arXiv:2605.02413cs.NIcs.LG2026-05

用时空学习实现低轨卫星网动态路由,提升吞吐量并减少拥堵。

Spatial-Temporal Learning-Based Distributed Routing for Dynamic LEO Satellite Networks

  • 融合图注意力与LSTM的深度Q网络,基于局部信息做分布式自适应路由。
  • 相比传统方法,吞吐量更高,端到端延迟更低,队列长度减少23.26%。
  • 计算开销低、碳排放极少,适合绿色智能卫星网络部署。

本文提出一种面向动态低轨卫星网络的时空学习型分布式路由框架,将图注意力网络(GAT)与长短期记忆网络(LSTM)集成于基于深度Q网络(DQN)的架构中,实现基于局部观测的分布式自适应路由决策。路由问题被建模为部分可观测马尔可夫决策过程(POMDP),以应对动态拓扑和时变流量下的部分可观测性。仿真结果表明,该方法在吞吐量、丢包率、队列长度和端到端延迟方面显著优于传统及学习型路由方案,同时实现主动拥塞避免,队列长度最高降低23.26%。此外,该方法计算开销极低,碳排放可忽略不计,体现了绿色AI的高效性。

原文摘要 · Abstract (English)

In this paper, we propose a spatial-temporal learning-based distributed routing framework for dynamic Low Earth Orbit (LEO) satellite networks, where graph attention networks (GAT) and long short-term memory (LSTM) are integrated within a deep Q-network (DQN)-based architecture to enable distributed and adaptive routing decisions based on local observations. The routing problem is formulated as a partially observable Markov decision process (POMDP) to address partial observability under dynamic topology and time-varying traffic. Simulation results show that the proposed method significantly outperforms conventional and learning-based routing schemes in terms of throughput, packet loss, queue length, and end-to-end delay, while achieving proactive congestion avoidance with up to 23.26% queue reduction. In addition, the proposed approach maintains low computational overhead with negligible carbon emissions, demonstrating its efficiency from a Green AI perspective.

卫星网络动态路由绿人工智能深度强化学习

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。