Mission-critical spectrum sharing with decentralized Multi-Agent Reinforcement Learning

2026-10-06

AI Summary

This study addresses spectrum congestion and transmission collisions between secondary and primary users in mission-critical scenarios by proposing a decentralized multi-agent reinforcement learning framework for dynamic spectrum access. Grounded in Markov potential game theory, the approach establishes a mapping between local policy updates and global performance optimization. Furthermore, it introduces a lightweight linear actor-critic algorithm that eliminates reliance on computationally intensive centralized architectures, making it well-suited for resource-constrained edge devices. Experimental results demonstrate that, compared to heuristic baselines, the proposed method reduces the overall transmission collision rate by up to 96.8% while effectively preserving network throughput.

Abstract

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 learners suitable for resource-constrained edge devices, rather than computationally intensive centralized or deep multi-agent architectures. Across spectrum environments with different incumbent activities, the learned policies adapt their transmission policy and waiting behavior to preserve throughput while greatly reducing transmission collisions relative to random and forecast-aware heuristic baselines. The results establish the value of decentralized MARL and shows up to 96.8% reduction in overall collisions.

Problem

• Spectrum Sharing
• Dynamic Spectrum Access
• Multi-Agent Reinforcement Learning
• Collision Minimization
• Mission-Critical Applications

Innovation

• Decentralized Multi-Agent Reinforcement Learning
• Dynamic Spectrum Access
• Markov Potential Game
• Linear Actor-Critic
• Edge Devices