信息物理融合系统的动态多优先级调度

信息物理融合系统(Cyber-physical Systems,CPS)的复杂和异构性给设计者带来了不少挑战,其中任务的多样性使得传统的调度策略不能满足CPS的性能需求.提出了专门针对基于大规模传感器网络的CPS的动态多优先级调度策略.根据任务类型分配4级缓存队列:第1级是来自控制器待处理的实时任务,拥有最高的可抢占式优先级;第2级是来自控制器待转发的实时任务,拥有次高的可抢占式优先级;第3级是来自其他节点待转发的非实时任务,拥有第三高的非抢占式优先级;第4级是来自本地待发送的非实时任务,拥有最低的非抢占式优先级.设计了抢占与非抢占混合的动态调度策略来减少任务的平均等待时间,加入了等待时间阈值机制来保证第4级任务的公平性.通过理论分析和仿真实验对调度策略的性能做了评价.仿真结果显示,动态多优先级调度策略在提高系统性能和稳定性上要优于传统优先级调度.

Dynamic Multi-priority Scheduling for Cyber-physical Systems

The complexity and heterogeneity of Cyber-physical systems (CPS) bring big challenges to system desig-ners.Conventional task scheduling scheme will not satisfy performance requirements of CPS due to the diversity of tasks.A dynamic multi-priority scheduling scheme for a class of CPS based on large-scale sensor network was proposed.In the proposed scheme,tasks reaching each node in the system are divided into four categories:real time tasks to be sent from local node which have the highest preemptive priority,real time tasks to be transmitted from other nodes which have the second highest preemptive priority,non-real time tasks to be processed from controller which have the third highest non-preemptive priority,non-real time tasks to be transmitted from other nodes which have the lowest non-preemptive priority.A mixed preemptive and non-preemptive priority scheduling method was designed to reduce the average waiting time of tasks in each queue.A waiting-time threshold mechanism was added to ensure the fairness of tasks.We analyzed task transmission delay of our scheduling algorithm.Simulation results show that the proposed dynamic multi-priority scheduling scheme outperforms conventional priority queue scheduling scheme.