DRL-Empowered Topology Control for Time-Critical UAV Swarm Networks in Low-Altitude Economy

Fueled by the widespread deployment of uncrewed aerial vehicles (UAVs), the low-altitude economy (LAE) is experiencing a period of rapid growth, marking the entry of industry and commerce into a brand-new era. Although UAVs can effectively collaborate to accomplish LAE missions, the design of UAV networks presents numerous challenges, including high agility requirements, dynamically varying communication environments, and stringent time constraints. Therefore, to address these challenges, this paper proposes a deep reinforcement learning (DRL)-based topology control approach for UAV swarms operating under prescribed time constraints, ensuring reliable formation tracking despite dynamic communication conditions and temporal limitations. Specifically, to enhance the robustness of UAV networks against unpredictable communication losses due to sensor faults, a DRL-based topology decision-making algorithm is developed. This enables the topology structure to be dynamically adjusted in response to intermittent communication link failures among UAVs over specific time intervals. Furthermore, a novel prescribed-time UAV formation controller is designed to propel networked UAVs to reach the desired trajectory within a specified convergence time. Simulation results demonstrate that the proposed method is effective and outperforms the baseline schemes.