随着云计算的快速发展,如何高效地进行云任务调度逐渐成为云计算研究的重点.任务调度问题属于NP优化问题,许多超启发式算法被应用到任务调度问题.针对蚁群算法在任务调度中存在收敛速度慢、局部搜索能力差和易于陷入局部最优的问题,将蚁群算法和模拟退火算法相结合,提出了蚁群模拟退火算法,拟解决云计算中的任务调度问题.在该算法中,以减少任务的完成时间和保证资源负载均衡为目标,根据蚁群算法构造局部最优解,利用模拟退火算法较强的局部搜索能力,将局部最优解作为模拟退火算法的初始解进行局部搜索并以一定的概率接受当前搜索结果,从而避免算法陷入局部最优.仿真结果表明,蚁群模拟退火算法的性能优于先来先服务(First Come First Served,FCFS)和标准蚁群优化(Ant Colony Optimization,ACO)算法.%With the rapid development of cloud computing,how to carry on task scheduling effectively is crucial in the research of cloud computing. Cloud task scheduling belongs to a NP-hard optimization problem,and many meta-heuristic algorithms have been proposed to solve it. ACO algorithm in task scheduling still has many shortcomings such as slow convergence speed,poor ability of local search and falling into local optimum easily. Therefore,a new algorithm-ACOSA is presented to solve task scheduling problem. In this algorithm,re-ducing task completion time and ensuring resource' s load balance as the goal,according to the local ant colony algorithm the optimal so-lution is constructed,and the strong local search capability of simulated annealing algorithm is applied to make the local optimal solutions as the initial solutions of simulated annealing algorithm and accept the results of current search to a certain probability in order to avoid falling into the local optimal. Simulation results show that ACOSA is superior to First Come First Served ( FCFS) and Ant Colony Opti-mization ( ACO) by reducing make span and achieving load balance.
展开▼
