摘要
任务调度是计算网格系统中极其关键的一部分,一种好的调度方法可以极大地提高整个系统的性能。针对蚂蚁算法在网格调度中早期信息素匮乏和蚂蚁分工单一的缺陷,提出了一种新的启发性智能调度方法。在调度过程前期,采用遗传算法为各网格节点生成丰富的信息素,作为调度中心进行任务调度的依据,然后在多群蚂蚁算法中,各种群的蚂蚁根据分工的不同在属于自己的空间中寻找最优解,从而缩小了搜索规模,加快了收敛速度,优化了调度性能。
Job ,scheduling is a key part in the computing grid system, a good scheduling method could enhance the performance of entire system. In this paper, a new heuristic intelligent scheduling method is proposed, which could solve the drawback of ant algorithm's pheromone lack when applied in job seheduling in the early stage and all ants carry our the .same task. At the beginning of job scheduling, create abundant pheromones for each network- node by using genetic algorithm. And the scheduling center performs job scheduling depending on these pheromones. Because of dividing the work, each colony's ant has different task, and they search for the most optimun sohtuion from the multi colony ant algorithm. As a rest,It, it reduces the range during do the searching job and accelerates the convergent speed. In other words, it optimizes the capability of job scheduling.
出处
《计算机技术与发展》
2006年第11期119-121,124,共4页
Computer Technology and Development
关键词
任务调度
信息素
并行遗传算法
多群蚂蚁算法
负载平衡
job seheduling
pheromone
parallel genetic algorithm
multi colony ant algorithm
lead balancing