TRAFFIC MODEL BASED ON TIME SERIES TO FORECAST TRAFFIC FUTURE VALUES WITHIN A WI-FI DATA NETWORK

This work aimed to show that time series are an excellent tool for data traffic modelling within Wi-Fi networks. Box-Jenkins methodology, which is herein described, was used to achieve this objective. Wi-Fi traffic modelling through correlated models, like time series, allow to adjust a great part of the data behavior dynamics in a single equation and, based on it, to estimate traffic future values. All this is advantageous when it comes to covering planning and resource reservation as well as performing a more efficient and timely control at different levels of the Wi-Fi data network functional hierarchy. An 18-order ARIMA traffic model was obtained as a research outcome, which predicted the traffic with relatively small mean square error values for a 10-day term.