UAV Trajectory Optimization for IoT Data Collection with User-level Fairness

Unmanned aerial vehicles (UAVs) have emerged as a cost-effective method to improve the coverage and reliability of wireless networks. Particularly, UAVs can be utilized as mobile base stations for data collection from low-powered devices such as those commonly utilized in Internet-of-Things (IoT) applications. In this work, we propose a reinforcement learning (RL) enabled Deep Q-Network (DQN) approach for trajectory planning in UAV-supported wireless networks. User fairness is considered in scenarios whereby the total demand of users cannot be satisfied within the available flight time. Results show effective trajectory planning is achieved that optimizes the total UAV's satisfied demand, while maintaining user fairness.