Conflict-Aware MADRL for AoI-Driven Collaborative Mission Scheduling in Aerospace Integrated Networks

The aerospace integrated networks (AINs), leveraging satellites and unmanned aerial vehicles (UAVs), offers a promising solution for large-scale Internet of Remote Things (IoRT), effectively ensuring information freshness, i.e., low Age of Information (AoI). However, in resource-constrained and dynamic AIN environment, a key challenge is how to achieve fresh data by efficiently resolving mission conflicts across multiple IoRT devices, which requires advanced scheduling design for collaborative UAVs monitoring and UAVs-satellites data transmission. In this paper, we first construct a mission scheduling framework for collaborative monitoring and transmission utilizing the wide coverage of low earth orbit (LEO) satellites and the mobility of UAVs. Then, by considering constraints such as mission conflicts, energy consumption, and motion characteristics, we characterize the relationship between UAVs trajectories and IoRT demands. Based on this, we propose a multi-agent deep reinforcement learning (MADRL) algorithm that jointly optimizes UAV trajectories and transmission scheduling. The algorithm incorporates a filter layer to optimize UAV cooperation, preventing redundant device IoRT selection and resolving mission conflicts. Simulation results indicate that the proposed algorithm can reduce 22.9% AoI compared to the benchmark.

Ҳали таржима қилинмаган