Scheduling Periodic and AperiodicTasks Using the Slack

This chapter discusses the problem of jointly scheduling hard deadline periodic tasks and both hard and soft deadline aperiodic tasks using xed-priority methods. A recently developed algorithm, the slack stealing algorithm developed by the authors, is introduced and its properties are discussed. The methods introduced provide a uniied framework for dealing with several related problems, including reclaiming unused periodic and aperiodic execution time, load shedding, balancing hard and soft aperiodic execution time, and coping with transient overloads. Over the last decade, scheduling methods have been introduced which allow for the design of real-time systems with predictable timing correctness. Moreover, these methods have become suuciently advanced so that many practical problems associated with these systems have been addressed successfully. The most complete theoretical results have been for the situation in which the system must process a signiicant number of periodic tasks, for example, tasks associated with monitoring in control systems. For this case, there are two popular approaches: (1) static or xed-priority algorithms, including the rate monotonic and deadline monotonic algorithms 10, 11] and (2) dynamic priority algorithms, including the earliest deadline algorithm 11]. Both approaches are becoming increasingly well developed, although at the present time the static priority theory is much more complete. For example, static priority methods can: handle transient overloads in a predictable fashion, 172 Sec. 8.1 Introduction 173 be combined with priority inheritance and the priority ceiling algorithm to prevent the unbounded priority inversions that can occur with task synchronization , permit the eecient joint scheduling of hard deadline periodic tasks and soft deadline aperiodic tasks, be analyzed to include the eeects of the operating system and hardware architecture on the timing behavior of the workload. The interested reader should consult a review article such as that by Burns 1] or Lehoczky 9] for a summary of the results available on xed-priority scheduling. Given the success of xed priority scheduling methods, it is natural to attempt to extend this theory to solve other important problems that arise in real-time systems. This includes problems such as simultaneously scheduling hard deadline periodic tasks along with both hard and soft deadline aperiodic tasks. Hard deadline aperiodic tasks are of special importance. They arise in a number of ways, for example, from alert conditions or from failures of hard deadline periodic tasks which fail the reasonability checks to validate their results and must be retried and completed before the original deadline …