Mohammedelkhatim M.Canbilen A.E.2026-03-102026-03-1020259798331597535https://doi.org/10.1109/ISMSIT67332.2025.11268082https://hdl.handle.net/20.500.13091/13077This paper presents a reinforcement learning (RL)-based dynamic positioning system for unmanned aerial vehicles (UAVs), aiming to improve wireless coverage, energy efficiency, fairness, and emergency response capabilities. Based on deep Q-learning (DQL) algorithms, specifically deep Q-network (DQN), double DQN (DDQN), and dueling DQN, the proposed system enables a UAV to autonomously optimize its position to provide coverage to dynamically distributed ground users. Specifically, the considered system defines a multi-component reward function that balances coverage, energy consumption, fairness, and fast response to emergencies. Simulation results show that dueling DQN outperforms DQN and DDQN, offering 94% coverage, 40% energy reduction, and a robust emergency response while achieving near-optimal fairness. The proposed approach adapts to dynamic environments, providing a scalable and efficient solution for scenarios such as connectivity continuity and disaster management in harsh conditions. © 2025 IEEE.eninfo:eu-repo/semantics/closedAccessDeep Learning (DL)Reinforcement Learning (RL)Unmanned Aerial Vehicle (UAV)Conference Object10.1109/ISMSIT67332.2025.112680822-s2.0-105031088048