A Reinforcement Learning Approach for Fair User Coverage Using UAV Mounted Base Stations Under Energy Constraints

Abstract

Unmanned Aerial Vehicles (UAVs) are gaining popularity in many aspects of wireless communication systems. UAV-mounted mobile base stations (UAV-BSs) are an effective and cost-efficient solution for providing wireless connectivity where fixed infrastructure is not available or destroyed. However, UAV-BSs have their limitations and complications, for instance, limited available energy. In addition, when several UAV-BSs are deployed to provide coverage to a specific area, the possibility of inter-UAV collisions and the interference to ground users increase. We propose Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL) based methods to deploy UAV-BSs under energy constraints to provide efficient and fair coverage to the ground users, while minimising inter-UAV collisions and interference to ground users. The proposed methods outperform the baseline methods by an average increase of 38.94% in system fairness, 42.54% in individual user coverage, and 15.04% in total system coverage, in comparison with the baseline methods.