云计算虚拟资源差分进化分配方法仿真  

Simulation of Differential Evolutionary Allocation of Virtual Resources in Cloud Computing

作　　者：高杨[1] 马贺男[1] 刘晓菲 GAO Yang;MA He-nan;LIU Xiao-fei(Public Security Information Technology and Information College,China Investigation Police University of China,Shenyang Liaoning 110034,China;Shenyang Aerospace University,Shenyang Liaoning 110136,China)

机构地区：[1]中国刑事警察学院公安信息技术与情报学院,辽宁沈阳110034 [2]沈阳航空航天大学,辽宁沈阳110136

出　　处：《计算机仿真》2023年第9期328-332,共5页Computer Simulation

摘　　要：云平台集群规模较大,集群节点之间存在一定的差异性,为云计算虚拟资源分配带来了一定困难。为降低数据中心能耗与时间复杂度,提出云计算虚拟资源差分进化分配方法。为实现全局能耗更新调节服务器工作电压和资源分配频率,使用动态能量感知策略计算资源分配的最短时间跨度与空闲分配时间,代入能量消耗时间变换因子,将分配所需能量转换成时间参数,使用二元组构建云计算资源分配模型;将资源节点拟作种群,通过资源-任务间接方式进行染色体编码,以轮盘赌原则为基础,采用差分进化算法更改种群染色体基因排列方式,推算个体适应度,通过变异概率自主调节种群个体,获取被标记的最优个体,实现云计算虚拟资源最优分配。实验结果表明,所提方法可以使用较少的服务器完成资源分配,云计算任务分配效率高且执行能耗低。The cluster scale of the cloud platform is large.And the difference between cluster nodes leads to the difficulties for allocation of cloud computing virtual resources.In order to reduce the energy consumption and time complexity of data centers,this article proposed a differential evolution allocation method for cloud computing virtual resources.In order to realize the update of global energy consumption,we adjusted the server voltage and resource allocation frequency and then adopted the dynamic energy-aware strategy to calculate the shortest time span and idle allocation time of resource allocation.After that,we introduced the energy-consumption time conversion factor to convert the energy required for allocation into the time parameter.Moreover,we used a tuple to construct a model for cloud computing resource allocation.The resource nodes are proposed as a population and coded the chromosome in indirect ways.Based on the principle of roulette,we adopted the differential evolution algorithm to change the arrangement of chromosome genes in the population and calculate the individual fitness,thus achieving the independent adjustment of population individuals through the mutation probability.Finally,the labled optimal individual was obtained.Thus,the optimal allocation for cloud computing virtual resources was achieved.Experimental results show that the proposed method can complete resource allocation with fewer servers.Meanwhile,the allocation efficiency of cloud computing tasks is high,but the energy consumption is low.

关 键 词：差分进化算法 云计算 虚拟资源 资源分配 能量感知 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]