离散微粒群优化算法在网格任务调度中的应用  

Application of Discrete Particle Swarm Optimization Algorithm to Grid Task Scheduling

作　　者：卜艳萍[1] 俞金寿[1] 

机构地区：[1]华东理工大学自动化研究所

出　　处：《计算机仿真》2008年第4期175-178,共4页Computer Simulation

摘　　要：网格任务调度算法是影响网格成功与否的关键技术之一。在研究现有任务调度策略的基础上,指出Min-Min算法的负载不均衡性。借鉴遗传算法中的交叉操作过程,提出了一种新的任务调度算法。该算法对传统的连续型微粒群优化算法进行改进,使其适用于网格任务调度问题的优化处理,实现网格资源的优化分配。仿真研究表明该算法更符合网格调度的复杂环境,能得到较短的任务执行时间和较好的负载均衡性。对比分析表明,离散微粒群优化算法所得结果优于常用的Min-Min调度方案,是一种高效的调度方法。Algorithm for grid task scheduling is one of the key technologies which influence grid success. Based on the research of existing scheduling strategy, a bad quality of load balancing of the most classical Min - Min was pointed out. By referring to the crossover operations in the genetic algorithms, this paper presents a new task scheduling algorithm. The algorithm adapts the classic particle swarm optimization （PSO） algorithm to the optimization of grid task scheduling problems, and optimizes the grid resource allocation. By utilizing this algorithm, shorter execution time and better performance of load balancing can be gained via the simulation studies, especially in grid scheduling environments. The comparison results show that the solutions produced by discrete particle swarm optimization algorithm are better than that of Min - Min scheduling policies. The new scheduling algorithm is validated by the simulation results.

关 键 词：离散微粒群优化算法 网格 任务调度 完成时间 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]