Resilience-Driven Topology Reconfiguration via Hierarchical Deep Reinforcement Learning in Low-Altitude UAV Networks

In low-altitude wireless networks, Unmanned Aerial Vehicles (UAVs) are susceptible to environmental disturbances that may disrupt topology and degrade coverage and backhaul performance. To address this, we propose UR-HDRL (UAV network Reconfiguration based on Hierarchical Deep Reinforcement Learning), a hierarchical framework that decouples safety constraints from communication optimization. By integrating Control Barrier Functions (CBFs) and Graph Neural Networks (GNNs), UR-HDRL ensures safety and enhances cooperative decision-making in obstacle-rich environments. Experimental results demonstrate notable improvements in transmission efficiency, coverage, and collision avoidance over baseline methods.