ANOVA Decision Tool for Determining the Best Estimation of a Large Scale IP Network Traffic Matrix Using ARIMA/GARCH Algorithm

This paper deals with the Estimation of the Traffic Matrix (TM) of the well-known American ABILENE IP Network also called Internet 2. This work which has been processed on real collected data, aims to select the most precise ARIMA/GARCH (Auto Regressive Integrated Moving Average/Generalized Auto Regressive Conditional Heteroskedasticity) model for estimating the T.M. of such a Big Data Network by using the analysis of variance (ANOVA). The results have shown among many models that: ARIMA(1,1,1)/GARCH(1,1); ARIMA (0,2,1)/GARCH(1,1);ARIMA(1,2,1)/GARCH(1,1) have earned the challenge by yielding the most satisfactory results in terms of accuracy and precision. This fact has also been confirmed by implementing several validation methods like: Linear Regression; Normalized Mean Square Error (N.M.S.E); Root Mean Square Error (R.M.S.E).