容错实时任务调度的DSPN建模与分析

DSPN Modeling and Performance Analysis of Fault-tolerant Real-time Task Scheduling

摘要:

复杂系统的形式化描述对新系统的设计以及现有系统的改进与评价都具有十分重要的作用;针对处理机系统容错实时混合任务调度,提出采用确定与随机Petri网进行建模与性能分析;首先,根据任务执行的优先级、周期性、容错性和实时性,将任务分为四类；然后,采用DSPN对任务调度执行过程,不同优先级任务抢占式调度,处理机故障及故障恢复过程进行建模,由此构成处理机系统容错实时任务调度过程的DSPN模型；最后,仿真实验结果表明,在负载相同情况下,处理机利用率基本相同,且具有容错的实时任务调度算法可以有效地降低任务错失率;容错实时任务调度DSPN模型可以为复杂任务调度系统的Petri网建模与分析奠定了基础,并为实际工程应用提供了理论指导。

Abstract:

Formalized description of the complicated system has the extremely vital role to design the new system, improve and evaluate the existed system. A detailed DSPN(Deterministic and Stochastic Petri Net) model and performance analysis of fault-tolerant real-time hybrid task scheduling in processor system is presented in this paper. Firstly, the tasks are divided into four kinds based on their priority, period, fault tolerance and real-time. Secondly, the behavior of scheduling execution of tasks, preempting resource of the higher priority tasks, interrupting and resuming of tasks, occurring and recovering of failure in processor system is accurately described by DSPN, and then the model of fault-tolerant real-time task scheduling of processor is constructed. Finally, the simulation results demonstrate that the utilization of processor is same at the same load, and the fault-tolerant real-time task scheduling algorithm can effectively reduce the task miss ratio. The DSPN model constructed can analyze the quantitative performance metrics of the fault-tolerant real-time task scheduling, which not only will be useful for constructing the Petri net model for complex processor system, but also be helpful for engineers and researchers.