The Characteristics Analysis And Prediction Model Study Of Campus Network Traffic

Along with the rapid development of the network and the scale of network expanding gradually, network control mechanisms and behavioral characteristics are becoming increasingly complex and difficult to understand. Because of the characteristics of network traffic data actually reflect interaction and impact between the network and itself, so through the in-depth analysis and modeling of the network traffic, an effective method of study on control mechanisms and characteristics of network is provided. The accurate and objective traffic model not only can help our understanding on complex network behavior characteristics, but also can play a very important role on improving network performance, monitoring network anomaly, etc.Wavelet has close contact with Fractal, the thinking of multiresolution of wavelet transform is similar to the partial and the whole idea of Fractal. So new ideas of network traffic prediction model can be provided by combining wavelet and fractal.Firstly, this dissertation analyses network traffic characteristics. Based on analysis and comparison of various methods of parameters estimation of the current self-similar network traffic, self-similar parameters of campus network traffic is estimated using R / S method, the results showed that the campus network traffic has obvious self-similar characteristics.Secondly, according to the campus network traffic characteristics and using the current traffic modeling method, ARMA modeling method based on wavelet decomposition is proposed. The experimental results show that prediction accuracy of this model is significantly higher compared with traditional ARIMA model.Finally, this dissertation analyses self-similarity of the impact on network performance, and proposes that a better approach to improve the network performance is to increase the network bandwidth.