Trajectory-Aware UAV-Enabled WPT Networks Based Grid World DRL Approach

In this research, we investigate a wireless power transfer (WPT) system involving an unmanned aerial vehicle (UAV) equipped with an array of antennas to wirelessly charge ground user (GU) devices. Our objective is to enhance the lowest GU energy levels by optimizing the UAV’s trajectory, beam-forming strategy, and transmission power simultaneously. Since optimizing the lowest GU energy presents a challenging non-convex problem, we reformulate it as a discrete-time grid world problem. We propose a deep reinforcement learning (DRL) approach to optimize this problem by determining the UAV’s movement direction, beam-forming angle, and transmit power level. We also integrate the water-filling algorithm with DRL to aid in determining the optimal hovering duration. Through simulations, we demonstrate that our approach significantly improves GU energy levels compared to the successive hover-and- fly algorithm while maintaining low computational complexity.