Reinforcement Learning-Based UAV Handover Algorithm in Cellular Networks: A Survey

UAVs execute handovers for stable communication while moving in the air. Conventional handover algorithms designed for ground users in cellular networks are not suitable as it can induce ping-pong handover on UAVs. Furthermore, the increase of altitude and speed of UAVs leads to handover failures. To address these issues for improving signal quality, research is underway. This paper discusses research trends on utilizing reinforcement learning in cellular networks to enhance UAV s handover performance. It proposes three handover methods using Deep Reinforcement Learning and Q-learning, respectively, and validates the reduction of hand over ratio by adjusting the weight co in each approach.