Synthesizing schedules through heuristics for hard real-time workflows

Task assignment and subsequent schedule synthesis in distributed real time systems is a problem that arises in many fields of industry, such as factory automation or the automotive or avionic sector. Monolithic and bus-based approaches, while easy to schedule in the aforementioned context, are giving way to more flexible hardware environments, either because of increased pressure for flexibility (factory automation) or hardware consolidation (automotive and avionics). This paper presents an approach for synthesizing schedules of a distributed hard real time workflow based on existing heuristics. This approach can find a solution for over 85% of all feasible tested system configurations while being orders of magnitude faster than an approach based on a satisfiability solver. We obtained these results by simulating over 1 million different workflows and system configurations.