Multiservice Networks using SuperJANET
Management of Multiservice Networks
5 January 1994
Wide area broadband networks, the intelligent services which extend their functionality and the applications which make use of these service are all complex distributed systems which need to be managed. This management involves interactions between multiple administrative domains -- end users, service providers and network operators. This collaborative project will investigate a range of research issues that apply to the management of multiservice networks, and in which the participants have strong and complementary expertise. An initial set of workpackages is proposed.
The participants are:
University College London (co-ordinator)
Imperial College
Lancaster University
Loughborough University of Technology
Oxford Brookes University
Cambridge University
The project objectives are to develop a set of management tools to perform the following management functions in a multi-domain environment:
i) Configuration Management - graphical tools to support the initial construction and subsequent dynamic change of the software components of the network services, the management system itself and distributed applications.
ii) Policy Based Traffic Management - This will provide a language and associated tools for specifying policy which can be used to modify the behaviour of automated manager agents to enable them to deal with network congestion in real-time.
iii) Quality of Service Management - this will permit multi-media, application specific QoS requirements to be specified and mapped onto the required support mechanisms in the underlying services.
iv) Traffic Monitoring and Measurement - distributed traffic monitoring using statistical algorithms will be provided as well as a traffic source based on a multi-bitrate video and audio service to evaluate performance.
v) Security - an architecture for assuring secure and authenticated signalling will be developed, as well as a unified approach for authenticating end-user traffic flows based on virtual channels.
Multiservice Networks using SuperJANET
Management of Multiservice Networks
5 January 1994
University College London (UCL)
Imperial College (IC)
Lancaster University (LANC)
Loughborough University of Technology (LUT)
Oxford Brookes University
Cambridge University
The co-ordinating university will be University College London (UCL).
The deployment of new wide area broadband networks will result in complex interactions between the management of the networks, the intelligent services based on these networks and the end-user applications using these facilities. Management complications arise because:
o New applications, such as interactive multi-media, place stringent quality of service requirements on the underlying support services.
o The network, the intelligent services and the applications may be complex distributed systems which results in problems of monitoring the system to determine a global state and performing control actions to modify the behaviour distributed components.
Multiple administrative domains are involved relating to end users, network providers and third party service providers
Security Multiple administrative domains create security problems both network provider and for subscribers.
The management systems must be automated, distributed and structured to cope with partitioned responsibility reflecting organisational or service boundaries.
This collaborative project will focus on the following management issues:
It is our view that today's centralised servers will quickly be replaced by multi-agent servers as the services they provide become more sophisticated and as the computational and bandwidth resources they demand increase. It is envisaged that many of the services provided by the SuperJANET network will require this multi-agent approach. Multi-agent servers have a complex internal structure which is not easy to manage using the conventional publish/locate interface binding techniques. Hierarchical composition is needed to permit a complex service, with a well defined interface, to be built as a composition of more primitive components (or objects). Similarly, the management system itself will be built from distributed intelligent manager agents. These distributed subsystems will not remain static but will evolve to incorporate new functionality. Configuration tools are needed to support the initial construction and subsequently dynamic change of these distributed services.
Management involves monitoring the activity of a system, making management decisions and performing control actions to modify the behaviour of the system. Management policy guides the decision making process. Policy is information which influences the behaviour of the components in the managed system. The size and complexity of large networks and distributed systems has resulted in a trend towards automating many aspects of management into distributed components. If the policy is coded into manager components they become inflexible and cannot be reused in different environments. There is thus a need to be able to specify, represent and manipulate management policies without building them into managers or manager agents. This permits manager objects to be reused in different environments and to be provided with the specific policies for each and enables dynamic changing of management policies for a system without changing the managers.
Interactive multimedia-based applications involving people working in groups will create new usage patterns which have yet to be understood. Dynamic QoS management of user traffic is an essential end-user need to handle the demands of such an application environment. To provide contracted-for QoS in the network, we need to understand its support in end-system environments, its specification at the user-network interface, and its enforcement in the network-network interface.
In order to make management decisions managers need up-to-date information on the current state of components under their control. This information must then be distributed to the various managers and intelligent manager agents which require the information. However, it is impractical to monitor and distribute all the information about every connection in the network. For this reason, selective acquisition and distribution of this information will be researched. In particular, current work on the statistical monitoring of multimedia type connections will be developed to enable an efficient cell level distributed monitoring system to be implemented for the network, using monitoring stations at the participant sites. This will be invoked, during the call establishment phase, for any connection by signalling between the communicating parties and will be independent of any central control, allowing expansion to larger networks. Processed information about the connections can then be passed to the manager agents and other entities (eg the policing function) efficiently.
v) Security
Authenticity, integrity, privacy including non-disclosure and prevention of traffic analysis are all aspects of security required in operational networks. They are required both by the provider with regard to running an open network service across multiple administrative domains and the subscriber, who has to be assured of at least the same level of security provided by postal mail. The virtual circuit model of SuperJANET provides a clean opportunity to fulfil such requirements.
In order to address the above issues, our approach will be to establish a number of services between the partners' sites. The services will be selected both to be representative of likely future commercial broadband services (e.g. multi-rate video and audio distribution), and also to stretch the limits of the network from an operational viewpoint. The characteristics of the services (data profiles) and the network performance will be studied and analysed in order to derive an understanding of QoS management requirements. In addition, the distributed locations of the consortium members will be exploited to enable the investigation of distributed systems management issues. Finally, the fact that the project is to be carried out across multiple administrative domains will enable us to test our models of security.
The research will be structured as four work packages each being the prime responsibility of one of the four participating sites, but with collaboration between partners as required. The work packages are provisionally defined as follows:
Configuration Management is concerned with defining, constructing and maintaining complex systems in terms of their structure. Systems are viewed as interconnected instances of components (or objects). Components may be primitive or constructed from other components.
We have developed a set of graphical tools, textual notations and design methods for configuring systems in terms of their structure. The configuration management tools facilitate the specification of a distributed service in terms of the required instances of (software) component types, bindings between interfaces and the allocation of software components to hardware nodes. Dynamic configuration permits both evolutionary change and recovery from component failures. These tools will be refined and extended for configuring the network management system, network services or distributed applications running over the network.
Domains will provide the means of grouping objects and partitioning configuration management responsibility, particularly in an inter-organisational environment such as SuperJANET.
These tools will be tested by experimenting with them for configuration of management and services within the other work packages.
Deliverable: Graphical Configuration Management Toolset
Major Partners: Imperial, UCL
A notation is needed for the specification of management policy which will guide the decision making in automated managers and intelligent manager agents. This work involves designing a policy language compatible with the OSI network management model. Imperial College will also be involved in this work and will provide input from policy notations and tools being developed within the SYSMAN & IDSM Esprit Projects (see section 4 below). These concepts of policy based management will be applied to automated traffic management as explained below
We are developing intelligent algorithms in end-to-end systems that control the rate at which packets are injected into the network; when packet loss occurs or performance degradation is encountered, the source will back off to a point where a minimum quality of service can be achieved for the type of traffic being transmitted; for example the rate at which file traffic is injected can be reduced to a far lower level than is possible for video traffic. Collectively, a group of routers must co-operate to ensure the optimum combination of thresholds and drop algorithms are applied.
We will also investigate how management agents controlling the network can be adapted so that they can deal with congestion problems in real time. Currently, agents send event reports about congestion in switches to managers, which evaluate the information and send back their decisions to the agents. However, in a congested network the messages will not be received by managers quickly enough for them to make timely decisions. Agents need to be become autonomous and intelligent. A manager's role would then be to download policies to agents, which then could themselves determine which drop algorithms should be selected.
Deliverables: Policy Specification Notation
Automated Traffic Management System
Major Partners: UCL, Imperial, Lancaster
The integration of new multimedia applications into a communications architecture that encompasses the new networks poses serious challenges, not only because of these data rate characteristics, but also because of new QoS requirements such as support for continuous media transfers, multimedia synchronisation, and dynamic QoS negotiation and re-negotiation. For applications relying on the transfer of continuous media, it is important that QoS is supported on an end-to-end basis, including application services, the transport protocol and the multiservice network, so that the QoS provided by the system is acceptable to the user.
The approach to be taken in this work package is to consolidate and extend current notions of QoS at the crucial system levels (i.e. operating system, transport and network level) and to define relationships, correspondences and mappings between related concepts and metrics. This work will be tested over the SuperJANET network in collaboration with Loughborough (in WP4) and over the campus wide local ATM infrastructure being established at Lancaster (the first in a UK university). This infrastructure is based around a number of commercial switch designs and both in-house and commercially developed workstation interfaces. The work will also build on ongoing research on continuous media storage servers. These servers will be used to store continuous media streams with varying QoS requirements which can be generated and monitored by the QoS control mechanisms to be developed in collaboration with Loughborough and Oxford Brookes.
A second aspect of the work package involves the handling and control of congestion within an ATM network which may arise due to the dynamically varying traffic load offered. Work at OBU will utilise a reconfigurable transputer array to simulate congestion scenarios and proposed control policies. This work will be undertaken in conjunction with the traffic monitoring studies in WP4, enabling congestion to be predicted where possible and controlled. This will provide an optimum underlying service to the QoS support mechanisms to be developed at Lancaster.
Deliverable: QoS specification and support mechanisms
QoS-controlled multimedia storage service
Congestion control mechanisms
Major Partners: Lancaster, Loughborough, Oxford Brookes
Loughborough will develop a distributed traffic monitoring mechanism for the network and a multi-bitrate video and audio service using multiple stream transmission to allow different qualities of service to be received according to the available bandwidth (e.g. due to local link rate, or network load). This service will provide a flexible traffic source with which to load the network in order to evaluate the performance of the management systems developed in the other workpackages.
The video and audio service will be developed from current work using M-JPEG (although this may be modified for the proposed work) compressed images within a PC environment with transputer enhancement to enable concealment of lost cells. The use of concealment to estimate lost (or untransmitted) cells and intelligent cell discard will allow a lower bitrate version of an image to be sent to a site with a lower capacity connection without the need for the more complex multi-layer coding proposed by certain groups. This approach will allow a practical demonstration to be developed relatively early in the initiative.
This aspect of the workpackage will be undertaken in conjunction with the congestion control mechanisms to be developed in WP3. The aim is to provide an "intelligent" cell (or block) discard mechanism, providing the optimum level of service given prevailing network conditions. It is anticipated that the end-to-end QoS will continually improve as the project matures (e.g. due to the availability of cell level monitoring information and refinement of the algorithms from Lancaster and Oxford Brookes in WP3.
A cell level monitoring system will then be developed building upon statistical monitoring algorithms currently under investigation as part of a SERC funded initiative at LUT. The system control will be distributed, and the optimum information to be passed to the management entities of the other workpackages will be investigated.
Deliverables: Traffic monitoring and measurement toolset
Multirate video and audio service.
Major Partners: Loughborough, Oxford Brookes, Lancaster
Future CCITT plans for B-ISDN network management and signalling indicate a shift towards solutions based upon ideas from open distributed processing, aiming for the benefits of modular and extensible system. Such an approach becomes particularly attractive when it presents the services and capabilities of the network tocustomers in such a way as to allow them to compose such network services to provide a solution to their information problem.
Two issues of security arise in the context of such an open management architecture: the first is concerned with the integrity of the control system, the second is the increasing demand from users for secure communications on demand.
Traditionally signalling systems have taken a simple view of authentication; either the communication is between a user (identified by their physical location) and the network and hence the user is restricted to a certain subset of operations, or between two entities in the network which trust each other. In opening up some of the control operations of the network to customers, there needs to be a finer granularity of authentication. This work package will concentrate on the design and implementation (on relevant elements of SuperJANET) of a suitable security model for secure and authenticated signalling.
Following from this work, there will be the desire for secure end-to-end communications on demand. In this respect, the ATM virtual channel provides a massive opportunity to build a security architecture starting from a very low level. If a general mechanism is provided for authenticating virtual channel connections and if virtual channel connections are not used to multiplex traffic for different authentication entities, then one can build up a new and unified approach to authentication. The aim is to use this simple building block to provide a range of end-to-end security mechanisms.
Deliverables: Security model and implementation for signalling Implementations of end-to-end security mechanisms
Major Partners: Cambridge, Imperial, UCL
All the institutions are involved in teaching courses on distributed systems, networks and network management. The research work undertaken in this project will feed into these courses and into new courses being developed. This input will be in terms of concepts, case studies and the tools developed could be used for practical work associated with the courses. In addition we anticipate having research students associated with the project.
UCL are involved in teaching the Martlesham MSc. This project is particularly relevant to the modules on Private Networks and Network Management. In addition a module on Virtual Private Networks and their management is under consideration.
We envisage that, towards the end of the project, limited access may be given to potential users of the services, that will have been developed; for example, to the Lancaster University team involved in the DELTA JITOL (Just in Time Open Learning) project and who are interested in using SuperJANET as a distance learning vehicle.
UCL have been involved in communications and distributed systems since 1976. The department participated in the Universe and Alvey Admiral Projects which experimented over wide area networks with distributed systems. The group had a long running DARPA funded program in areas concerning traffic management and routing. UCL currently is involved in two RACE projects in network management: the PREPARE project in Virtual Private Networks is building a prototype LAN/MAN/WAN interconnect together with heterogeneous management systems, and using a system built at UCL to combine these. The ICM Project is working on a network management platform. Previous projects, Nemesys and CAR, provided early work on management of QoS and in multimedia application pilots respectively. Work under a SERC project and in collaboration with Bellcore and Xerox Europarc have contributed to our understanding of Multimedia and in particular multicast video and audio traffic requirements.
The Esprit project MIDAS, involves Generic Management Systems. The Esprit MICE project is enhancing the CAR Multimedia System to run over very large networks (with up to several thousand simultaneous conference members, and a million terminal systems).
Jon Crowcroft is responsible for a number of European and US funded research projects in multimedia communications. A recent project worked on protocol migration (Internet and OSI). He has been working in these areas for over ten years. For the last two years he has also been consulting to the Bloomsbury Computing Consortium as a senior systems analyst on the installation of a multi-campus distributed system. He is general chair for the SIGCOMM 94 symposium.
Graham Knight is a Senior Lecturer in the Department of Computer Science. His research has been mainly in the area of computer communications, beginning with the Universe project which studied the interconnection of LANS via a satellite link. Subsequently he led the UCL team in the ESPRIT INCA project. UCL assumed the technical leadership of INCA and their activities included early work on X.400, X.500, ODA and OSI Network Management. Graham Knight then led the UCL team in the ESPRIT PROOF project which studied the use of OSI and other applications across LANs and ISDN. The early work on the OSIMIS management package was undertaken under the PROOF project and one strand of this work is being further developed in the ESPRIT MIDAS project which Graham Knight is coordinating at UCL. In addition to the management work, Graham Knight is actively involved in the use of narrow-band ISDN for teleworking and the interworking between LANs and ISDN.
George Pavlou is a senior research fellow in the Department of Computer Science, where he has been responsible for European funded research projects in the area of network management for future generation high speed networks. He is also actively involved in the OSI to Internet management coexistence work under the auspices of the OSI NMF and he has been a member of the BSI group on network management. He has been doing research in the area of computer communications and distributed systems for the last 8 years with emphasis in the areas of broadband technologies, upper layer architectures and network management. He has also architected the OSIMIS platform which provides a development environment for OSI and Internet management services.
James Cowan is a research fellow in the Department of Computer Science, University College London, where for the last 3 years he has worked on a DARPA video conferencing project and subsequently on two EEC RACE network management projects (Nemesys and ICM). Prior to that he worked in industry for nine years on a variety of software development projects, including compiler development, interactive video systems and treasury debt management systems.
Dr. Jon Crowcroft
Department of Computer Science
University College London,
Gower St.
London, WC1E 6BT.
tel: +44 71 380 7296
fax: +44 71 387 1397
email: J.Crowcroft@cs.ucl.ac.uk
The Imperial College group have extensive research expertise in the areas of configuration management and management policy for distributed systems and are applying this expertise to the management of multiservice networks. The Conic environment for the construction and management of distributed systems has been developed and distributed to about 25 sites around the world. Recently, the group has been investigating the support of heterogeneity, programmed change, and support for more of the software lifecycle under the ESPRIT II project, REX (2080), on Reconfigurable and Extensible Parallel and Distributed Systems. The group have also been involved in the DTI/SERC funded DOMINO project (Domain Management for Open Systems). The DOMINO work on Domains and Policies is being used in two ESPRIT Projects on Management of Distributed Systems SYSMAN (7026) and IDSM (6311). The SysMan Esprit project is defining notations and domain based tools for the specification of both authorisation policy which specifies what a manager is permitted to do and obligation policy which defines what management activities a manager must (or must not) do. This work will be applied to the automated management of network services within the BT project.
The group have extensive experience in the provision and use of languages, execution environments and general tool support for the design and construction of parallel and distributed systems. Members of the distributed software engineering section have been involved in many collaborative projects with industry within the UK and Europe. They are world leaders in two key areas of distributed systems: configuration and distributed systems management.
Jeff Kramer is a Reader and Director of Studies in the Department of Computing. He is head of the Distributed Software Engineering research section. He was principal investigator of the TARA project on Tool Assisted Requirements Analysis, and of the various research projects which led to the development of the Conic environment for distributed programming. More recently, he was the technical director of the REX Esprit Project. He is co-author of a book on Distributed Systems and Computer Networks.
Jeff Magee is a Senior Lecturer in the Department of Computing. He has been a principal investigator in the various research projects which led to the development of the Conic environment for distributed programming, and in the REX project which developed Conic's successor. He is currently a principal investigator in the SYSMAN and IDSM distributed systems management projects. He has authored many papers relating to configuration management and to the design and development of parallel and distributed systems.
Morris Sloman is a Reader in the Department of Computing. He was technical director of the DOMINO project, and is currently technical director of SysMan and chairman of the Technical Committee co-ordinating the joint work between SysMan and IDSM. He has been member of the BSI Open Distributed Processing Committee and is a member of the SERC Communications and Distributed Systems Committee. He is co-author of a text book on Distributed Systems and Computer Networks and has authored many papers relating to configuration and management of distributed system.
Kevin Twidle is a senior research fellow in the Distributed Software Engineering Section. He has been involved in the Conic, REX, Domino and SysMan projects and has co-authored a number of papers relating to configuration and management of distributed systems.
Dr. Morris Sloman
Imperial College of Science Technology and Medicine
Department of Computing
180 Queen's Gate
London SW7 2BZ, U.K.
Phone: + 071 589 5111 Ext 5041
Fax: 071 581 8024
Email: mss@doc.ic.ac.uk
LANC background
For the past three years the main focus of the group's research has been on distributed multimedia systems. Recent projects have included: i) the ESPRIT OSI 95 project; ii) the SERC funded Multimedia Network Interface, Zenith and TEMPO projects (all in collaboration with BT Labs.); iii) a SERC Multimedia requirements project in collaboration with ICI.
Currently running projects are as follows:- i) the SERC QoS-Architecture project (in collaboration with Netcomm Ltd) which is developing a Quality of Service Architecture for multimedia applications over high speed networks and in which BT Labs are associated by funding a CASE PhD student; ii) the SERC SMSS project (in collaboration with Olivetti Research Ltd.) which is extending a prototype multimedia storage server developed at Lancaster to provide a scalable service through the use of co-operating servers; iii) work on extending the Chorus microkernel to support continuous media in an ODP environment (funded by SERC and France Telecom); iv) a new SERC funded project on multicast services for continuous media.
Other relevant work includes the development of hardware systems based around intelligent I/O devices and cell based switching technology within workstations. This will allow closer coupling with cell based networks such as B-ISDN and ATM systems. A particular area of interest is the use of AAL 5 conformant I/O devices.
LANC profiles
Doug Shepherd extended his research in communications into the area of multimedia distributed systems whilst at the IBM Networking Center in Heidelberg. This is now his main area of interest. He is currently Director of Information Systems Policy for the university. Professor Shepherd is pioneering the introduction of ATM as a campus backbone network. He is also chairman of the UK SERC/DTI Communications and Distributed Systems Club.
David Hutchison is currently a professor and Head of the Computing Department. He has been seconded part time by the SERC to the DTI as a technical consultant in Communications and Distributed Systems (CDS), and was an official on the JFIT CDS Committee from 1988 until 1992. The SERC/DTI work involved assessing research applications from academics and industrialists, advising committee members and officials on technical matters, and preparing cases for new collaborative R&D programmes. From September 1993 he will be an academic member of the CDS Committee.
Gordon Blair is currently a senior lecturer in the Computing Department at Lancaster University and has been actively involved in research in distributed systems for the last eleven years. He has been responsible for a number of research projects at Lancaster in the areas of distributed systems and multimedia support and has published over seventy papers in his field. He was an invited expert on the Basic Research Actions Workshop for ESPRIT III. His current research interests include distributed multimedia computing, operating system support for continuous media, the impact of mobility on distributed systems and the use of formal methods in distributed system development.
Andrew Scott has been involved in several research projects with the Distributed Computing Group at Lancaster. The first of these projects examined the application of object-oriented computing to distributed process control applications. He has worked on several contracts to develop industrial applications, especially in robotics and process control systems. His research over the last five years has largely concentrated on distributed multimedia systems and the hardware/software needed to provide multimedia network interfaces. He is the joint holder of a number of SERC grants and is currently involved in the DELTA JITOL project.
Geoff Coulson has collaborated closely with CNET, Paris, on the use of Chorus to support distributed multimedia applications. He has also had a significant involvement with the research work in the Esprit OSI 95 project at Lancaster and with the Touring Machine group at Bellcore, New Jersey, USA. Geoff Coulson has published a wide range of papers in his area and has co-authored conference papers for ACM SIGCOMM, IEEE Infocom and the ACM/IEEE International Workshop on Network and Operating System Support for Digital Audio and Video.
LANC contact details
Professor David Hutchison
Computing Department (SECAMS),
Engineering Building,
Lancaster University,
Lancaster, LA1 4YR
email: dh@comp.lancs.ac.uk
tel: 0524 65201 x3802
fax: 0524 593608
LUT background
LUT (Electronic Engineering Department) have been involved in multimedia network research for in excess of ten years. Video services were developed for the UNIVERSE and UNISON networking projects (SERC and ALVEY funded) and a network monitoring system was developed for the Megastream based UNISON project. Work has also been undertaken in the area of packet video characteristics (BT funded). Much of the work at LUT has involved both simulation studies and practical implementation of systems.
LUT is currently a member of the RACE funded CHARISMA consortium. This project is concerned with case handling and workflow over networks. Other UK participants include Rutherford Appleton Laboratories (RAL), Buckingham University and Oxford Brookes University (OBU). LUT are providing fibre based physical layer connectivity at 200Mbit/s for a PC based ATM card which will provide networked communications. In addition, LUT are developing M-JPEG based video for the project. This includes fundamental studies of packet video to allow image reconstruction in the face of lost cells.
A related project to that above is the SERC MASTER project. A low-cost multimedia workstation using video similar to that described above is under development with OBU. The architecture includes Transputers for error concealment, bandwidth management and segmentation/reassembly. Communication within the wide area is provided by ISDN.
There is also currently a SERC funded project into Call Monitoring for Charging and Network Management Purposes. This research has BT involvement and aims to develop an efficient statistical cell monitoring mechanism to enable a 'picture' of an ATM call to be produced by monitoring a subset of the cells. Data produced could be used for charging users or to assist in the policing function as an input to a policing algorithm. The work will culminate with trials over a high speed Orwell Ring network developed by LUT with BT funding. This network supports a range of multimedia services.
LUT profiles
David Parish leads a team of 7 researchers and has worked with BT on a number of research projects over a period of nearly 10 years. He has supervised the UNISON network monitoring activity and the development of a variable bitrate video codec for BT. He leads all of the current research projects described above and has also experience of communications applied to other fields including vehicles and pipelaying. Recently he has developed a multidisciplinary course in Systems Engineering for British Aerospace. He is also involved with the organisation of the future EUSIPCO conference to be held in Edinburgh.
Mike Woodward has interests in network simulation and performance analysis of network access protocols. He is involved with the investigation into the monitoring of ATM calls described previously and is shortly to publish a book in the area of network performance.
LUT contact details
Dr. David Parish
Dept. Electronic Engineering
Loughborough University of Technology
Loughborough
email: D.J.Parish@uk.ac.lut
tel: 0509 222832
fax: 0509 222854
CAMB Background
The University of Cambridge Computer Laboratory has been active in communication research in particular ATM and distrbuted multimedia since the late 1970's. More recent and ongoing projects have included the Pandora Project (digital video to the desktop) in collaboration with Olivetti Research partially funded by the SERC; the Fairisle ATM Switch Project in collaboration with HP partially funded by the SERC; the Fairisle Extension (an ATM network used for experiments in call acceptance control and performace) in collaboration with BTL; the Pegasus project (an ESPRIT Basic Research Activity) in collaboration with the University of Twente investigating fundamental operating system requirements and design for the support of open distributed multimedia applications. Smaller projects funded by the SERC include the Attachment of Continuous Media Devices Using ATM and Operating System Support for the Desk Area Network.
As far as we know, we are the only university department in the UK to have designed, built and used our own ATM switches. The concept of the Desk Area Network (DAN) was invented in the Laboratory. Commercial exploitation of some of these ideas has already taken place. The Pegasus Project is constructing open platforms on modern workstations on which multimedia applications will be built. Pegasus will use an ATM network as its local (and possibly wide area) network.
The Laboratory also has a great deal of experience with the ANSA distributed systems architecture which is having a profound effect on the thinking within the ODP and telecommunication communities.
The Laboratory also has an active group in security, in particular authentication protocols.
Cambridge Profiles
Roger Needham has a wide variety of research interests including communication operating systems and security. Of particular interest here is his work in authentication protocols where he has published a number of papers including work on logic systems for examining the properties of authentication protocols.
Ian Leslie has worked on a number of large communication collaborations including Universe, Unison, and the MultiMed RACE program. His first work on multiservice networks was done in 1979 in an experiment with voice on the original Cambridge Ring. He is currently working on operating systems which support quality of service guarantees to applications; protocols for ATM networks; and host interface design.
Derek McAuley interests include operating systems, distributed systems, and ATM networks. He has an extensive knowledge of remote procedure call systems (in particular ANSA) and is a pioneer of the Desk Area Network.
Cambridge Contact Details
Dr Derek McAuley
University of Cambridge
Computer Laboratory
Pembroke Street
Cambridge CB2 3QG
Phone 0223 334416
Fax 0223 334678
Email Derek.McAuley@cl.cam.ac,uk
Oxford Brookes Background
The Distributed Systems Research Group has worked on a number of ATM-related projects over nearly a decade and is therefore well placed to work in the area of the SuperJANET initiative.
The group has a number of successful contracts with BT Laboratories for two years (1991-1993). The work has been close to the ATM standardization process and the experimental results have been input to the standardization committees.
In addition to the projects carried in conjunction with BT, the group has worked on the Alvey-funded UNISON project (1984-1988), MultiMed, which was partially funded by the CEC under the RACE programme (1988-1991), MASTER, a teleconferencing project funded by SERC (1990-1993) and CHARISMA, a RACE II project which is developing a multimedia case server (and client) for use on an ATM network.
Over the last two years, the Distributed Systems Research Group has conducted a series of investigations into congestion control within ATM telecommunications networks. The group has used a fully connected, re-configurable transputer array, programmed in occam for this work, and has found it to a flexible platform well suited to these types of investigation.
Oxford Brookes Profiles
Oxford Brookes Contact Details
Dr A.G Tagg
School of Computing and Maths
Oxford Brookes university
Gypsy lane
Headington
Oxford OX3 0BP
Tel: 0865-483651
Fax: 0865-483666
Email: P0054531@cs3.oxford-brookes.ac.uk