Dual priority scheduling

In this paper, we present a new strategy for scheduling tasks with soft deadlines in real-time systems containing periodic, sporadic and adaptive tasks with hard deadlines. In such systems, much of the spare capacity present is due to sporadic and adaptive tasks not arriving at their maximum rate. Offline methods of identifying spare capacity such as the Deferrable Server or Priority Exchange Algorithm are unable to make this spare capacity available as anything other than a background service opportunity for soft tasks. Further, more recent methods such as dynamic Slack Stealing require computationally expensive re-evaluation of the available slack in order to reclaim such spare capacity. By comparison, the Dual Priority approach presented in this paper provides an efficient and effective means of scheduling soft task in this case.