access to the proposed infrastructure with CLRC-RAL at level 2, leased line.

As indicated in the "Justification", the creation of the proposed network and the involvement of the SuperJANET operator will enable us to realise the original vision by the transfer of knowledge, skills and possibly facilities from the experimental phase to operational use.

Existing University Work

University of Bristol, Centre for Communications Research

The main specific areas of investigation relevant to this proposal have included:

Application of formal specification methods to the design and analysis of distributed protocols and algorithms, Establishing object-oriented distributed processing models and standard for control of audio/visual systems for domestic entertainment applications. A common theme in this work has been specification and implementation of protocol test suites and interoperability test specifications, and investigations of integration via residential gateways.

Open service access and execution platforms, including service and connection management following the TINA model, studies of integration of Intelligent Network solutions in UMTS for mobile multimedia services.

Participant: Alistair Munro

Edinburgh University, Department of Computer Science

BT Labs - CASE studentship on modelling and simulation of ATM networks; proposal would allow better validation and calibration.

Anticipated studentship (via Ian Marshall) on use of soft circuitry in active networking components.

SHEFC - high-speed ATM link between brain scanner and visualisation boxes - currently restricted to MAN, but could be extended.

EPSRC - current bid in for Internet/other network large-scale simulation; more calibration and validation possible. –

current grant (with BT Labs) on workstation interface to high-speed optical LAN; possibilities for extension to wider area.

EC - current grant on telematics, mainly human factors in teleconferencing, but could benefit from better infrastruture.

Participant Gordon Brebner

University of Essex, Computer Science Department.

We are interested in quality of service, adaptability, and resource management in IP networks. Our research focuses particularly on producing resilient systems that allow good-quality video to be transmitted over heterogeneous, traffic-laden, networks. We therefore study problems related to packet-loss and delay-variations. Carrying out controlled measurements on the existing SuperJANET network has proved extremely difficult, not to say inappropriate. Our research would greatly benefit from having access to the wide-area captive network (INR) envisaged in this JIF proposal.

There are three end-user multimedia contexts for our work:

real-time video conferencing;

telepresence and remote robot control;

full-motion interactive video for distance learning.

Most of our research is directed at the third context. Amongst our current collaborative projects are the following:

HIGHVIEW, funded by BT Laboratories, with UCL

HIGHVIEW is part of the UMBRELLA research consortium, funded by BT.

PIPVIC-2 funded by UKERNA, with UCL

We have also received support in the form of hardware and software from Oracle (eg beta test site for Oracle Video Server products), BT (eg connection to the 155Mbits/sec. LEARNet) and nCube Ltd. (eg support for parallel video server). We are connected to LEARNet via a CAIRN router that permits experimentation with partner-sites at UCL, Cambridge and BT (Martlesham). Our video servers are able to deliver multiple, controllable, full-motion video streams to LEARNet, for test purposes.

Participant: Professor Simon Lavington

Essex University, EE Department

Essex EE are working on sending video over ATM networks supporting ABR services. The video rate in response to a feedback from the network is controlled by a video transcoder. This will minimise the risk of network overloading. EE are developing a video distortion meter, that can measure subjective quality of video, which is very similar to what the human observer can see. This is particularly interesting for measurement of video quality due to packet/cell loss. The device can be used in various parts of the network, such as routers, to decide more subjectively which packet is more appropriate to be dropped, in case of congestion. The other project is the modelling of video traffic, such that the generated traffic is very close to that generated by the real-time MPEG-2 codec. By this method we can very easily generate video traffic at any rate, with a simple PC. Of course this is needed if you want to fill a network with say with 155 Mbit/s video traffic. Finally, there is ongoing work for several years on the design of layered/scalable video codecs, that can be used with various priority levels to improve QoS.

Participant: Professor Mohammed Ghanbari.

University of Glasgow, Department of Computing Science

The MINIMS project involved the research of real-time networked audio services to address real-time distributed audio and music related applications. The associated NETMUSE project involved the deployment of ATM based networks and networking research from MINIMS to provide audio applications between 5 music departments in Scotland using the MANs and MAN interconnect.

The ENTROPIC project applied technology using Large Deviation Theory for ATM call admission and "application admission" in operating systems. This project benefited particularly during the brief period of access to the Super-JANET network where a connection with the partners in Cambridge was possible.

The REVELATION project is a SHEFC funded Research Development Grant which has assembled samples of high fidelity information capture, storage and presentation systems. The platform enables research into new models of computer-assisted consultation, collaboration and distance learning, all relying on leading edge internetworking techniques.

The department also has an Industrially funded Ph.D. student (by Sun) who is looking at evolving congestion control techniques to take account of macroscopic network measurements and statistics in order to improve response and useful throughput without degrading network fairness.

Pegasus II is an ESPRIT funded long term research project researching operating system support for distributed multimedia. Though particularly concentrating on the end system, this project would benefit from access to the distributed collaborators (through JAMES or its successors) which only the INR as proposed could deliver.

The academic staff member listed is a recent appointment at Glasgow. He has previous networking research experience at Cambridge where he was involved in the Fairisle ATM switch development projects, ATM protocol and end-system development, the Desk Area Network, the PNO and James research network testbeds, and the development of B-ISDN terminal equipment software.

Participant: Richard Black

Imperial College, Department of Computing

Pro-active Role Based Management for Distributed Services:

To refine and extend the tool support for the specification of policies, roles and relationships.

Specification and analysis of security policy for distributed systems:

To provide a framework for managing security in large, multi-organisational distributed systems.

Distributed link services for mobile environments (LinkMe) :

In collaboration with Southampton University, Ordnance Survey and HP Labs, Bristol. To provide a distributed link service for mobile users of distributed hypermedia (Web) information systems. Ordnance Survey's geographic information system will be used for a demonstrator and critical evaluation.

Control and Co-ordination of Complex Distribution Services (C3DS):

To exploit distributed object technology to create a framework for complex service provisioning. The C3DS approach to building a framework for complex service provisioning, unlike other approaches, will be based on unifying three technologies: software architecture based development environments, software agents and transactional workflow management systems.

Participant Professor Morris Sloman

Lancaster University.

Quality of Service Management: Lancaster's work on Quality of Service architecture (QoS-A) is internationally respected. In several inter-related projects we have developed services and mechanisms to allocate and manage the flow of continuous media traffic across an inter-network including the Internet and local area ATM networks. With funding from EPSRC, BT Labs and the European Commission we are investigating QoS management in multi-service networks, in which we apply our QoS-A work to a model for service provision that distinguishes the roles of end-user, service broker, network provider and information provider. The major focus of this research, which strongly relates to the QoS-A work, is the development of services and support mechanisms for multipeer (ie group) communications. We have developed QoS filtering mechanisms for adapting continuous media flows, and have shown how these can be integrated with Internet routers that use RSVP for resource reservation. This work is continuing within a collaboration that includes partners in the best networking research labs throughout Europe.

Integrated Multimedia Services for the Internet: Much of this work is now done within the context of the latest Internet Protocols, such as IPv6 and RSVP. We currently run one of the largest IPv6 testbeds and have developed the first implementation of Mobile IPv6. This Mobile IPv6 implementation has also been used to support a locally developed IPv6 Video on Demand service, now used by companies such as BT and Telebit. Current projects are investigating the mapping of Integrated and Differential Services on to networks such as ATM, and on to Active Networks where routers intelligently process the contents of data streams.

A Specification Architecture for the Validation of Real-Time and Stochastic Quality of Service This project continues from previous work in defining a language independent specification architecture for (open) distributed multimedia systems. In collaboration with the University of Kent at Canterbury, this phase of the research continues our work on providing explicit support for the specification and validation of quality of service properties, and will also develop more extensive software tools based on the new approach.

ReTINA's aims are to develop an industrial strength, TINA-compliant Distributed Processing Environment, together with its specifications and an associated set of development tools.

Participant: Professor W.D. Shepherd, Professor D. Hutchison,Professor G.S. Blair, Dr G. Coulson, Dr. N. Davies, Dr. A. Friday, Dr. A. ParkesDr. A.C. Scott.

Computer Science & Services at Loughborough University

The Department of Computer Science has a track record of work on computer networking stretching back for over ten years. Our focus has been on `middleware' technologies such as WWW caching, indexing, cataloguing and replication/mirroring. In most cases, the primary motivation for these projects has been to find ways of routing around the problem of wide area bandwidth availability and cost. To appreciate why this work is necessary, consider that a single entry- level PC can trivially saturate a 10Mbit/s Ethernet.

In addition to local services, Computing Services is also a partner in the consortium which provides the JANET Web Cache Service, along with the University of Manchester. This JISC funded project is effectively the single largest node on the JANET network, currently accounting for:

up to 33% of the chargeable JANET World-Wide Web traffic

160GB of data transferred per day

22 million requests serviced per day

a pooled cache of some .5TB of data

Whilst nominally a service rather than a research project, much research and development is necessary in order to facilitate service provision. For example, we have done much of the pioneering work on the new Cache Digest protocol, and were one of the first JANET sites to connect to the 6bone - the experimental global IPv6 Internet. In the near future, we are particularly interested in IP based resource reservation and queuing/scheduling developments - e.g. QBONE, RSVP, RED. We would expect to use these over the INR to experiment with dedicated bandwidth for live Cache Service traffic, since WWW caching technology is well suited to application level routing with failover. Although in theory the JANET ATM network should be able to provide this, in practice it has not been forthcoming.

Electronic and Electrical Engineering, Loughborough University.

Much of our current work is directed towards the monitoring and interpretation of network and application performance. We have developed a range of tools to enable the performance of different network types to be monitored and subsequently processed to allow "interesting" aspects of network operation to be easily identified. Work on application performance has centred on attempting to understand how the low level performance of a network influences the subsequent user perceived performance of a networked application. This relationship is often complex, and tools to present the predicted performance of such an application to a user have been demonstrated.

Specific Research Projects include: Monitoring and Grading Application Quality of Service. (EPSRC), Network Performance Monitoring (Part of the BT funded URI) and Monitoring the SMDS Network (BT).Other Research projects include: ATM Telemetry (TCD/Hytec Electronics), and Generic modelling of Network Scenarios (EPSRC).

Participant: Dr. D.J.Parish

RAL

The inter-departmental Advanced Communications Unit and the Information Technology Department at Rutherford Appleton Laboratory have been active in telecommunications services and network development, world wide web technology, database application development, and user interface design for over 15 years, with funding from national and European Union sources. The ACU at RAL was a principal contributor to the development of the technical strategy underlying the original SuperJANET initiative. RAL has also provided the co-ordinator for the MNA programme in recent years. Recent communications-oriented projects have focused particularly on exploitation of emerging communications technology through new applications. Applications include:

MANICORAL (EU Telematics, developing a distributed co-operative visualisation environment for a user group of geo-scientists).

INK (EU ESPRIT, developing multimedia networked kiosks for use in hotel rooms: integrated terminals providing real-time TV, video-on-demand, off-air digital recording and playback, and Internet email and world wide web access).

AQUARELLE (EU Telematics, developing web-oriented access to heterogeneous cultural archive databases across Europe).

Participant: Chris Cooper

University of Surrey, Centre for Communication Systems Research

The Centre for Communication Systems Research (CCSR) (http://www.ee.surrey.ac.uk/CCSR) at the University of Surrey belongs to the School of Electronic Engineering, Information Technology and Mathematics. The Department of Electronic and Electrical Engineering received the highest 5* rating in HEFCE's 1996 research assessment exercise.

IthACI - The overall scope of the IthACI (Internet and the ATM: Experiments and Enhancements for Convergence and Integration) ACTS project is to evaluate and contribute to the different technologies that together permit the efficient transport of IP traffic over a public or a private ATM backbone network.

BISANTE - The BISANTE (Broadband Integrated Satellite Network Traffic Evaluation) ESPRIT project aims to support network and service providers in management of their services and capacity planning of their resources, by providing them with a simulation methodology realised in a workbench.

REFORM - The REFORM (REsource and Fault restORation and Management in ATM) ACTS project investigates fault and resource management in ATM networks through close interaction between the control and management planes.

FlowThru - The FlowThru ACTS project aims to provide a service and network management system integrating management applications that originate from previous ACTS projects such as VITAL, REFORM and Prospect. In this case, integration brings together aspects of order handling for service access, network planning, configuration management, Quality of Service management, problem handling and accounting processes.

MIAMI - The MIAMI (Mobile Intelligent Agents in the Management of the Information Infrastructure) ACTS project aims to investigate the benefits gained from agent mobility and intelligence in the context of telecommunications network and service management, information services and electronic commerce..

VITAL - The VITAL (Validation of Integrated Telecommunication Architectures for the Long-term) ACTS project aims to validate the emerging Telecommunications Information Networking Architecture (TINA) as the vehicle for providing future advanced telecommunications services.

Participant: Professor George Pavlou

Sussex COGS

We currently hold two projects and are in the process of bidding for two more which will be able to make use of the proposed Infrastructure for Networking Research.

The EPSRC funded Lowband project has produced a language, SafetyNet, to allow resource controlled user computation within the network. This language allows applications to execute computation within the network and so facilitates adaptive objects in simple best effort networks. SafetyNet is currently the subject of two further proposals:

We also hold a grant with UCL to investigate the mental models users hold of networked applications. As part of this project we will be re-engineering the front-end to video conferencing tools, and we will be running user trials over the network. Controlled and consistent network conditions are necessary to ensure the validity of our results.

Participant: Dr Ian Wakeman

UCL CS

Notably, BT/JISC have funded the LEARNET family of projects (JAVIC, HICID and HIGHVIEW) which provides network between BT, Cambridge University, UCL and Essex and may be used for delivering sections of the telecom and BT Martlesham MSc and MRes programs as well as the 5 year MMN project, which involves 4 of the other universities in this proposal. This is just one of a family of projects in the general area of Resource Management (QoS, routing, multicast etc). The US agency, ARPA funds the RADIOACTIVE project which is addressing the needs for application level active networking. We are also funded by ARPA to work on Policy Based Resource Control of Networks, with particular application for Multimedia Conferencing. The EU funded the HIPPARCH project investigating High Performance Protocol Architectures using ILP and ALF. This has led to possible folllow-up projects with BT such as the directly funded ALPINE project, with Imperial and Lancaster and Sussex Universities and some planned Framework V proposals including follow-ups to the current COIAS ACTS project on the Convergence of IP, ATM and Satellite.

A number of our projects are motivated by our primary business in the University Sector, dissemination of learning and research (the Americans call this "Scholarly Pursuits"). These include several that are under the broad heading of Technology In Learning & Teaching (TILT) across UCL. A number of projects in CS and between CS and partners exist that have motivation from delivering teaching and learning (and research collaboration and results) to the wider world. Recent work includes ReLaTe - and TACO Other existing projects centred in CS include: Relate and Prospect and MECCANO. Meccano is a follow-up to Merci all still running, and all having teaching based demonstrators. MERCI continues the work begun by the successful MICE. A lot of the work is centred around multicast/internet technology -= we have some proposals in the pipeline for Next Generation Internet technology work (of which the LEARNET work is a piece), across Europe - some of them include Eutelsat who already provide a 2Mbps satellite feed for multicasting.Network Management has been a major topic of interest at UCL for many years. Current projects include the Flowthru project.and the VITAL project.

Participant: Professor Jon Crowcroft

UCL EE

The LEARNet project was, form the first, envisaged to be a key factor in progressing the distance teaching enterprises of UCL (EE/CS) with BT (both the MSc Programme at BLT and BT Masters Program).

A core concern of the group is the engineering of systems for Management and intelligence for Public Network Operators. This has a number of aspects including: System integrity, functional design and performance characterisation - of the systems, the network and their interworking dynamics. Core developments in the integrity area is derived from EU project TRUMPET and is being developed further under EPSRC funding: "The development of Integrity Metrics". Studies in this direction are also components of LEARNet related projects concerned with the development of co-operative 'complex adaptive' systems for fault finding and load balancing and management issues for large scale QoS based IP networks, funded by BT/CASE and Nortel/CASE studentships. These projects will also consider problematic network system dynamics such as churn and fractal traffic; and the impacts on these on network planning. New projects are being developed to explore these issues further within FW5. These will focus on core integrity issues and the development of adaptable co-operative systems. A specific specialised network management issue for networks is planning for resilience and load balancing. These issues are tackled for ATM within the REFORM (AC208) and (to some degree) MISA (AC408) projects above. Building on studies of SDH restorable networks; new studies for planning for restorable networks with IP over WDM within the context of interconnect services is under way, funded by BT/CASE. This project should also consider questions of the affect churn and other novel traffic patterns.

On the impact of impairments on optical fibre systems, looking in particular at the effect of interferrometric noise in optical networks. Techniques to increase the capacity of high-speed long haul systems to multi-terabits/fibre, looking at bandwidth efficient modulation schemes and the impact if fibre non-linearity on these schemes in a DWDM environment. Short-haul, low-cost optical systems, aimed at the LAN and display interconnect market. The work in these areas is/has been funded by EPSRC, EU, DTI Foresight and direct industrial funding (e.g. IOC Ltd, Nortel). For access technology activity is focussed on the further development of xDSL technologies with a goal of supporting STM-1 rates over the copper local loop, the work is looking at echo cancellation, equalisation, coding and modulation. We also consider fibre supported mm-wave radio access with a goal of several 100 Mbit/s to the customer at 30-60 GHz (this work links closely to the radio heading as well). The work is/has been funded by EU, EPSRC and direct industrial funding (e.g. Fujitsu, BT). For Mobile and Radio activity addresses such issues as improving the efficiency of inter-PLMN GPRS interconnect, dynamic channel allocation in satellite PCN systems, handover strategies in S-PCN and satellite based packet data systems using mostly industry funding (e.g. Nokia).

Participant: Dr Lionel Sacks

University of Kent

ROPA: Diagnosis of congestion in ATM networks

This project has built upon locally available ATM measuring and monitoring equipment to carry out investigations of the timing patterns which characterize various kinds of network queuing and congestion behaviour. It has identified observable timing patterns that can be used to deduce possible

causes of congestion.The results have been tested in the inter-institutional ATM environment in Canterbury area, but there is comparatively little traffic on this at present. Access to a research infrastructure where monitors could be attached to significant trunks would allow wider and more realistic tests.

Coordination and Control Interfaces in Multimedia Information Systems

This built on the results of a JISC-funded Video file server project (SERVICE), which produced a flexible and efficient ATM video store with powerful synchronization facilities. It was a technology transfer project under HPIP, and resulted in the production and field trialing of a highly interactive video information system for tourists in Tunbridge Wells. It culminated in a public user trial of the interactive video system, which was successful. However, the need to move the team and support environment across the county resulted in a short, intensive trial under difficult conditions. More extensive results could have been obtained from a longer trial if a research infrastructure with managed quality of service had been available, so that the trial equipment could be supported and monitored from UKC.

Monitoring and event detection in ATM networks

There have been several projects with HP Labs, using and developing out ATM monitoring capabilities to include the detection in real time of significant events in the traffic passing over an ATM network. The system produced allows dynamic reconfiguration of a state-based even recognizer. The projects have been restricted to local area tests. The next step would be to extend the work to the monitoring of a wider range of traffic in larger scale networks.

Participant: Professor P.F.Linington

UKERNA

UKERNA manages the UK Academic and Research Network, JANET. We develop and provide advanced communication facilities for use in the education and research community and our mission includes research into relevant fields. Major service contributions have been through SuperJANET, particularly in deploying high-capacity IP-based networking (based on ATM, SMDS, and leased-line bearer services) and video-conferencing facilities. We are planning for the next phase of development (SuperJANET IV), due to be operational in 2-3 years. We have significant interest and experience in the underpinning technologies, and their operations. We are also involved in the development of promising new applications.

Participants:James Hutton, Business Director &Bob Day, Network Development Director and Deputy Chief Executive.