采用优先级排序的数据流驱动调度算法

Priority-Based Data Stream-Driven Scheduling Algorithm

针对事件触发并发数据流(event-triggered concurrent data flow,ECDF)调度算法中多个组件间相互切换导致的计算开销增大问题,提出采用优先级排序的数据流驱动调度(priority based data flow,PBDF)算法。研究了信号处理平台中组件间数据依赖关系、预期处理开销和通信开销对系统组件优先级的影响。在适用于信号处理平台数据流驱动模型的基础上,采用表调度的思想,结合ECDF调度算法的要求,建立有序组件执行顺序。最后依据数据流驱动算法调度优先级列表。仿真结果表明,与现有算法相比,该算法提高了调度效率,有效降低了应用响应时间。

To address the problem of computational overhead increasing caused by switching between multiple components in event-triggered concurrent data flow (ECDF) scheduling algorithm, a priority-based data stream-driven scheduling (PBDF) algorithm is proposed. The effect of data dependency, expected processing cost and communication overhead on component priority of signal processing platform is studied. On the basis of the data stream driven model applied to signal processing platform, the idea of table scheduling and the order of ECDF scheduling algorithm are established. Finally, the priority list is scheduled according to the data flow-driven algorithm. Simulation results show that compared with the existing algorithms, the algorithm improves the scheduling efficiency and reduces the application response time.