Wireless MAC Protocol Synthesis and Optimization With Multi-Agent Distributed Reinforcement Learning

In this letter, we propose a novel Multi-Agent Deep Reinforcement Learning (MADRL) framework for MAC protocol design. Unlike centralized approaches, which rely on a single entity for decision-making, MADRL empowers individual network nodes to autonomously learn and optimize their MAC from local observations. Our framework is the first of a kind that enables distributed multi-agent learning within the ns-3 environment, and facilitates the design and synthesis of adaptive MAC protocols tailored to specific environmental conditions. We demonstrate the effectiveness of the MADRL framework through extensive simulations, showcasing superior performance compared to legacy protocols across diverse scenarios.

I. Introduction

Wireless networks have to satisfy multitude of requirements, ranging from high-reliability and low-latency connectivity to support of bandwidth-intensive applications such as virtual reality (VR), gaming, and holographic video. Thus, there is a growing need for adaptable and application-specific channel access mechanisms and resource allocation. The large number of configurable parameters and their entangled inter-dependencies in the medium access control (MAC) layer pose challenges to optimize and fine-tune protocols with traditional methods due to the dynamic nature of wireless networks [1]. This is especially pronounced in uncoordinated environments such as unlicensed bands, where there is no centralized authority to regulate channel access across all network nodes.