摘要
合理地进行任务调度是云计算长期以来存在的挑战。云任务的调度过程具有动态性的特点,仅从单一方面来优化调度策略已不能满足用户需求。针对上述问题,从任务完成时间、任务完成成本、资源利用率三个方面出发,提出一种基于遗传与粒子群算法融合的多目标任务调度算法。在遗传算法的变异操作中引入粒子群算法,既可以发挥遗传算法全局搜索能力强的优势,又可以利用粒子群算法的反馈特性改善变异操作提高收敛速度。通过Cloud Sim平台进行云环境仿真实验,将此算法与遗传算法(GA)和粒子群算法(PSO)进行比较。实验结果表明,在相同的条件设置下,该算法在用户满意度和资源利用率方面都优于遗传算法和粒子群算法,是一种云计算环境下有效的任务调度算法。
Howto schedule tasks reasonably remains a long-standing challenge in cloud computing.The process of the cloud task scheduling has the characteristics of dynamic,so to optimize the scheduling strategy only from a single aspect cannot meet the needs of users.To solve the above problem,from three aspects of task completion time,task completion cost and resource utilization,a multi-objective task scheduling algorithm based on genetic algorithm and particle swarm optimization algorithm is proposed.Particle swarm optimization algorithm is introduced into mutation operation of genetic algorithm which can not only give play to advantage of quick global searching speed for genetic algorithm,but also apply particle swarm optimization algorithm 's feedback characteristic to improve mutation operation and convergence rate.Cloud Sim is adopted to simulate the cloud environment,and the GA and PSO is compared.The simulation results showthat under the same conditions,the combined algorithm outperforms other two algorithms on task completion time,task completion cost and resource utilization.It is an efficient task scheduling algorithm in the cloud computing environment.
出处
《计算机技术与发展》
2017年第2期56-59,共4页
Computer Technology and Development
基金
湖北省教育科研计划指导性项目(B2015373)
关键词
云计算
任务调度
多目标
遗传算法
粒子群算法
cloud computing
task scheduling
multi-objective
genetic algorithm
particle swarm optimization algorithm