Intelligent UAV Trajectory and Power Control in 6G Non-Terrestrial Networks Using Deep Reinforcement Learning

Unmanned Aerial Vehicles (UAVs) have started to become a primary driver for 6G Non-Terrestrial Networks (NTNs) in providing rapid, agile, and area-wide connectivity in challenging conditions. Smart management of the mobility of UAVs and transmission power, however, still remains a significant challenge, particularly under energy constraints and changing user locations. This work introduces a Deep Reinforcement Learning (DRL) approach on the basis of Proximal Policy Optimization (PPO) to both optimize the UAV trajectory and power control in a case of an NTN. A modified simulation platform is developed to simulate the UAV-user interaction, where the agent collects reward signals that trade-off signal-to-noise ratio (SNR) improvement and energy efficiency. Our proposed system enables the UAV to learn optimal flying and transmission policies without prior knowledge of the environment. Through extensive training and testing, we demonstrate spectacular improvement in reward stability, UAV motion intelligence, and constant power consumption. Visualization tools such as learning curves, action distribution histograms, and UAV trajectory plots verify the effectiveness of the learned policy. This research provides an scalable and self-sustaining solution for future wireless communication aided by UAV in 6G NTN networks.