论文部分内容阅读
多QoS约束下的工作流调度是网格计算中难以求解的问题.在深入剖析该问题难解性基础上,采用克隆选择算法求解该问题.首先通过增加网格服务的唯一标识,简化工作流调度的编码方式.其次,提出QoS偏好的概念,将调度问题的目标函数转换为适应值函数.该算法具有QoS属性的可扩展性.最后通过大量实验,优化算法参数,与基于遗传算法、蚁群算法的调度算法对比,克隆选择算法求解效率较优.在扩展情况下,与单一QoS约束下的时间、费用贪婪算法对比,克隆选择算法能进行最优调度.
Multi-QoS constrained workflow scheduling is a difficult problem to solve in grid computing.On the basis of deeply analyzing the difficulty of the problem, a clone selection algorithm is used to solve the problem.Firstly, workflow is simplified by increasing the unique identification of grid services Scheduling algorithm.Secondly, the concept of QoS preference is put forward and the objective function of scheduling problem is transformed into fitness function.The algorithm has scalability of QoS attributes.Finally, through a large number of experiments and optimization of algorithm parameters, Compared with the clustering algorithm, the clonal selection algorithm is more efficient in solving the problem.In the extended case, the clonal selection algorithm can be optimally dispatched in comparison with the time and cost greedy algorithm under the single QoS constraint.