Representation Learning for Spatial Reuse in IEEE 802.11ax-Compliance Edge Intelligence

IEEE 802.11ax standard supports dense deployment of access points (APs)/edge devices, with a focus on robustness and uplink transmission. Dense deployments of IEEE 802.11ax APs use virtual carrier sensing to mitigate the effects of interference. Other challenges of IEEE 802.11ax compatible edge devices under dense deployment include homogeneous and heterogeneous coexistence and backward compatibility with legacy devices. To address these challenges, in this paper, two representation learning approaches based on graph neural network (GNN), called as direct-affinityGNN and skip-affinityGNN. Extensive evaluations demonstrate the effectiveness of both the approaches to enable high-capacity edge intelligence.