基于资源回收的实时任务同步算法

Task Synchronization Scheduling with Slack Reclaiming in Real-Time Systems

在服务器算法中,传统的任务同步机制很少考虑空闲预算回收的问题,造成处理器资源的极大浪费,或在资源回收算法中,若任务相互独立,则很难满足实际情况的需要。针对此问题,通过分析带宽遗传算法,提出基于资源回收的实时任务同步算法。在资源回收算法的基础上引入任务同步机制,利用带宽遗传算法同步任务,同时将服务器中的空闲预算优先分配给有资源限制的任务。仿真证明,算法可以减少系统延迟,提高任务调度成功率,避免优先级翻转和死锁的发生。

The traditional task synchronization mechanism rarely considers the issue of slack reclaiming among the server-based real-time algorithm,or assumes that tasks are independent which does not meet the actual need.A new algorithm named SRBI is proposed to solve these problems.SRBI uses the bandwidth inheritance for task synchronization,and allocates the slack to the task with resource constraints.Simulation results show that this algorithm can reduce the OS latency,promote the task success rate as well as reduce the probability of priority inversion and deadlock.