A multi-agent based modeling and execution framework for complex simulation, control and measuring tasks

The paper presents a modeling concept and a supporting runtime environment, which enables running simulation, control and measuring (data processing) tasks on distributed implementation platforms. Its main features: (1) it is scaleable in various application domains; (2) it has a model based system architecture; (3) it has a direct link to a real-time fast prototyping environment; (4) it is open to integration with external (e.g. legacy) systems. The approach followed relies on a multi-agent based modeling framework, which decomposes the problem in hand into highly autonomous entities. These entities are communicating via abstract sensors and actuators. This modeling concept also enables the fully distributed execution of the system designed via a properly designed backbone data model and supporting runtime environment. As a result measuring (data processing, control) systems ranging from single-processor to networked symmetric multiprocessor implementations can be built up using the proposed architecture as a backbone. To target real-time concerns was an explicit issue allover the development. Open standards based components provide great degree of hardware and software platform independence.