Combining wavelet transform and Bayesian least squares supportvector machines to predict network traffic
LIU Yuana,b,WANG Penga
(a.School of Information Engineering,b.Digital Media Research Center, Jiangnan University, Wuxi Jiangsu 214122, China)
Abstract:In order to improve the precision of the network traffic prediction,this paper proposed a new method of network traffic prediction combining wavelet transform and least squares support vector machines (LS-SVM).The original network traffic time series was decomposed into approximate series and several detail series. The result of single branch reconstruction of each decomposed series was more unitary than the original series in frequency, and it could be built traffic model with LS-SVM.The Bayesian evidence framework was applied to LS-SVM in order to determine the regularization parameters and kernel parameters effectively. The prediction of the original series could be obtained by the synthesis of each reconstructed series’ prediction result. The simulation results indicate the high accuracy and speed of this method. ......