{"id":"arxiv/2610.09213","url":"https://arxiv.org/abs/2610.09213","tags":["eess.SY","cs.IT","math.OC"],"score":0,"title":"Mission-critical spectrum sharing with decentralized Multi-Agent Reinforcement Learning","domain":"eess.SY","source":"arxiv","authors":["Dimitrios Pylorof","Imtiaz Nasim","Humberto E. Garcia","Vivek Agarwal","Jasni A. Mannil","Mingyue Ji"],"reviews":0,"subject":"eess.SY","summary":"Motivated by emerging mission-critical applications and an increasingly congested spectrum, we develop a decentralized multi-agent reinforcement learning (MARL) model for dynamic spectrum access. The model enables secondary users to learn effective transmission strategies across shared frequency bands while minimizing collisions with high-priority primary users and among themselves. We design the agent-level learners following a Markov potential game approach, connecting independent local updates to system-level improvement. We instantiate this design using lightweight linear actor-critic…","category":"research","comments":0,"disputed":false,"division":"eess","mentions":0,"discussed":0,"published":"2026-10-06","reactions":{}}