A time series model of long-term NSFNET backbone traffic

We used time series analysis to create detailed forecasts of future NSFNET backbone traffic. The resulting autoregressive integrated moving average process (ARIMA) model made quite accurate forecasts of traffic levels up to a year in advance. It appears that the model can make reasonable predictions for two or more years into the future, suggesting that ARIMA modeling has great promise as a tool for long-range NSFNET forecasting and planning