19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Workshop 2
Preemptive Behavior Analysis and Improvement of Priority Scheduling Algorithms
Denver, Colorado
April 04-April 08
ISBN: 0-7695-2312-9
Xiaoying Wang, Northeastern University, P.R.China; Changshu Institute of Technology, P.R.China
Hai Zhao, Northeastern University, P.R.China
Most of embedded real-time systems only deploy necessary resources so that the extra preemption overheads among tasks debase the system performance terribly. Through scheduling process analysis of periodic task, this paper presents the waiting limit formula of each task in ready queue while guarantees its deadline. In addition, some properties such as final preempt time is deduced and the necessary condition of preemptive behavior of periodic tasks is quantitatively described. Based on them, a micro scheduling preempt model for periodic tasks in ready queue is put forward, which decreases preempt amount and optimizes system performance through change preempt relationship. The model cannot only decreases preempt amount effectively but enhances processor utilization for static priority scheduling algorithm such as rate monotonic scheduling, which is demonstrated by case study and simulation.
Citation:
Xiaoying Wang, Hai Zhao, Wenbo Zhang, Zhenyu Yin, "Preemptive Behavior Analysis and Improvement of Priority Scheduling Algorithms," ipdps, vol. 3, pp.130a, 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Workshop 2, 2005