摘要
随着云计算的不断发展,传统的单目标优化下的任务调度已经不能满足用户的服务质量要求。论文选取运行时间、费用和负载均衡建立多目标优化的云任务调度模型,提出一种改进的多目标小生境Pareto遗传算法(NPGA),采用相似任务序列交叉操作加快进化,再采用位移变异避免算法过早收敛。此外,通过自适应选取比较集合规模和小生境半径提高算法的收敛速度。仿真结果表明,改进后的NPGA算法在云调度中保持Pareto最优解的多样性和分布性更优。
As cloud computing continues to evolve, task scheduling under the traditional single-objective optimization has been unable to meet the user＇s requirements for quality of service. This paper selects the running time and cost and load balance of establishing a multi-objective optimization of cloud task scheduling model, an improved multi-objective niche Pare- to genetic algorithm（NPGA） is proposed to speed up evolution and avoid premature convergence through a similar task se- quence erossover（STOX） operating and shift mutation. In addition, the size of comparison set and niche radius are selected a- daptively to improve convergence speed. Simulation results show that improved NPGA algorithm is better to maintain diver- sity and distribution of Pareto optimal solutions in the cloud scheduling.
出处
《计算机与数字工程》
2015年第7期1196-1201,1216,共7页
Computer & Digital Engineering
