Abstract: Traffic modeling and understanding are imperative to network design and simulation, to providing the quality of service(QoS), and to network management and control. In this paper, a practical conclusion has been arrived at that we can transform the nonstationary and growing network traffic to the domain of the stationary signal by the ARIMA modeling. By using this approach, we analyze the real traffic and the results prove the effectiveness and practicability of this approach. It is effective approach to traffic prediction in the middle/long-stage design of networks.

Key words: network traffic, ARIMA modeling, data analysis