QoS-Aware UAV-BS Deployment Optimization Based on Reinforcement Learning

We propose an unmanned aerial vehicle mounted base station (UAV-BS) deployment optimization scheme using reinforcement learning (RL). We formulate the objective function considering quality-of-service (QoS) of ground user equipments and fairness among them. To solve the optimization problem using RL, we propose a deep Q-network (DQN) model by defining the state, action, reward, and train UAV-BS to find the best movement direction. The simulation results show that the proposed DQN model is trained well, and the UAV-BS reaches the optimum position through our proposed DQN model.