Systems Integration Initiative

1st Community Workshop - A Summary

 

Mark Wilkins, ESPRC IT & CS Programme Manager and Vince Osgood, EPSRC Engineering Programme Manager, opened the day by summarising the aims of the programme (see the 2nd call for proposals) and describing the industrial developments which motivate the initiative. (A downloadable version of the EPSRC PowerPoint presentation is in preparation.)

They also pointed out that we are now seeking to enrich the programme with a wider range of types of project, such as Feasibility Studies, Networks, Visiting Fellowships and associated Responsive Mode projects.

Bill Bardo, Technical Director of GEC-Marconi, and chairman of the Foresight Aerospace & Defence Panel, explained the concerns of Foresight about UK Systems Engineering, with tighter integration of contributors to major systems projects in an increasingly fluid environment, which inspires their support for the initiative.

Bob Malcolm sketched the outcome from the first call for proposals in the initiative, and then introduced presentations from the five successful teams in the first call.

Three industrial speakers opened the after-lunch session, to explain what they wanted of academic research in systems integration, and, in some cases, what they did not want.

John Gallagher, of Nestlé, gave a sometimes hilarious yet nevertheless very serious exposition of the problems of integrating factory shop-floor information with business systems. In particular he emphasised that data integration does not provide information integration - which is what they seek. This drew out the need to recognise the important role of context in the interpretation of data, and the difficulty of capturing and using context in this way.

Colin Piddington, of Computer Sciences Corporation (UK), described the needs of aerospace manufacturing, a sector which has gone through the Extended Enterprise and is now moving into the Virtual Enterprise. He set a number of challenges, not the least of which is the need for any academic research to be relevant in their context of projects with 50-year cradle to grave (recycling plant) life, while the technology transforms possibilities every two or three years, and industry restructures almost as frequently. (Download 17KB a zipped version of Colin's PowerPoint presentation.)

Tony Morgan of Unisys then gave the academic software engineering and computer science community a stirring call to arms - to 'get real'. He pointed out that industrial software concerns lie closer to those of Flexible Manufacturing Systems than the topics typically addressed in software engineering research. (As a former member of the Alvey Directorate, Tony speaks from a position of knowledge.) His position paper and presentation (download 11KB) describe the mismatch as he sees it and suggest appropriate research issues.

The discussion helped clarify some of the aims of the programme. In particular ...

The phrase 'heterogeneous systems of systems', from the call information, is a good indicator of the class of system with which we are concerned: where integration cannot wait for open architectural standards to cope with the heterogeneity. (Though the solutions themselves could well form the basis of such standards.)

The fact that a number of the successful proposals are concerned with supply chain management should not be taken as an indication that this is the main area of concern of the initiative. We are equally interested in embedded, operational, and manufacturing systems. In particular, technology integration has received less attention than we expected, though this may be accounted for by the (quite proper) focus on 'process' - one cannot look at any of these integration issues without considering the necessary changes in the contextual processes.

Also, 'operational' systems include complex integrated information systems (see the examples associated with the call information). A specific recurring theme in discussions about complex information systems engineering is the importance of context - as exemplified in John Gallagher's talk. A phrase being bandied about is 'semantic integration', to convey the differentiation from data integration and to avoid confusing 'information integration' with data integration. Nobody is quite sure what 'semantic integration' means, or if it is possible, but everybody wants it. This is a clear topic of research which would seem well-suited to the IT & CS community.

So the absence of projects with significant involvement of the IT & CS research community does not mean that information systems engineering is outside the scope of the initiative. One hypothesis is that it does reflect the 'disconnect' between academic research and industrial concerns. Apart from Tony Morgan's talk, this subject has received considerable airtime in recent EPSRC strategy workshops and in the Software Engineering Association mail-list. About a third of the proposals submitted in the first call had an 'IT flavour', but there was a greater preponderance of proposals from this community that failed to follow the guidelines. Proposals from the IT & CS community were in general less well-formulated than those from the manufacturing engineering community.

This is not the place for a full discussion of the issues, but the purpose for stimulating discussion at the workshop was not to discourage IT & CS researchers, but to encourage serious thought about the appropriate nature of research in this field. As Tony Morgan said, industry does not want more research on programming languages: researchers need to identify what is required. The rapid development timescales of the IT industry does not preclude research either. Automotive manufacturing has undergone dramatic changes in recent times and the associated research community is very actively supporting those changes and shaping future change.

We are keen now to redress the balance in the programme and we would like to see more proposals from the IT & CS community, while taking the considerations above into account.

It was reiterated that the desire for industrial relevance does not mean that we are not interested in theory. Ultimately we want to understand better how to design and build complex systems - for this we need theory. But we want the theoretical quest to be motivated and focused by the real industrial problems.

The term 'industry' is shorthand: we include public sector organisations, such as health administration, which is facing serious industrial scale information integration problems.

In similar vein, while the first successful proposals are concerned primarily with products and their manufacture, we are equally interested in services. Indeed, an emerging theme in a number of projects is a shift of focus from products and product development projects to services and service evolution. Essentially this is a change of mind-set to support continuing flexibility and to avoid freezing the features of a system. (The death knell of the System Requirement Specification as we know it?)

In response to the observation that so many of the first call proposals failed to address the criteria, and appeared to ignore the additional guidance, it was suggested that we should mandate that the first draft of the proposal must be written by industry. This would be impossible to implement - but it is certainly recommended. What is required is certainly much closer to an industrial business case than a conventional research proposal. The industrial partners on some projects are going a significant step further and putting in their own project managers. Again this is warmly welcomed.

The question of participation of non-UK organisations arose. There is a misunderstanding in some quarters that this is frowned upon. On the contrary, we have repeatedly said that we seek world-class research, and projects should be deploying the best brains in their field, from wherever. This can be achieved through fellowships, workshops, exchanges, or even plain correspondence. Similarly the industrial market leaders in a project's application domain may not be UK companies. Maximum impact with any results may be best achieved by involving foreign companies. Obviously care needs to be taken with IP protection, but this is the case anyway.

The points which arose from this discussion were used to revise and clarify the information for the second call.

Following the workshop, Bob Malcolm and the EPSRC Programme Managers and Assistant Programme Managers stayed behind for over an hour to run an informal 'surgery' to advise potential proposers.


GO TO .... | ideo home | SII home | EPSRC SI programme home | SII call for proposals | SII proposers' guidance | SII contacts