DNN-based Outdoor NLOS Human Detection Using IEEE 802.11ac WLAN Signal

Recently, WLAN-based wireless sensing technologies, which utilize WLAN devices widely used in many environments, have been focused because of their low deployment cost. We have presented an outdoor human detector using IEEE802.11ac channel state information (CSI) for line-of-sight (LOS) scenarios in our previous work [1]. In this paper, we extend our previous work and present a CSI-based human detection system for outdoor non-line-of-sight (NLOS) scenarios. The key idea is to utilize CSI retrieved by multiple devices and extracted key features using principal component analysis (PCA) for sensing to avoid unstable detection performance. Experimental evaluations revealed that our human detection system for NLOS scenarios successfully located a human with an accuracy of 99.58 % using four WLAN stations.