Priority Assignment for Event-triggered Systems Using Mathematical Programming
Martin Lukasiewycz, Sebastian Steinhorst, Samarjit Chakraborty · 2013
This paper presents a methodology based on mathematical programming for the priority assignment of processes and messages in event-triggered systems with tight end-to-end real-time deadlines. For this purpose, the problem is converted into a Quadratically Constrained Quadratic Program (QCQP) and addressed with a state-of-the-art solver. The formulation includes preemptive as well as non-preemptive schedulers and avoids cyclic dependencies that may lead to intractable real-time analysis problems. For problems with stringent real-time requirements, the proposed mathematical programming method is capable of finding a feasible solution efficiently where other approaches suffer from a poor scalability. In case there exists no feasible solution, an algorithm is presented that uses the proposed method to find a minimal reason for the infeasibility which may be used as a feedback to the designer. To give evidence of the scalability of the proposed method and in order to show the clear benefit over existing approaches, a set of synthetic test cases is evaluated. Finally, a large realistic case study is introduced and solved, showing the applicability of the proposed method in the automotive domain.