基于粒子群离散优化的网格资源分配方法

Grid Resource Allocation Method based on Discrete Particle Swarm Optimization

为提升网格计算的资源分配效率和调度精确性,提出了一种粒子群离散优化的网格资源分配方法.该方法首先基于网格的离散特性给出了粒子位置和速度的定义;接着,基于网格参量推导了粒子优化矩阵,分析了离散优化的实现步骤,并通过粒子速度的归一化处理,有效缓解POS后期粒子的局部早熟问题.最后,通过DridSim平台构建的10个资源节点分别针对利用效率、时间消耗进行了仿真分析,结果表明本文方法的时间利用效率提升了近5%.

In order to promote the efficiency of resource allocation and scheduling of grid computing accuracy, this paper proposes a grid resource allocation method based on discrete particle swarm optimization This method firstly gives the definition of discrete particle position and velocity; Second, the particles optimization matrix is deduced based on grid parameters. The paper analyzes the implementation steps of discrete optimization. The particle＇s local early--maturing problem in the late of POS is effectively relieved based on the particle velocity normalized processing. Finally, through the DridSim platform built 10 nodes, respectively. The utilization efficiency of resources and time consumption is carried on the simulation. And the results show that the method of time use efficiency increased by almost 5 %.