针对不确定执行时间任务的实时调度方案

为了解决在实时调度系统中,任务执行时间不确定性所带来的问题,提出了基于时间预测的调度方案。该方案设计了VSM(vectorspacemodel)模型、Markov模型和MVSM(Markovvectorspacemodel)模型。对这3种模型的比较表明:基于MVSM模型的调度方案可以很好地保证实时系统的效率和稳定性,即使在处理器超载的情况下,也能自动调节,超过99%的作业可以在时间期限之前完成。采用时间预测的方法,可以较好地解决任务执行时间不确定性所带来的影响,为不确定环境下的实时调度系统提供一种很好的参考解决方案。

Real-time scheduling for tasks with uncertain execution times

A time prediction method was developed to predict the execution time of real-time tasks during execution as the basis for a real-time scheduling system for tasks with uncertain execution times. Comparison of three time prediction models, a VSM-based model, a Markov-based model, and an MVSM-based model, showed that the MVSM-based model most effectively guarantees CPU utilization and the stability of the real-time system. Even with overloads, the scheduling system automatically adjusts, with more than 99% of the tasks completed before their deadlines. The time prediction method resolves the problem of uncertain execution times and provides a reference solution for real-time scheduling systems in uncertain environments.