计算机应用 ›› 2012, Vol. 32 ›› Issue (07): 1916-1919.DOI: 10.3724/SP.J.1087.2012.01916

• 先进计算 • 上一篇    下一篇

云资源中多目标集成蚁群优化调度算法

左利云1,左利锋2   

  1. 1. 广东省石化装备故障诊断重点实验室(广东石油化工学院),广东 茂名525000
    2. 郑州宇通客车股份有限公司 新能源产品部,郑州450016
  • 收稿日期:2011-12-09 修回日期:2012-02-27 发布日期:2012-07-05 出版日期:2012-07-01
  • 通讯作者: 左利云
  • 作者简介:左利云(1980-),女,河南周口人,副教授,主要研究方向:云计算、任务调度;左利锋(1985-),男,河南周口人,工程师,主要研究方向:Matlab、Simulink、CloudSim建模仿真应用。
  • 基金资助:

    广东省科技计划项目(2007B010400042);广东省自然科学基金资助项目(06029274);茂名市科技计划项目(2011008)

Multi-objective integrated ant colony optimization scheduling algorithm based on cloud resource

ZUO Li-yun1,ZUO Li-feng2   

  1. 1. Guangdong Province Key Laboratory of Petrochemical Equipment Fault Diagnosis (Guangdong University of Petrochemical Technology), Maoming Guangdong 525000, China
    2. New-energy Bus Department, Zhengzhou Yutong Bus Limited Company, Zhengzhou Henan 450016, China
  • Received:2011-12-09 Revised:2012-02-27 Online:2012-07-05 Published:2012-07-01
  • Contact: ZUO Li-yun

摘要: 针对云计算环境的复杂性和云资源的不确定性,提出多目标集成蚁群优化调度算法。采用熵度量云资源的不确定性,进行信息素全局更新,以提高算法收敛速度;将Min-min算法得出的任务预期最小完成时间作为启发信息,以实现最小调度时间;在信息素局部更新时加入负载系数,根据当前负载情况调节信息素,满足负载均衡需求,同时在更新时考虑信息素扩散因素,不仅计算当前节点还考虑周遭节点信息素情况,可增强蚂蚁间协作,提高最优解的性能。改进后算法比原始蚁群算法降低了算法复杂度,提高了最优解精度。云仿真系统实验测试表明改进算法在调度时间、负载均衡等方面表现均优于其他算法。

关键词: 云计算, 多目标集成, 最优解,

Abstract: With regard to the complexity of cloud computing and uncertainty of cloud resources, an integrated multi-objective ant colony optimization scheduling algorithm was proposed. Using entropy to measure the uncertainty of cloud resources, and updating the global pheromone, the algorithm's convergence rate could be improved. In order to achieve the minimum activation time, the expected minimum completion time calculated by Min-min algorithm was used as heuristic information. When the local pheromone was updated partly, the load factor was added according to the current load regulation of pheromone, to achieve the load balance. Collaboration between ants could be enhanced and the performance of the optimal solution would be improved, considering the pheromone diffusion factors, that is to consider not only the current node pheromone but also neighbor node pheromone. The advanced ant colony algorithm reduced the complexity of the algorithm better than the original algorithm, and improved the optimal solution accuracy. The experiments of the cloud simulation system prove that the performance of time scheduling and load balancing of the proposed algorithm is better than other algorithms.

Key words: cloud computing, multi-objective integration, optimal solution, entropy

中图分类号: