Forecasting of Data Traffic in Sudan Internet Exchange Point Using Autoregressive Integrated Moving Average Model

Abstract

Network traffic prediction is an important issue that has received much interest recently from computer network community. The network traffic prediction is one of the typical issues useful for monitoring network, network security, avoid congestion and increase speed of networks. Different techniques are used for network traffic prediction; first are linear time predictors such as Last Value (LV) Predictor, Windowed Moving Average (MA), Double Exponential Smoothing (DES), Auto Regressive (AR), Autoregressive Moving Average (ARMA) and Autoregressive Integrated Moving Average (ARIMA). The Second techniques are nonlinear time predictor such as generalized autoregressive conditional heteroscedasticity (GARCH) and the last one is hybrid model techniques, and it is combination between two or more models. From literature review, the ARIMA model is the best model to predict data traffic that has time series specification and linear growth. Data traffic collected from Sudan Internet Exchange Point (SIXP) and ARIMA is used to model data and forecast the next five years values. In addition to forecasting values, upper, and lower values are also founded. Furthermore, ARIMA model findings are compared with Monte Carlo forecast and it is found that the results are nearly typical. Finally, when the forecasted results are compared to actual data traffic on January 2020, the predicted values are shown to be very close to the measured ones.