Deep-Learning Based Proactive Handover for 5G/6G Mobile Networks using Wireless Information

Future blockage prediction-based proactive handover (HO) is an essential feature in 5G/6G wireless networks to establish reliable connectivity between user equipment (UE) and base stations (BSs). As the 5G/6G networks employ millimeter wave (mmWave) frequencies, which are highly sensitive to blockages, it is imperative to devise solutions for proactive blockage and handover predictions. Traditional approaches to ensuring seamless connectivity include the UE-based monitoring of the link status and identifying future blockages using machine learning (ML) models. Other techniques are multi-base station connectivity to the UE, multi-cell multi-beam switching, HO execution between BSs using non-RF sensors, light detection and ranging (LiDAR), and visual information for accurate prediction of blockages. However, underutilization of network resources and heavy computational overhead were the limitations of the above approaches. In this paper, we propose a deep-learning-based solution for predicting a mobile UE’s future line of sight (LoS) link blockage in a dynamic urban environment. We utilize only wireless information to minimize the high computational requirements of proactive HOs for mobility management. Our proposed deep-learning algorithm uses an observation set of past link status to predict future link blockages. Thereby enabling proactive HO to ensure a reliable communication link. The results show the prediction accuracy of blockages is almost 90% by using wireless signatures only, which is 18% greater than the reported deep learning models.