Performability Comparison of Schedulability Conditions in Real-Time Embedded Systems

In modern safety-critical real-time embedded systems, performance and reliability are two most important parameters. The joint consideration of these two parameters is called performability. This paper evaluates and compares different schedulability conditions, which are used for scheduling tasks with fault-tolerant Rate-Monotonic (RM) algorithm, with respect to their performability. The task re-execution scheme is used to implement fault tolerance in the RM scheduling algorithm. To evaluate the schedulability conditions, a software tool is developed to simulate an embedded processor with m schedulable periodic tasks. The results show that among five different schedulability conditions used in this evaluation, the best schedulability condition has on average 3.5% higher performability than the time the other conditions are used. Also, the simulation results show that the next best schedulability condition can on average increase the performability of the embedded processor 2.5% more than the other conditions.