A New Method for Communication Traffic Prediction and Anomaly Detection

In this paper, a new method based on Auto Regressive(AR) model is proposed to predict the communication traffic data and detect if the traffic data is anomaly. The method includes four phases: extraction of traffic data characteristics，data analysis and model determination，determination of model parameters, traffic prediction and anomaly detection。With the simulation, it is obvious that our method can predict future traffic data with high consistency. Moreover, the computational complexity of our algorithm is acceptable for real work.