Forecasting 802.11 Traffic Using Seasonal ARIMA Model

Based on the analysis to the collected traffic from many WLAN testbed, a statistical model is proposed to predict the short-term traffic in IEEE 802.11 networks. By large numbers of differencing and sampling to the original data sequence, the season property was found and verified. Then, a time series model was given which can accurately predict the WLAN traffic, multiple seasonal arima model (0, 1, 1) (0, 1, 1). After iterative computation, the model was transformed into an MA model and the parameter of it has been estimated using the character of MA model. Finally, a prediction to the random selected WLAN traffic has been finished through the difference function. The result of the prediction present that the employed model can short-term forecast the WLAN traffic and obtains a better result with a tiny average relative error.