Multi-Agent Reinforcement Learning for Dynamic Satellite Constellation Management
Low Earth Orbit (LEO) mega-constellations are becoming central to global broadband delivery, yet their scale and constantly shifting topology make centralized, ground-based control impractical. This paper proposes a multi-agent reinforcement learning (MARL) framework in which each satellite acts as an autonomous agent...