Proximal Policy Optimization-Based Hierarchical Distributed Auction for Dual-Layer Inter-Layer Link Scheduling in Remote Sensing

With the growing demand for remote sensing observations, massive high-resolution data generated by low Earth orbit (LEO) remote sensing satellites (RSSs) urgently needs to be transmitted to ground stations (GSs) through a three-layer space information network composed of RSSs, communication satellites, and GSs. However, limited communication resources and complex system structures pose severe challenges for efficient scheduling of massive remote sensing data. This paper addresses the dual-layer inter-layer link scheduling in remote sensing problem (DILLS-RSP) and establishes a scheduling model that aims to maximize download task benefits while comprehensively considering energy constraints, time window constraints, and the dependency relationship between task upload and download phases. To tackle the DILLS-RSP, a proximal policy optimization based hierarchical distributed auction algorithm (PPO-HDAA) is designed, the proposed PPO-HDAA introduces a learning-assisted adaptive strategy selection mechanism and designs a task splitting mechanism to optimize scheduling schemes through hierarchical interactive auction mechanisms in both upload and download phases. Extensive experimental results demonstrate that PPO-HDAA outperforms various comparative methods in both solution quality and computational efficiency, while exhibiting robustness.