Application specific performance indicators for quantitative evaluation of the timing behavior for embedded real-time systems

In the design and development of embedded real-time systems the aspect of timing behavior plays a central role. Especially, the evaluation of different scheduling approaches, algorithms and configurations is one of the elementary preconditions for creating not only reliable but also efficient systems - a key for success in industrial mass production. This is becoming even more important as multi-core systems are more and more penetrating the world of embedded systems together with the large (and growing) variety of scheduling policies available for such systems. In this work simple mathematical concepts are used to define performance indicators allowing to quantify the benefit of different solutions of the scheduling challenge for a given application. As a sample application some aspects of analyzing the dynamic behavior of an combustion engine management system for the automotive domain are shown. However, the described approach is flexible in order to support the specific optimization needs arising from the timing requirements defined by the application domain and can be used with simulation data as well as target system measurements.