摘要： 提出了基于并行遗传算法的网格资源分配方法，并采用粗粒度模型设计了该并行算法。为了评估该并行算法的性能，在PC 集群上实现了该并行算法和一个串行遗传算法。通过比较两个算法的执行时间和解的质量，说明了并行算法极大地提高了求解的速度和质量，是一种高效的资源分配方法。

关键词: 网格；资源分配；并行遗传算法

Abstract: Based on the attribute of natural parallelism of the genetic algorithm, a parallel-genetic-algorithm-based approach to resource allocation in the grid environments is proposed. The coarse-grained model is adopted to develop the parallel algorithm. In order to evaluate the performance of the parallel genetic algorithm, both the algorithm and a serial genetic algorithm are programmed in the cluster of PCs. Through comparisons of the execution time and the quality of the solutions of the two algorithms, it is concluded that the parallel-genetic-algorithm-based approach improves the speed of resource allocation and the quality of the solutions and is an effective approach to grid resource allocation.

Key words: Grid; Resource allocation; Parallel genetic algorithm