基于离散时间距的在线可重构任务调度算法

On-line Scheduling Algorithm for Reconfigurable Tasks Based on Discrete Time-interval

任务调度是影响动态可重构系统性能的重要因素.针对现有预约算法中由于维护预约资源逻辑单元信息而带来系统额外开销大、任务调度自私性等问题,提出一种基于离散时间距的非预约调度算法.算法的特点在于通过任务紧迫度和时间距信息能够动态更新任务优先级和设置任务的启动时间,从而有效避免了复杂的系统开销和任务调度的自私性.实验表明,该算法能提高任务的调度成功率,而运行时间开销没有明显增加.

Task scheduling has an important effect on performance of dynamic reconfigurable system. Considering some defects in existing planning algorithm, such as heavy systemic overhead and selfish scheduling among tasks, this paper proposes a time-interval scheduling algorithm. The novelty of our algorithm resides in setting the dynamic priorities and starting time for tasks by using the laxity and the time interval respectively. The approach could avoid the heavy systemic overhead in planning algorithm and produce a fair scheduling order. The simulation results show the proposed algorithm gains a prominent improvement in successful scheduling ratio of task, while the run-time efficiency is preserved.