Preliminary Design of Models for the AIDA tool - set , Working document

This report describes a modelling framework that is to be used in the AIDA decentralisation tool-set for design and early analysis of control system implementations in distributed computer architectures. The modelling framework includes three basic groups of models: application, computer system and mechanics. Each model group is divided into three modelling levels: the structural, the behavioural and the functional. Furthermore, the modelling framework makes it possible to describe systems in different levels of abstraction—either detailed models are defined, or simpler model versions can be used. The application models describe the control system and are designed to be a complement to the control design in CACE tools like e.g. Matlab/Simulink. The models given here are more focused on implementation requirements. At the structural level, the modes of the control system are defined. Each mode is associated with a data-flow diagram and a number of timing and triggering diagrams. The timing and triggering diagram specifies the control design’s timing requirements including period times, tolerances to jitter, admissible control delays and precedence relations. A variant of the timing and triggering diagram is used to feed back results regarding achieved timing from an implementation analysis to the control designer. Also, event triggered activities are, to a limited extent, supported by the application models. The computer system models describe the computer hardware structure with its processing elements and communication links. The software parts of the system such as the implemented processes, their allocations, their timing and precedence relations and their communications are described. Also the operating system’s scheduling and synchronisation policies for the processors and communication links are defined. The mechanical models describe the physical placement of the processor nodes in the mechanical system. It is concluded in a discussion that the models will need some more extensions to support e.g. synchronous communication, specification of event triggered systems and error handling characteristics. However, the modelling framework should provide a good base for multi disciplinary development and thereby give possibilities to develop high performance systems at a minimum cost.