Global guided-DDPG based spatiotemporal optimization of UAV swarm networks for disaster emergency communications

Frequent natural disasters often cause traditional public communication networks to fail, as infrastructure damage severely limits disaster perception and emergency rescue efficiency. Unmanned Aerial Vehicle (UAV) Swarm Networks have emerged as a promising communication solution, offering rapid deployment, strong adaptability, and the ability to establish temporary networks without infrastructure, making them ideal for post-disaster rescue and emergency communication. However, optimizing the dynamic deployment of UAV network nodes especially to ensure both network efficiency and reliability remains one of the most significant challenges. We propose a spatiotemporal dynamic deployment method for UAVs using the Global Guided Deep Deterministic Policy Gradient (GGDDPG) algorithm, designed to optimize UAV positioning in disaster zones and improve communication quality between ground terminals and the network center. By integrating communication rate modeling based on real-world scenarios with reinforcement learning, this study designs an adaptive deployment framework. The algorithm dynamically adjusts the UAV movement path based on global location information to maximize network transmission efficiency and ensure stable connectivity for ground terminals. Experimental results show that the proposed algorithm outperforms other methods in various environments, significantly improving UAV network transmission quality and stability, and providing important theoretical and technical support for UAV-based disaster emergency communication networks.