An interleaved channel state information clustering scheme for wireless LAN-based object detection systems

Abstract

In this article, we propose a frequency-domain feature clustering scheme for a machine-learning based object detection utilizing channel state information (CSI) in wireless local area network (WLAN) systems with multiple antennas where CSI frames are captured from nearby wireless devices. In this scheme, all subcarriers (their CSI) are divided into multiple clusters in an interleaved manner and the existence of object is detected by integrating majority decision of cluster-by-cluster machine-learning results. Simulation results show that the proposed interleaved clustering scheme achieves better object detection probability than cases with a localized clustering scheme and without clustering in an indoor propagation environment.