SmartSARIMAX: An Advanced Model for Bandwidth Prediction in Data Networks

Abstract

In the data network field, particularly in the domain of fast evolving data networks, it is necessary to have a good bandwidth estimating for resource planning and user guarantee because the system of weft or dynamics between data flows is increasingly and progressively complex in the structures topology. In this paper, we present a novel forecasting method known as SmartSARIMAX (S_SARIMAX) based on the Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX) approach to estimate bandwidth consumption. S_SARIMAX incorporates additional variables, including user behaviour patterns and historical bandwidth trends, to accurately simulate complex seasonal and network traffic trends. Our model is rigorously tested with real-world datasets, dramatically improving prediction accuracy over standard methods. The results show that the S_SARIMAX model provides reliable predictions to support strategies to stimulate network management processes with an MAE and RMSE as forecasting metrics and the proposed model outperforms the comparable model by more than 90%. This study presents essential contributions to bandwidth prediction, offering a strong asset for network operators to predict the demand, plan capacity and develop the users' Quality of Experience (QoE).