关键词: 地震救援及灾后重建, 空天地一体化网络, 卸载决策, 深度强化学习

Abstract: In earthquake rescue and post-disaster reconstruction scenarios, a plethora of real-time tasks emerge within integrated air-ground-space networks. Leveraging task offloading decision-making techniques to enhance task processing speed is highly meaningful. To address the task offloading decision-making problem in scenarios involving multiple users and multiple servers, this paper designs a system model, communication model, computing model, and task offloading model, proposing a task offloading decision-making scheme based on deep reinforcement learning. Experimental results demonstrate that this scheme can reduce task offloading latency and system energy consumption in earthquake rescue and post-disaster reconstruction scenarios, thus exhibiting promising prospects for practical applications.

Key words: Earthquake Rescue and Post-Disaster Reconstruction, Space-Air-Ground Integrated Network, Offloading Decision-making, Deep Reinforce Learning Algorithm