A Globally Optimal Algorithm for Scheduling Both Hard Periodicand Soft Aperiodic Tasks

Tasks that form the base load of a real-time system usually come with hard deadlines and need to be executed periodically. In addition to these periodic tasks, aperiodic tasks may be invoked randomly in response to environmental stimuli. These randomly-arriving aperiodic tasks usually come with soft deadlines, but their response times may still be important to the overall system performance. It is therefore important to schedule tasks so as to provide responsive services for soft aperiodic tasks without missing any periodic task deadline. In this paper, we propose an algorithm, called the last chance (LC) algorithm, that meets this objective. The LC algorithm is globally optimal in the sense that it provides the minimal aperiodic task response time while meeting all periodic task deadlines among all possible periodic and aperiodic task scheduling methods. Since the LC algorithm may require a signiicant amount of on-line computation, we also propose a suboptimal algorithm whose on-line computation requirement is low. Any opinions, ndings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reeect the views of the funding agencies.