Active RIS-Assisted UAV-Enabled Fair Data Collection Systems Under Energy Constraints: A Hierarchical DRL-Based Approach

In this paper, we investigate an energy-constrained unmanned aerial vehicle (UAV)-enabled data collection system assisted by an active reconfigurable intelligent surface (RIS), where the UAV departs from a helipad, sequentially collects data from multiple ground Internet of Things nodes (GINs), and reaches a designated data center while ensuring fairness among GINs. The crux of the problem is to ensure that the UAV, under a limited system energy budget, precisely completes its flight from the helipad to the data center while maximizing system throughput and maintaining fairness. Specifically, the objective is to jointly optimize the UAV trajectory, user association, and RIS phase shifts to maximize system throughput and data collection fairness, while considering the system's limited energy resources. Given the highly non-convex nature of the problem, we propose a hierarchically decomposed deep deterministic policy gradient with active RIS (HD-DDPG-AR) framework, where the upper layer utilizes the deep deterministic policy gradient algorithm to optimize the UAV flight trajectory, while the lower layer derives closed-form solutions for user association and the RIS phase configuration. To verify the superiority and effectiveness of the proposed framework, nine benchmark schemes are designed for comparison, and experiments are conducted under varying system energy constraints to evaluate the performance of the proposed HD-DDPG-AR framework. Extensive simulations demonstrate that the proposed framework achieves overall performance improvements of up to 15.23% over passive RIS systems, up to 20.86% compared to systems without RIS, up to 18.92% over fixed-speed baselines, and up to 22.24% over traditional dynamic programming algorithms.