张红.基于改进差分进化算法的云计算资源调度策略[J].中南民族大学学报自然科学版,2020,39(5):532-538
基于改进差分进化算法的云计算资源调度策略
Cloud computing resource scheduling strategy based on improved differential evolution algorithms
  
DOI:10.12130/znmdzk.20200515
中文关键词: 云计算  差分进化算法  资源调度
英文关键词: cloud computing  differential evolution algorithm  resource scheduling
基金项目:甘肃省工业和信息化发展省级专项基金资助项目( 23051358) ; 甘肃省引导科技创新发展专项基金资金项目( 2018ZX—05)
作者单位
张红 甘肃中医药大学 网络与信息管理中心兰州 730000 
摘要点击次数: 68
全文下载次数: 51
中文摘要:
      在云计算环境之中,计算资源会动态地发生变化,差分算法通过选择、交叉和变异操作对云计算任务资源调度问题实现寻优,具备较高的前期寻优效果,但它的全局搜索能力较差,后期对最优解的搜索速度变慢。提出了一种基于变异概率自适应调整的改进差分进化算法(Adaptive Tuning of Mutation Probability Improved Differential Evolution, ATMPIDE)的云计算资源调度策略,其中,交叉操作选择种群的个体来执行多点交叉操作,改变染色体的基因排列,生成更多不一样的个体,确保群体的多样化。变异操作能够根据适应度的值自动设定阈值,在个体基因上随机选择多个位置,在每个位置进行小范围变异,根据变异的阈值,若产生的随机数小于阈值,则发生变异,否则不发生变异。仿真实验证明:所提出的改进算法能够加强全局搜索能力及加快搜索速度,最终找出最优解,不管虚拟机数量与任务数量之间如何变化,该算法在负载均衡方面和任务完成时间等方面都能取得较好的效果。
英文摘要:
      In cloud computing environment, computing resources will change dynamically. Differential algorithm can optimize the task scheduling problem of cloud computing through selection, crossover and mutation operation, which has a higher pre-optimization effect, but its global search ability is poor, and the search speed of the optimal solution becomes slower in the later stage. An improved Differential Evolution based on Adaptive Tuning of Mutation Probability (ATMPIDE) in cloud computing resource scheduling strategy is proposed, in which crossover operations select individuals of the population to perform multi-point crossover operations, change the gene arrangement of chromosomes, generate more different individuals, and ensure the diversity of the population. Variation operation can automatically set threshold according to fitness value, randomly select multiple locations on individual genes, and mutate in a small range at each location. According to the threshold of variation, if the random number generated is less than the threshold, the mutation will occur, otherwise no mutation will occur. The simulation results show that the improved algorithm proposed can enhance the global search ability and speed up the search speed, and finally find the optimal solution. No matter how the number of virtual machines changes with the number of tasks, the algorithm can achieve better results in load balancing and task completion time.
查看全文   查看/发表评论  下载PDF阅读器
关闭