AI-Based Mitigation of Coverage Holes Through UAVs Path Planning

This paper proposes an efficient path-planning scheme for unmanned aerial vehicles (UAVs) aimed at addressing coverage holes in wireless networks. Coverage holes can undermine the quality of service (QoS) of terrestrial cellular networks where they cause outage times longer than a threshold value dictated by the different application requirements. The proposed approach leverages the self-organizing map (SOM), an unsupervised machine learning technique, to design a UAV trajectory that minimizes the flight path length, while ensuring a coverage hole-free cell or guaranteeing a maximum outage time across the existing holes. The designed path also satisfies constraints on minimum and maximum UAV velocity. Simulation results show that for realistic scenarios, we can practically eliminate all coverage holes when one UAV travels over the designed path. For more extreme scenarios, we show that we need to deploy multiple UAVs to satisfy the QoS requirements where each UAV covers a partition of the holes. To achieve optimal partitioning, we utilize the ant colony algorithm.