Spline-Extrapolation Method in Traffic Forecasting in 5G Networks

This paper considers the problem of predicting self-similar trac with a signicant number of pulsations and the property of long-term dependence, using various spline functions. The research work focused on the process of modeling self-similar trac handled in a mobile network. A splineextrapolation method based on various spline functions (linear, cubic and cubic B-splines) is proposed to predict selfsimilar trac outside the period of time in which packet data transmission occurs. Extrapolation of trac for short- and long-term forecasts is considered. Comparison of the results of the prediction of self-similar trac using various spline functions has shown that the accuracy of the forecast can be improved through the use of cubic B-splines. The results allow to conclude that it is advisable to use spline extrapolation in predicting self-similar trac, thereby recommending this method for use in practice in solving trac prediction-related problems.