改进的异构多处理器的实时任务调度算法研究

Improved real-time task scheduling algorithms for heterogeneous multiprocessors

现有的很多调度算法存在时间复杂度过高或调度成功率低的问题。提出一种新的调度算法(HRT-SA),提高实时任务的调度成功率。HRTSA首先通过METC策略初始化分簇,降低算法的时间复杂度;再在放置任务时根据处理器的负载均衡进行处理器负载的有效控制;最后通过任务复制调度以提高任务调度成功率。对比实验分析表明提出的HRTSA算法时间复杂度与RTSDA相比较低,调度成功率较高。

Present scheduling algorithms have many problems,for example,high time complexity and low guarantee ratio.This paper proposed a new scheduling algorithm （HRTSA）,to improve the guarantee rate of real-time task.Firstly,HRTSA applied METC tactics to initialize cluster computing in order to lower the time complexity.Secondly in the step of put task on the processors use load balancing to control the tasks in processors.Lastly used the task duplication in order to enhance the success rate of task.By comparing experimental results can know that our scheduling algorithm achieved not only the lower time complexity but also the better guarantee ratio compare to RTSDA（real-time scalable duplication based algorithm）.