Hybrid precoding algorithm for Wi-Fi interference suppression based on deep learning

Interference among wireless access points (APs) in Wi-Fi systems limits the throughput of multi-AP massive multiple-input multiple-output systems, and as the AP density increases, the increased interference leads to a significant loss of spectral efficiency of the system. Suppose interference is suppressed by obtaining information about all interfering channels, although the spectral efficiency of the system is greatly improved. In that case, the communication overhead between APs is too huge and consumes too many resources for coordinated transmission, and the performance improvement obtained is negligible. Based on this, a new deep learning hybrid precoding technique based on local channel information is proposed in this paper, where APs use local channel state information for direct hybrid precoding, which can effectively suppress inter-AP interference in dense wireless local area network and improve the reachable rate of the system through the characteristics of deep learning networks. Through multi-AP system-level simulations, it is demonstrated that this non-collaborative hybrid precoding method based on deep learning greatly suppresses interference and effectively improves the spectral efficiency of the system.