一种基于小波变换和PSR-LSSVM的P2P网络流量预测模型

提出一种基于小波变换、相空间重构理论和LS-SVM的P2P流量预测模型。首先将P2P流量分解为小波系数和尺度系数,然后分别对各个系数进行相空间重构,将重构的分量分别通过LS-SVM模型进行预测,最后用小波方法将各个分量的预测值进行再重构,得到原始流量的预测结果。仿真结果表明该模型的预测结果较传统的LS-SVM模型有更高的精度。

A p2p Network Forecasting Model Based on Wavelet Decomposition and PSR-LSSVM

Propose a p2p traffic forcasting model based on wavelet transform,phase space reconstruction theory and LS-SVM.First, p2p traffic is decomposed into wavelet components and scaling components ,Then use the phase space reconstruction theory to determine the optimal delay time and minimum of each component.And we use the LS-SVM model to train and predict.Finally, reconstruct each component to get the predicting result of the oringin p2p traffic using wavelet method.Simulating results show that the prediction of this new model approaches higher accuracy than the traditional model.