资源灰预测的反馈任务调度算法

Task scheduling algorithm based on resources grey prediction feedback

为了达到网格环境下任务调度时的负载平衡,针对此环境下的资源的自治性、异构性和分布性等特性,利用改进的灰预测模型GM(1,1)预测方法,设计了资源实时预测模型,可在较小的开销下取得满意的负载平衡。基于该模型的资源灰预测反馈任务调度算法RGP-FB是把资源预测融入到网格环境下的任务调度策略中,从而使系统调度的综合效率提高。仿真实验证明了该算法的合理性和有效性。

In order to achieve load balance in task scheduling under the grid environment, in view of the resources＇ autonomy, isomerism and distribution under this circumstance, a model of resources real-time prediction was designed based on improved GM （1, 1 ） prediction method, which could get satisfied load balance with smaller expenditure. The task scheduling algorithm named RGP-FB based on resources prediction model integrated the dynamic forecast into the grid environment under the task scheduling strategy, then it could enhance the comprehensive efficiency of the scheduling system. Simulation results show that the algorithm is valid and effective.