Evaluation and comparison of task allocation and scheduling methods for distributed real-time systems

Automatic tool support for scheduling of distributed real-time applications on various platforms is of critical importance both to guarantee critical application demands, such as safety requirements, as well as to keep development costs down. It is, however difficult to choose a proper scheduling technique from the plethora of more or less practical approaches that are available, as means for objective evaluation and comparison of scheduling approaches are lacking. In this paper we propose a framework for description, evaluation and comparison of methods for task allocation and scheduling on distributed real-time system. By using this framework it is possible to perform studies, both of existing techniques as well as of new combinations of existing techniques. Our framework uses a decomposition approach wherein existing scheduling methods have been analyzed and broken down into a set of independent parameter choices.