一种基于自动回归的改进网格主机负载预测模型

提出了一种基于自动回归（AR）改进的主机负载预测模型，它不仅具有AR模型本身的计算成本小、预测性能稳定的优点，还对AR模型只对未来某个固定时间段的负载预测进行了改进，使之能根据作业的预测执行时间进行主机负载动态预测，同时该模型还充分体现了主机负载变化的自相似性和长期依赖性。实验结果表明，该模型达到了预期的效果。

Improved AR-based model of host load prediction in computing grid

As the main part of resources prediction in the grid computing, host load is particularly important. An improved AR-based model of host load prediction was presented, The improved model not only has the advantages that AR model has, such as, low computing cost and high stability, and has the advantages that AR model has not. It can predict the dynamic interval host load that some job will encounter in the future, and process some characters of the host load variance, for example, self-similarity and long-range dependence. Experiment results show that the improved model has come to an anticipant result.