Abstract: How to schedule masses of tasks efficiently is an important issue to be resolved in cloud computing environment. An algorithm combining PSO and GA is brought up for user satisfaction and the needs of cloud providers. The algorithm, according to the characteristics of cloud environments to classify virtual machine resource, introduces the concept of the task-resources satisfaction distance and the comprehensive performance of resource. Then it optimizes the operation of initial particles to improve the quality of the particle in PSO. The algorithm fuses the operations of GA’s crossover and mutation and expands the search space of the particle to overcome the particles trapped in local optimal solution. The simulation result shows that the proposed algorithm is efficient to improve user satisfaction and reduce the completion time of the task.

Key words: cloud computing, task scheduling, Genetic Algorithm（GA）, Particle Swarm Optimization（PSO）

摘要： 对云计算任务调度进行了研究，针对用户满意度和云提供商利益需求，提出一种融合粒子群和遗传算法的PSOGA改进算法。根据云环境特点对虚拟机资源进行分类，同时引入任务-资源满意度距离、资源综合性能概念；对粒子群初始粒子操作进行优化，来提高粒子质量；为克服粒子易陷入局部最优解问题，加入遗传算法（GA）的交叉、变异操作，扩展粒子的搜索空间。仿真结果表明，该调度策略提高了用户满意度的同时减少了任务的完成时间，是云平台下一种有效的任务调度策略。

关键词: 云计算, 任务调度, 遗传算法, 粒子群算法