面向异构信号处理平台的负载均衡算法

A Load Balancing Algorithm for HeterogeneousSignal Processing Platform

针对当前异构信号处理平台中信号处理应用的调度算法优化目标单一且调度结果中处理器负载不均衡的问题,提出了一种基于蚁群优化算法的负载均衡算法。该算法结合蚁群优化算法的快速搜索能力和组合优化能力,以信号处理应用的调度长度和处理器负载均衡为优化目标,对初始信息素矩阵和蚂蚁的遍历顺序进行改进,提出调度长度启发因子和负载均衡启发因子对处理器选择公式进行改进,利用轮盘赌策略确定信号处理应用各子任务分配的处理器,完成信号处理应用的调度。仿真结果表明,该算法得到调度结果在调度长度和负载均衡方面均有改进,可以充分发挥各处理器性能,提高异构信号处理平台的整体效率。

For the problem that the scheduling algorithm of signal processing application in the current heterogeneous signal processing platform has a single optimization goal and the processor load is unbalanced in the scheduling results,a load balancing algorithm based on Ant Colony Optimization(ACO)algorithm is proposed.The algorithm combines the fast search ability and combinatorial optimization ability of ACO algorithm,takes the scheduling length of signal processing application and processor load balance as the optimization goal,improves the initial pheromone matrix and ant traversal order,proposes scheduling length heuristic factor and load balance heuristic factor to improve the processor selection formula,and uses Roulette strategy to determine the processor assigned to each subtask of signal processing application,thus completing the scheduling of signal processing applications.The simulation results show that the scheduling results obtained by the algorithm are improved in terms of scheduling length and load balancing.It can give full play to the performance of each processor and improve the overall efficiency of heterogeneous signal processing platform.