![]() |
||
| Structure | ||
|
|
|||||||||||||||||||||||||||||||||
|
Computer scientists understand much about structuring computer systems. In DIRC, our interdisciplinary approach has enabled us to extend such research and to find ways to apply similar thinking to wider computer-based systems. Dependable structures must limit the propagation of errors. In technical systems, we are familiar with internal checks and exception handling. In human systems we see layers of judiciary (courts, appeal courts and the Lords), processes such as financial audits etc but it appears that society all too often devises checks and boundaries only in response to failures. DIRC research is seeking ways to design and control detection/correction boundaries. This approach is essential when considering a complex computer-based system like a hospital. Strongly related to this is DIRC's study of the role of “processes”. Computer programs are the ultimate process but many human actions are also governed by procedures – a pilot landing an aircraft follows a detailed check list. Where computers excel at slavishly following instructions, people are often able to see when a set procedure fails to anticipate a possible exception and take appropriate action. A successful computer-based system has to utilise the strengths of both computers and people.We have made significant progress in research on determining the specifications of control systems. Once an initial specification is agreed, there are formal approaches to designing programs which satisfy the specification but in many computer-based systems it is extremely difficult to get the right initial specification. We have published a method where a description of a wide system is used to determine the specification of what might be the software control component. All significant systems change; those involving humans (either as players or as external forces like legislators) undergo continuous evolution. It is a major challenge to design a system that can cope with reasonable evolution. DIRC’s researchers have investigated the role of generic systems which can be instantiated to particular applications as they clearly support evolution over some range. These ideas and others are explained in the book Structure of Computer-based Systems and Dependability. LinksTime bands Diversity Psychology and software development
References |
|||||||||||||||||||||||||||||||||
| Page Maintainer: webmaster@dirc.org.uk | Credits | Project Members only | Last Modified: 28 November, 2006 | ||