Sum Secrecy Rate Enhancement in Low-Altitude Intelligent Networks with Mixed Obstacles

This paper investigates the key challenges of trajectory planning and resource allocation for unmanned aerial vehicles (UAVs) providing secure communication services in a low-altitude intelligent network with mixed obstacles. By utilizing physical layer security techniques, the goal is to maximize the long-term sum secrecy rate for ground users, while strictly ensuring both UAV flight safety and communication security. To achieve this, we formulate a joint optimization problem that incorporates realistic constraints, including three-dimensional nofly zones, ground obstacle areas, and limited onboard energy. Due to the high dimensionality, non-convexity, and coupling characteristics of the problem, we propose a solution based on the multi-agent deep deterministic policy gradient (MADDPG) framework. Specifically, we design local observation and action spaces for the UAVs and adopt a centralized training with decentralized execution (CTDE) mechanism to enable distributed decision-making. Additionally, we analyze the computational complexity of the proposed algorithm and demonstrate its scalability. Extensive simulation results confirm the superiority of our proposed scheme. Compared with four state-of-the-art schemes, it significantly improves the sum secrecy rate. Furthermore, we explore the impact of key network parameters on secure communication performance, providing practical insights for real-world deployment of low-altitude intelligent networks.