Abstract: The heuristic method is problem-dependent and fits only a particular of problems while the meta heuristic method has the problems of incomplete search space or low search efficiency in the complete space,and redundant space in traditional one-dimensional coding.To fill the gaps,a Two-stage coevolutionary Genetic Algorithm with Two Dimensional coding (TDTSGA) was proposed for cloud workflow scheduling.In this algorithm,a new two-dimensional individual coding method was used,the crossover and mutation methods based on two-dimensional hierarchical sorting and topological sorting were designed,and a two-stage coevolution strategy was employed.Extensive experiments were conducted on various workflow applications,and the result demonstrated the performance of TDTSGA was better than that of existing classical approachs.

Key words: workflow, cloud computing, scheduling optimization, genetic algorithms, coevolutionary

摘要： 针对当前启发式算法依赖于特定问题,元启发式方法存在搜索空间不完备或在完备空间上搜索效率不高,以及传统一维编码存在冗余空间等问题,提出一种基于二维编码两阶段协同进化遗传算法(TDTSGA)的云工作流调度优化方法。在TDTSGA中采用一种新的二维个体编码方法,设计了基于二维层次排序和拓扑排序的交叉变异方法,同时采用了两阶段协同进化策略。通过在各种工作流应用案例上进行广泛实验,验证了TDTSGA的优越性。

关键词: 工作流, 云计算, 调度优化, 遗传算法, 协同进化