Abstract: First，the characteristic of grid resource scheduling and limitation of current genetic algorithm are analyzed.And genetic algorithm is improved.A grid resource scheduling algorithm based on improved GA is proposed.In the algorithm，a lot of QoS parameters are considered，such the quota of tasks and resource，the latency time of tasks and the utility of resource etc.And the author proved correctness of the algorithm by Markov theory.Finally，the comparison of two algorithm performance is presentesd by simulation experiment.The results indicate that the improved algorithm can improve the performance about execution time consumption，and achieve high resource utilization.

Key words: grid, resource scheduling, Genetic Algorithm（GA）, Markov theory

摘要： 首先对网格资源调度的特点、现有遗传算法的局限性进行了分析，在此基础上对遗传算法进行改进；提出一种基于改进遗传算法的网格资源调度策略（GRSS_IGA），该算法综合考虑资源任务分配量、任务截止时间、任务等待时间及资源利用率等QoS参数；并用马尔可夫理论证明了算法的正确性；最后通过仿真对改进前后两种算法的性能进行比较，实验结果表明改进后的算法降低了时间消耗，提高了资源利用率。

关键词: 网格, 资源调度, 遗传算法, 马尔可夫理论