AIDA II Automatic control in distributed applications

Machine industry faces a shift where more and more functionality is implemented in software on distributed computer systems. Many difficult problems then face designers including system modelling, analysis and management of multi disciplinary design teams. There is a need to transfer theoretical results to industrial practice, necessitating the integration of existing work from separate disciplines. Industrial practice, on the other hand, needs to be shifted towards model building, analysis and prediction in order to avoid clashes in system integration. Several useful modelling approaches and computer aided tools exist for the design of control, software and mechanical systems. These support tools however fall short when the applications are implemented in embedded and distributed real-time computer systems. A number of additional design issues then face designers including • Structuring, partitioning and allocation, i.e. how to establish structures and the mappings between them (e.g. functions/threads/processors). • Execution and communication policies. The definition of policies and mechanisms for triggering, scheduling, synchronisation and communication. • Performance requirements. The definition of timing requirements and related trade-offs. • Error detection and handling policies. In particular additional failure modes of distributed real-time systems must be identified and handled. The provision of a modelling framework that enables interdisciplinary work is a prerequisite in order to support these design issues.