Resource Allocation Using Deep Learning in Uplink 802.11ax Networks

In the IEEE 802.11ax wireless local area network (WLAN) standard, multi-user multiple-input multiple-output (MU-MIMO) and orthogonal frequency-division multiple access (OFDMA) are two key technologies to enable concurrent transmissions of multiple data streams on the same time-frequency resource unit (RU). In a MU-MIMO-OFDMA enabled WLAN, finding the optimal resource allocation over time, frequency and space usually involves exhaustive search, which can be computationally prohibitive. In this paper, we study the optimal resource allocation to maximize the network throughput of an UL 802.11ax network. In particular, we propose resource allocation using deep learning (RAuDL) that jointly considers stations' channel conditions and their distributive power constraints in UL. Simulation results show that RAuDL can find near-optimal solution with few training samples.