Agentic Hierarchical Multi-Team MARL for Joint AI-Driven MLO and MAP-Co in Wi-Fi 7 Networks

The forthcoming IEEE 802.11be (Wi-Fi 7) standard introduces Multi-Link Operation (MLO) and Multi-AP Coordination (MAP-Co), including Coordinated Spatial Reuse (C-SR), to enable ultra-reliable, low-latency, and high-throughput wireless connectivity. However, existing approaches treat these mechanisms independently, limiting performance in dense, interferenceprone environments where link activation and channel reuse decisions are interdependent. This paper proposes an Agentic Hierarchical Multi-Team Multi-Agent Reinforcement Learning (A-HiMT-MARL) framework that jointly optimizes AP-STA association, transmit power control, traffic distribution, and multiAP coordination. The framework adopts an agentic design with specialized agents and orchestrator agents operating across two hierarchical levels, a controller level and an AP level, combining global coordination with local adaptation through structured interaction and sequential cooperative learning. The proposed architecture enables scalable and low-latency decision-making in dynamic environments. Simulation results show significant improvements in throughput and latency over state-of-the-art rule-based and MARL methods, demonstrating the effectiveness of agentic MARL for next-generation Wi-Fi networks.