一种基于灰色模型的资源可用性预测模型

网格环境中资源可用性信息的掌握是网格中间层应用的基础和依据,对于系统性能的提高和资源的有效利用发挥着重要作用。针对当前资源可用性预测难以有效指导网格应用的缺点,从应用的角度评估资源可用性并提出了一种基于灰色模型的资源可用性预测模型(Model for Resource Availability Prediction Basedon Grey Model,MRAPGM)。首先,结合当前监测方法分析资源可用性历史数据中噪声的存在。然后,在MRAPGM中,利用小波分析方法对资源可用性历史数据进行噪声过滤,再利用灰色预测模型预测出资源在未来某一时间段内的资源可用性。通过大量实验确定了模型中使用的小波以及相关参数,并且实验结果表明提出的预测模型是合理和有效的。

Model for Resource Availability Prediction Based on Grey Model

The knowledge of resource availability, which plays an important role in improving system performance and resource utilization, is the basis for Grid middleware applications. For the current resource availability prediction models cannot provide effective instruction for applications in Grid middleware, resource availability from application's perspective was evaluated and a Model for Resource Availability Prediction Based on Grey Model (MRAPGM) was proposed. It was analyzed that noise was exist in availability historical trace got from currently measurement methods. Therefore, in MRAPGM, noise was filtered out from resource availability historical data employing wavelet analysis, and then Grey model was utilized to predict resource availability during a future period. Extensive experiments were conducted to determine the wavelet and related parameters in MRAPGM. Experiment results show that MRAPGM is reasonable and achieves good prediction accuracy.