Computer control systems are increasingly required to be highly dependable and to have deterministic timing properties. Distributed architectures have the potential to meet this challenge. The advantages of distributed computer control systems include the possibility of composing large systems out of pre-tested components with small integration effort, their well-defined fault containment properties and their capacity to make effective use of mass-produced silicon chips.
The IFAC Workshop series on Distributed Computer Control Systems (DCCS) highlights and traces the growth of key concepts in this field at their various stages of development. Theoretical and practice-oriented viewpoints receive equal emphasis and there is a creative blending of the disciplines of computer science and control engineering.
The 1998 DCCS Workshop was notable for the attention given to true real-time communication networks and protocols. The complexity of the trade-off between services, dependability mechanisms and system-level properties was highlighted, and rigorous modelling and analysis methodologies were discussed. Event-triggered and time-triggered protocols were contrasted.
Models for analysing and predicting response times in distributed systems and for predicting the effect of response-time jitter on the performance of feedback control loops were presented.
The application of formal methods to the specification and development of safety-critical control software also received much attention. Distributed object methodologies and object request brokers were also highlighted as being promising approaches for the programming of large-scale, heterogeneous distributed systems.
Applications reported included control systems for traffic lights, jet engines, automobiles, fully-automatic trains and flexible manufacturing systems.
