Periodic and aperiodic task scheduling in strongly partitioned integrated real-time systems

Abstract

To facilitate the integration of real-time applications in a common platform, temporal and spatial partitioning should be provided. A strongly partitioned integrated real-time system (SPIRIT) is reported in this paper that adopts a two-level hierarchical scheduling mechanism to ensure temporal partitioning. At the lower level, multiple partitions (applications) are dispatched under a cyclic scheduling, whereas, at the higher level, multiple periodic tasks of a partition are scheduled within the partition according to a fixed priority algorithm. The proposed Distance Constraint guaranteed Dynamic Cyclic (DC2) scheduler applies three basic operations, left-sliding, right-putting and compacting, to dynamically schedule aperiodic tasks and, in the meantime, guarantees the distance constraint characteristics of a partition cyclic schedule. In addition, the slack time calculation of these dynamic operations can be applied for scheduling hard aperiodic tasks. With simulation studies, we observe that the DC2 algorithm can result in a significant performance enhancement in terms of the average response time of soft aperiodic tasks and the acceptance rate for hard aperiodic tasks.