Traditionally, the design of a cellular network focuses on the optimization of energy and resources that guarantees a smooth operation even during peak hours (i.e. periods with higher traffic load). However, this implies that cells are most of the time overprovisioned of radio resources. Next generation cellular networks ask for a dynamic management and configuration in order to adapt to the varying user demands in the most efficient way with regards to energy savings and utilization of frequency resources. If the network operator were capable of anticipating to those variations in the users’ traffic demands, a more efficient management of the scarce (and expensive) network resources would be possible.
Current research in mobile networks looks upon Machine Learning (ML) techniques to help manage those resources. In this case, you will explore the possibilities of ML to detect abnormal behaviors in the utilization of the network that would motivate a change in the configuration of the base station.
The purpose of this model is to accurately as possible predict the abnormal behaviors in the utilization of the network that would motivate a change in the configuration of the base station.