Systems Integration Initiative

The Systems Integration Initiative – a technical résumé

Through the mid-to-late 1990’s, a consistent theme of Technology Foresight panels was that systems integration – whether for military aircraft, construction, or health-care - is difficult. As a consequence, the Engineering & Physical Sciences Research Council launched in 1998 the Systems Integration Initiative. This is a major research programme in which around £20M has been committed to address the problems of complex systems integration.

While the original motivation for the Systems Integration Initiative was concern for the integration of complex artefacts - whether military aircraft or major construction complexes - it was soon accepted that integration of artefacts requires integration of the processes involved in their development. This in turn requires integration of the information systems which underpin the processes. And integration of processes ultimately entails integration of all the organisations involved in the ‘supply mesh’ (see 'Systems Integration 2005'). The initiative includes research into all these varieties of systems integration.

Heterogeneity

One way to simplify integration is to establish standards for architectures, components, and interfaces. However, while standards may eventually emerge, the fundamental problem is that of integration in the absence of standards. There is, therefore, an emphasis on heterogeneous systems of systems.

Research themes

The programme has attracted research from communities associated with manufacturing, information engineering, and organisation and management.

Starting with research coming from the manufacturing community, I see three main strands: technology integration, architectures for integration, and process integration. Technology integration has not been a strong theme in this programme, and I think it was never intended to be such. We are not addressing such things as electro-mechanical systems, chemo-electronic systems, or even how to glue plastics to metal. I think it was never our intention to work at those levels, however scientifically interesting it may be for other research programmes. Interestingly though, we have one project to explore the integration of models from different technological domains. This might expose a seam of interesting work on the integration of models from different domains. (This is not just a matter of coupling data channels together. There are some hard and quite philosophical problems concerned with the context in which data is used and interpreted.)

There has been no direct attack from the manufacturing community upon (technical) architectures for integration. Perhaps this was because such a line of work would be considered to be concerned with standardisation (consider ‘modular avionics’) or maybe because, given the wide diversity of technologies and organisations associated with the construction of real, physical systems, it was thought to be just too hard.

The main thrust from the manufacturing community has been on ‘process integration’. At its simplest, this has taken the form of integration of process models. This in turn has led to interest in the integration of the manufacturing information systems which embody those process models and support the relevant processes. Concerns about process integration have also inspired a quite separate line of research into business and organisational integration.

The information systems engineering community has concentrated upon architectures which facilitate integration and upon the problems of process integration. Research into architectures for integration is primarily concerned with resilience to change - that is to say support for evolution throughout specification, design, implementation, use, and maintenance. With regard to process integration, there is some research into ways to design the processes for information systems design and integration more effectively: this is based upon an analysis of the decision-making processes concerned with such activities.

One new and interesting line of research is in ‘service architectures’. Rather than viewing information systems as support for processes, the researchers are viewing the integration of information systems in the context of and as a means to deliver the integration of services for which the owners of those information systems are responsible. While this approach was conceived to support the integration of information systems, the conceptual framework which it provides may have applicability in other domains. It is, for instance, being considered as a way of articulating better the relationship between physical architecture and the functionality of buildings required by and provided by the organisations using them. It offers a way of analysing the interoperation of organisations - not just the relationships between discrete organisations, but the relationships between parts of an organisation.

Analysis of a system - or an intended system - in terms of ‘services’ is similar in many respects to analysis in terms of ‘processes’. Indeed, the richer views of ‘process’ are quite similar to those of ‘service’. However, the differentiation between process and service does seem to offer some advantages in terms of facilitating the analysis of a wider range of situations in terms of services. It has yet to be determined why this is so – perhaps the ‘process’ conceptual framework makes some presumptions about the nature of organisational systems, and can therefore be an unwelcome constraint; perhaps it is just that the ‘services’ terminology and conceptual framework are more intuitive and thereby facilitate mapping to real-world situations.

Integration versus evolution

Integration is unlikely to be a once-off event. The class of integrated systems of concern is characterised by continual evolution, whether of product, product range, or services, and restructuring of businesses, industrial sectors, and public services. Much of the research is therefore concerned with architectures which support evolutionary integration, and the corollary, the components which ‘plug in’ to an appropriate architectural framework and the connectivity of those components.

Researchers working on such themes then have an interesting problem with respect to ‘systems integration’. If they are successful in providing some framework within which one may more readily define, construct, and evolve systems, then the ‘integration’ problem is effectively ‘disappeared'.

The problems of systems integration with which we began arose from the difficulty of integrating bits of systems when inadequate forethought had been given to how those bits might be put together - whether that inadequacy was necessary or accidental. Earlier we had rejected standards and open architectures as a solution, on the basis that if we had such standards and associated architectures then we would not have the integration problem in which we were interested. So, is research which aims to ‘disappear’ the problem of systems integration still concerned with systems integration?

Back to EPSRC SI programme home


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