Forecasting Of Wireless Networks' Traffic Based On Minimax Probability Machine

As a new network access technology, the wireless network has widely used in every areas, attributing to its flexible and convenient. Meanwhile, the security and network management have become the core studies for the wireless network. According to the analysis, at present, the researches of the traffic forecasting were mainly about the wired network, while there are few studies on the wireless network. Because the wireless network has more complicated and more unstable than the wired network, the traditional forecasting models are not fitting to wireless local network. Those methods can not display the characteristic of the wireless network, lead to the poor accurate and result in the worse efficiency of the traffic anomaly detection. So, this paper did some researches on the wireless network traffic analysis and traffic forecasting:Firstly, this paper employed the marginal distribution, self-similarity analysis, fractal analysis and so on, to prove the wireless network have burstly and chaos multifractal characterize. Based on the characters of the network traffic analysis and the study of the forecasting theoretic, we introduced the Minimax Probability Machine (MPMR) as a forecasting method for wireless network traffic.Secondly, analyze and improve the forecasting method. The experiments mainly included the length of data set, kernel function, epsilon values and other study about MPMR's parameters; using the best parameters, forecasted the traffic for the wireless network by MPMR algorithm, and compared with SVM algorithm; improved the forecasting model. Through studied the forecasting method, we did some improvements'on the distance between vector, and take the residual analysis, model amend to improve the forecasting accuracy. At the experiment, we verified the model and its improvements were efficacious, and it could make forecasting of wireless traffic more accuracy.