基于Q学习和双向ACO算法的云计算任务资源分配模型设计

发布时间：2018-01-28 20:55

本文关键词： 云计算蚁群优化算法 Q学习资源分配　出处：《计算机测量与控制》2014年10期 　论文类型：期刊论文

【摘要】：云计算异构环境中由于计算和存储资源物理分布的不一致性,往往容易导致在应用传统的调度算法进行任务资源分配时存在调度效率低和负载不均衡的问题,为此,设计了一种基于Q学习和双向ACO算法的云计算任务资源分配模型;首先,引入了基于主从结构的调度模型,并综合考虑任务计算完成时间、网络带宽和延迟等因素设计了资源分配目标函数,然后,设计了基于Q学习的云计算资源初始分配方法,将其获得的最优策略对应的Q值初始化网络中节点的Q值,最后,设计一种结合前向蚂蚁和后向蚂蚁的双向ACO算法实现任务资源的最终分配,并对算法进行了定义和描述;在CloudSim环境下进行仿真实验,结果证明文中方法能有效实现云计算异构环境下的任务资源分配,且与其它方法相比,负载均衡离差值平均约为0.071 5,是一种适用于云计算异构环境的有效资源分配方法。
[Abstract]:Due to the inconsistency of the physical distribution of computing and storage resources in the heterogeneous environment of cloud computing, it is easy to lead to the problems of low scheduling efficiency and unbalanced load when the traditional scheduling algorithm is used to allocate task resources. Therefore, a resource allocation model of cloud computing task based on Q learning and bidirectional ACO algorithm is designed. Firstly, the scheduling model based on master-slave structure is introduced, and the objective function of resource allocation is designed considering the completion time of task calculation, network bandwidth and delay. This paper designs an initial allocation method of cloud computing resources based on Q learning, and initializes the Q value of nodes in the network by the Q value corresponding to the optimal policy. Finally. A bidirectional ACO algorithm combining forward ant and backward ant is designed to realize the final assignment of task resources, and the algorithm is defined and described. The simulation results in CloudSim environment show that the proposed method can effectively implement task resource allocation in heterogeneous cloud computing environment, and compared with other methods. The average value of load balancing distance is about 0.071 5, which is an effective resource allocation method for cloud computing heterogeneous environment.
【作者单位】：盐城工学院信息工程学院;
【分类号】：TP18;TP3
【正文快照】： 0引言云计算[1-2](cloud computing)是继并行计算、分布式计算和网格计算的一种新的计算模式[3-4]。云计算平台的体系结构[5]主要是由三个部分组成即基础设施即服务、软件即服务和应用级服务,但其需要解决的问题是对于新到达的服务请求,需要将虚拟资源池中的资源分配给该服务

【参考文献】