多处理器混合关键性系统中的划分调度策略

发布时间：2018-03-01 01:25

本文关键词： 混合关键性系统多处理器划分调度 EDF(earliest deadline first)　出处：《软件学报》2014年02期 　论文类型：期刊论文

【摘要】：多核处理器正越发广泛地应用到现代嵌入式系统的设计与实现当中,其强大的计算能力为将多个不同关键性级别的功能子系统集成到统一的共享资源平台提供了支持.混合关键性系统的调度问题即便在单处理器平台中都极具挑战性,在多处理器平台则更为困难.将目前资源利用率最高的单处理器混合关键性调度算法EY-VD扩展到多处理器平台中.首先,结合传统的划分调度策略提出了适用于多处理器混合关键性系统的MC-PEDF(mixedcriticality partitioned earliest deadline first)划分调度算法.尽管比之前的算法有更好的可调度性能,但传统的划分策略不能有效地平衡不同关键性级别下的负载,故其不完全适用于混合关键性系统.为了克服传统策略的不足,提出了划分调度策略OCOP(one criticality one partition).OCOP允许系统在关键性模式切换时对实时任务集进行重新划分,进而更好地平衡各个处理器在不同关键性模式中的资源利用率.基于OCOP,提出了第2种划分调度算法MC-MP-EDF(mixed-criticality multi-partitioned EDF).基于随机生成任务集的仿真实验结果表明,与MC-PEDF和已有的算法相比,MC-MP-EDF能够显著地提高系统的可调度性,尤其是在处理器数量较多的系统中.
[Abstract]:Multi-core processors are increasingly widely used in the design and implementation of modern embedded systems. Its powerful computing power provides support for integrating multiple functional subsystems at different critical levels into a unified shared resource platform. Scheduling problems for hybrid critical systems are extremely challenging even on a single processor platform. It is even more difficult on multiprocessor platforms. We extend the single processor hybrid critical scheduling algorithm (EY-VD), which has the highest resource utilization, to multiprocessor platforms. Combined with the traditional partitioning scheduling strategy, a MC-PEDF(mixedcriticality partitioned earliest deadline firstscheduling algorithm for multiprocessor hybrid critical systems is proposed, although it has better schedulability than previous algorithms. But the traditional partitioning strategy can not effectively balance the load at different critical levels, so it is not fully applicable to hybrid critical systems. A partitioning scheduling policy, OCOP(one criticality one partition).OCOP, is proposed to allow the system to repartition the real-time task set when the critical mode is switched. Furthermore, the resource utilization of each processor in different critical modes is better balanced. Based on OCAP, a second partition scheduling algorithm, MC-MP-EDF(mixed-criticality multi-partitioned EDF, is proposed. The simulation results based on randomly generated task sets show that, Compared with MC-PEDF and existing algorithms, MC-MP-EDF can significantly improve the schedulability of the system, especially in systems with a large number of processors.
【作者单位】：东北大学信息科学与工程学院;
【基金】：国家科技支撑计划(2012BAF13B08) 中央高校基本科研业务费项目(FRFCUN100204001,FRFCUN110804003) 国家自然科学基金(61300022)
【分类号】：TP332

【共引文献】