Network traffic forecasting by support vector machines based on empirical mode decomposition denoising
Yuan Qian, Jingbo Xia, Ke Fu, Rui Zhang · 2012
Network traffic forecasting plays an important part in network control. A new method based on the Empirical Mode Decomposition (EMD) denoising and Support Vector Machines (SVM) is developed to improve the accuracy of the traffic prediction. Firstly, network traffic data are preprocessed by EMD to remove noise. Then the denoised data are processed by phase space reconstruction to form the training samples. Last the SVM model is constructed to forecast the real network traffic. The results show that the new method is more effective for extracting noise and prediction precision is high.