基于蚁群算法的多QoS约束海量数据网格任务调度  

Modified ant colony algorithm based grid task scheduling on multi-QoS constraint mass data

作　　者：胡毅[1] 龚斌[1] 刘运臣[1] 

机构地区：[1]山东大学计算机科学与技术学院,山东济南250100

出　　处：《华中科技大学学报（自然科学版）》2007年第S2期90-93,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家高技术研究发展计划资助项目(20060101A1003);山东省自然科学基金资助项目(Y2004G08)

摘　　要：针对海量数据计算网格任务调度问题,提出一种多Qos约束网格任务调度蚁群改进算法,优先考虑与数据存储网格结点间数据传输效率高、完成任务概率高的网格计算资源,通过调整蚁群算法中信息素挥发因子和传输导向因子的更新方法来提高算法的收敛速度和全局最优解搜索能力,从而达到调度目标最优.仿真实验结果表明,该算法与QoS-Min-min和QoS-Sufferage等同类算法相比,具有较好的综合性能,特别是在计算数据文件较大时,该算法在收敛速度和全局最优解搜索能力方面有明显的提高.Aimed at problems existing in mass data grid scheduling,this paper proposes an algorithm based on modified ant colony algorithm for grid task scheduling of multi-QoS constraints.Through improving the efficiency of data transferring between nodes of computing resources and data storage,it accomplishes the scheduling task with high efficiency and short time.The algorithm is based on transferring orientation factor to modify the searching behavior of ants,and modifies the value of pheromone according to the object function,which accelerates the convergence rate and improves the ability of searching an optimum solution.Simulation shows that it performs better than the QoS-Min-min and QoS-Sufferage in the integrated performances,especially for large computing data file.

关 键 词：网海量数据网格 蚁群算法 任务调度 服务质量 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]