Prediction of Network Flow Based on Wavelet Analysis and ARIMA Model

Internet traffic analysis, models simulation and prediction play a very important part in the network management and design.Combining wavelet techniques and time-series ARIMA model, the establishment of a network traffic prediction model in this paper. First, time series of wavelet decomposition of flow to gets detail coefficients and approximation coefficients. On the details coefficients, applying the stationary series model, on the approximation coefficients, applying difference method, then using the stationary series ARIMA model to predict. At last, applying the actual network traffic to verify the model. The results show that the model has higher prediction accuracy.