网格计算中资源负载均衡问题的优化调度模型及其算法.docVIP

· PAGE 52· 通 信 学 报 第29卷 第3A期 舒万能：网格计算中资源负载均衡问题的优化调度模型及其算法 · PAGE 53· 第29卷第3A期 通 信 学 报 Vol.29 No.3A 2008年3月 Journal on Communications March 2008 网格计算中资源负载均衡问题的优化调度模型及其算法 舒万能 (中南民族大学 计算机科学学院, 湖北 武汉 430074) 摘 要：针对网格计算中多个独立任务在多个异构的资源上处理时，资源的负载均衡为最小非抢先调度的问题，建立了一类资源负载均衡问题的优化调度模型。该模型引入了资源的动态负载权值表示方案，并采用量子遗传模拟退火算法对模型进行求解。该算法采用量子编码来表征染色体，先通过选择，交叉，变异等遗传操作来产生一组新的个体，然后再独立地对所产生的各个个体进行模拟退火，直到退火温度不能再降低为止，从而求得问题的最优解。理论分析和实验结果表明这种算法优于其他调度算法。 关键词：网格计算；负载均衡；量子遗传模拟退火算法；遗传算法；模拟退火算法 中图分类号：TP393 文献标识码：A 文章编号：1000-436X(2008)3A-0048-06 Optimal scheduling model and its algorithm for resource load balancing problems in grid computing SHU Wan-neng (College of Computer Science, South-Central University for Nationalities, Wuhan 430074, China) Abstract: A novel optimal scheduling model and its algorithm were developed for a kind of resource load balancing problems, in which many independent tasks should be scheduled many isomerous available resources to make the whole task fulfilled in the shorten time. This model introduced dynamic load weight of resource nodes and quantum genetic simulated annealing algorithm was applied to solve the scheduling model. By adopting the qubit chromosome as a representation, it first generated a new group of individuals through genetic operation such as reproduction, crossover, mutation, etc, and than simulated annealing independently all the generated individuals respectively. When the temperature in the process of cooling no longer falls, the result was the optimal solution on the whole. From the analysis and experiment result, it concluded that this algorithm was superior to other algorithm. Key words: grid computing; load balancing; quantum genetic simulated annealing algorithm; genetic algorithm; simulated annealing algorithm 1 引言 收稿日期：2008-01-04 基金项目：国家自然科学基金资助项目 Foundation Item: The National Natural Science Foundation of China 在网格计算中，任务调度的实质就是将n个相互独立的元任务分配到m个异构的资源上，使得总的任务完成时间最小，资源得到充分的利用