Professor Jack Howlett is a consultant with ICL Limited. He is acknowledged as one of this country's leading experts on communications. He also edits ICL's Technical Journal. Prior to joining ICL he was the director of the Atlas Computing Laboratory belonging to the Science Research Council at Chilton in Oxfordshire.
I spent the two weeks from August 21 to September 3 in a relatively unknown part of South West France, as co-Director with Dr. Kenneth Beauchamp of Lancaster University of a NATO Advanced Study Institute on Information Technology and the Computer Network. This was the third that we had run on the general theme of computer communications and networking, and before going any further let me say something about these ASIs, as they are always referred to.
NATO is, of course, a military alliance between the USA and some 14 countries mainly in Western Europe but including Greece and Turkey. With what continues to strike me as remarkable broadmindedness and farsightedness, obligations to non-military, civil science were written into the treaty and are embodied in the NATO Civil Science Committee; this was established in 1958 with practical action guaranteed by the setting up of the International Scientific Exchanges Fund, ISEF. An important part of the committee's activity is the ASI programme; in the committee's own words, The purpose of the Advanced Studies Institutes Programme is to contribute to the dissemination of advanced knowledge and the formation of international contacts among scientists. An ASI is primarily a high-level teaching activity at which a carefully-defined subject is presented in a systematically and coherently structured programme. The presentations are made to other scientists already specialised in the field or possessing an advanced general scientific background.
The committee goes on to say that most fields of science, including the social sciences, are eligible for support, that an ASI will be a relatively small gathering. ideally of between 70 and 100 participants, lasting two to three weeks and held in a secluded and quiet location with all the participants accommodated, if not literally under one roof, at least all on one site - a university campus, for example. These domestic conditions are imposed in the interest of ease and informality of discussion and collaboration; having become thoroughly disenchanted with the huge international jamborees, with their thousands of delegates and bewildering programmes of parallel sessions, I applaud them heartily. One scarcely needs to add that there are no security restrictions whatever; in fact, everything is published openly: NATO are very insistent on rapid publication and have made special arrangements with a number of publishers.
Clearly, the AS! programme is an important source of support for specialised scientific meetings. The total number held since its inception is around 250 and they do indeed cover the full spectrum of pure and applied science: Particle Physics and Animal Husbandry might serve to indicate extremes, or Pattern Recognition and Engineering Mechanics. In 1977 Kenneth Beauchamp - who, by the way, is Director of Computing Services at Lancaster and has published several books on signal processing - asked me to collaborate with him in organising one on a subject he felt to be of great and growing importance, the interlinking of computer networks. This was to be held in the Chateau de Bonas in the Gers Department of France - the relatively unknown place I referred to in my opening sentence.
We recruited (on paper) a team of lecturers and put together a programme, submitted this to NATO and got their support - meaning, approval to go ahead and a grant to finance a meeting in the summer of 1978. It proved very successful, with all the interaction between participants one could have wished, and ended up with a strong recommendation that there should be a successor meeting in about three years' time - by which time there are likely to be commercial realisations of some of the possibilities now opening up and commercial solutions to some of the problems. Here, possibilities meant actual connections between real, working networks and the problems were mainly those of interface standards and protocols.
In 1980 therefore we set about planning and organising a second meeting, this time choosing the title New advances in distributed computers to allow for a rather broader coverage. Again we got support and the meeting was held, again at Bonas, in June 1981. A great deal had happened in the field of computer communications and networking in the intervening three years, and I said in my introductory survey that I, at least, felt that we were in a different world from that of 1978.
The ideas of the layered architecture for a communications system, in particular the ISO 7-layer model, and of the local area network (LAN), in particular the Ethernet and the Cambridge Ring, had come into prominence. We had papers on these, on standards and protocols, on mathematical modelling of distributed and communicating systems; and a group of papers by Wyn Price of NPL on security and encryption which dealt with, among other topics, the recently-announced Data Encryption Standard of the US National Bureau of Standards and the ideas underlying public key encryption. We also had a session on the penetration of the concepts of distributed processing into the domain of computer architecture, with the consequential development of what are now being called non-von Neumann architectures, such as array processors and data-flow machines, the importance of which, by the way, was emphasised in the Japanese Fifth Generation Report.
Once again the participants recommended a successor meeting, stressing the fast pace at which the subject was moving and its importance increasing; and the by now well-established firm of Beauchamp & Howlett (not without some feelings on the part of the Howlett half of pushing our luck) made a third submission to NATO. This was approved and resulted in the 1983 meeting at Bonas, where by now we were feeling thoroughly at home, with the title I gave at the start. Let me now summarise this meeting.
One of the most striking events in the Information Technology world of the last few years has been the Japanese announcement of the Fifth Generation Computer programme. This has stimulated questioning and discussion of national programmes all over the world, and one outcome in Britain has been the Alvey study, followed by the setting up of the Alvey Directorate with its £350 million programme for IT developments involving industry, research establishments and academic bodies in collaboration. This must surely count as the strongest acknowledgement ever made by any British government of the importance of the subject.
We had included a discussion session on the Alvey activity in our programme, and in the event we had the great good fortune to have Brian Oakley; Director of the Alvey Programme, to tell us about the aims and current state of this. He was actually holidaying in southern France at the time and most generously offered to break into his holiday to spend a day with us. He outlined the main heads under which the programme is organised: VLSI, CAD (for IC design), Software Engineering, Expert Systems and Interactive Knowledge-Based Systems (IKBS), Man-Machine Interface (MMI). Infrastructure ( eg a communication network and a mailbox service) and Information dissemination. Directors for these have now been appointed, several on secondment from industry - a welcome state of affairs. He ventured the view that there is a revolution coming in IKBS, though the timescale isn't clear; and that in the MMI field, whilst considerable: advances in voice input and output could be foreseen, the achievement of computer understanding of normal speech was a very long way off.
A significant part of the time at the ASI was given to factual reports on some of the main networks now in operation or in an advanced state of planning, with emphasis on standards and protocols and on management issues. Quite severe management problems are thrown up and the participants said they found it valuable to hear of how these were being tackled. Giraudbit (Paris) spoke about the SITA international airlines' network and Kuo (SRI International) about developments in the USA following on the pioneering ARPANET. Participants from the British university community had a lot to contribute here: with the stimulus and support of the Computer Board and the Research Councils - especially the Science and Engineering Research Council, SERC- a great deal has been done here in the provision of networks both for services and for research into networking techniques and technologies. Several very extensive networks serving groups of universities are in non-stop operation and have become an essential part of the UK university infra-structure - for example, the SW Universities Computer Network, described by the Director, Jim Brooks; and plans for a national network linking all the universities are well under way. Christopher Adams of SERC's Rutherford-Appleton Laboratory described the very interesting UNIVERSE project, in which ring LAN technology and the European Space Agency's Orbital Test Satellite are combined in an experiment on high speed communication between computers at seven sites: the universities of Cambridge, Loughborough and London (University College), SERC (Rutherford Appleton Laboratory), British Telecom (Martlesham Heath Research Laboratories), GEC (Marconi Research Laboratory; Chelmsford) and Logica, a very welcome collaboration between the academic and industrial worlds.
Local Area Networks have developed at a tremendous rate during the past few years: the term hadn't been invented at the time of the 1978 meeting. Technological advances such as in LSI, VLSI and optical fibre transmission lines have made it possible to produce very high speed systems at relatively low cost and many manufacturers are now offering such systems as standard products. The LAN is now becoming the standard means for linking all the information gathering, processing and transport equipment on a defined site such as a university campus, an office complex or a factory. Problems of interface standards and protocols are now becoming important, because of the need to link groups of LANs together and to link them to wide area networks (WANs) such as the PTT public systems. Danthine (Liege) spoke about the interlinking of LANs and also about the problems of assessing performance, and Bannerjee (Orbis/Acorn) about the possibilities for slotted-ring LANs of much higher speed than those now in use. The topic of standards in networking and in information transport generally was the subject of several papers - for example, on the current activities of the relevant committees and working groups of ISO and CC ITT; but it entered into every topic discussed, either explicitly or implicitly. The virtually completed acceptance of the achievement of Open Systems Interworking as a goal whose attainment will benefit everyone, including manufacturers, has made the definition and implementation of interface standards and protocols a matter of prime importance and the subject of a great deal of international collaborative activity. A paper by Helms (EEC Research Centre, Ispra) included a valuable directory to this activity.
Several other topics were discussed, performance prediction for example, but I will mention only one more. Wyn Price of NPL again gave us, in effect, a short course on data encryption and security measures, up-dating what he had told us two years previously. The need for protection of cipher keys in transit and for the authorisation of messages - that is, of providing assurance that a message has actually come from the source from which it purports to have come - have risen in importance, and some very sophisticated means have been devised to satisfy these needs. With the world's banking and other financial institutions now practically completely dependent on telecommunications for their operations, which can include the transfer of very large sums of money by coded messages, the need for the most stringent protection of these messages is only too clear.
And finally, Bonas. This is a minute hamlet in the Gers, a Department that starts just south of the Garonne and finishes a little north of the Pyrenees; no large towns, rolling green countryside, very agricultural, very French. It had a colourful history but fell into a decline in the present century; becoming almost a ruin in the 1960s. In 1972 it was bought by Professor J-C Simon who has a Chair in computer science at the Pierre et Marie Curie Universite in Paris ("Paris-VI") and he and his artist wife Francoise d'Origny restored it to its present charm and elegance. They formed the idea of making Bonas a meeting place for scholars and a centre for seminars and similar gatherings, and set up what they called the Association Scientifique, Educationale et Culturelle de Bonas-ASCEB. It has been a regular venue for NATO ASIs since 1974. As I have said, Beauchamp and I have held three meetings there over the past five years, and we have found it ideal.
Jack Howlett
