An optimal algorithm for scheduling soft-aperiodic tasks in fixed-priority preemptive systems

A novel algorithm for servicing soft deadline aperiodic tasks in a real-time system in which hard deadline periodic tasks are scheduled using a fixed priority algorithm is presented. This algorithm is proved to be optimal in the sense that it provides the shortest aperiodic response time among all possible aperiodic service methods. Simulation studies show that it offers substantial performance improvements over current approaches, including the sporadic server algorithm. Moreover, standard queuing formulas can be used to predict aperiodic response times over a wide range of conditions. The algorithm can be extended to schedule hard deadline aperiodics and to efficiently reclaim unused periodic service time when periodic tasks have stochastic execution times.<
>