Machine learning based traffic analysis and forecast for 5G Systems

Mobile traffic forecasting is a relatively new research area, which is becoming of fundamental importance for next-generation networks. Proactively knowing the user demand allows the system to allocate resources and apply energy-saving decisions properly. Classical models are limited by the stationary assumption of time sequences and fail to take correlations into account. This work presents results on cellular network traffic analysis and prediction, providing a novel, robust, and precise machine learning model to efficiently and dynamically manage network resources in 5G systems.