基于粒子群的网格任务调度算法研究

Study on PSO algorithm in solving grid task scheduling

为了更好地解决异构动态环境下的资源管理问题,提出了一种网格环境下的任务调度模型。该模型考虑了当前网格虚拟组织下的计算资源、存储资源和带宽资源,模型的最优化目标是实现三者利用率最高和代价最低,即构造min-max函数。与遗传算法相比,利用粒子群优化算法对min-max函数求解提高了资源的利用率和任务的执行效率,同时在随着迭代次数增加的情况下,搜索速度、寻优率和避免早熟方面也有明显的提高。

In order to resolve the resources management in dynamic heterogeneous environment, a kind of task scheduling model for grid environment was proposed. The model considers the computing resources, storage resources and bandwidth resources of current virtual organization in grid, and the optimal target of the model is to achieve the max-ratio and the min-cost of the above three kinds of resources, viz to build the min-max function. To compare with GA（genetic algorithm）, PSO（particle swarm optimization） was applied in solving the rain-max function so that the ratio of using resources and the efficiency of scheduling task are enhanced. Meanwhile, with the rise of iterative times, the searching speed, optimization ratio and avoiding pre-maturity are also distinctly enhanced.