A UAV-Assisted Handover Scheme for Coverage Maximization against 5G Coverage Holes

The Handover (HO) problem is widely explored by the research industry. In the dense traffic, the vehicles move at a lower speed, which means the vehicles will spend more time in the coverage holes. The absence of a communication link from the Next Generation NodeB (gNB) will degrade the Quality of Service (QoS) requirement of users. This motivates us to propose Unmanned Aerial Vehicles (UAVs) (e.g., drones) as temporary base stations to serve the traffic of User Equipments (UEs) in peak hour conditions. To overcome the HO delay, we propose a machine learning-based proactive HO scheme. In this paper, we train a Long Short-Term Memory (LSTM) model using Reference Signal Received Power (RSRP) values to predict and optimize HO decisions. Experimental results show that a UAV-assisted HO strategy can significantly enhance network performance in terms of the reduction of both Ping-Pong Rate and End-to-End Delay as performance metrics.