An adaptive overload management strategy for periodic real-time tasks

Abstract

By analyzing the relationships of preemption and abortion among the jobs of the real-time tasks, the overload management strategy (OMS) divides the jobs into two types: steady-finished-jobs and absolute-aborted-jobs. While scheduling the jobs, the steady-finished-jobs are released and the absolute-aborted-jobs are rejected. In this way, the cascaded preemption among the jobs could be avoided. Also, the system resource could be efficiently utilized. The results of the simulation show that, the OMS could adaptively deal with the static or dynamic task set. Additionally, by applying OMS, the scheduling algorithms can obviously improve the performance of the real-time system.