Awareness of Line-of-Sight Propagation for Indoor Localization Using Hopkins Statistic

The non-line-of-sight (NLOS) propagation degrades the performances of mobile computing applications such as indoor localization when the line-of-sight (LOS) propagation between Access Point and client is blocked due to the mobility of users and building structure. Thus, the ability of sensing environment, i.e., awareness of LOS/NLOS propagation is significantly important. Many previous systems relied on channel state information statistic to construct LOS identification schemes by using the commodity Wi-Fi network. However, the vulnerability of these systems is that the robustness is not considered carefully since the system configuration such as operating channel affects statistic results. Up to now, the estimation of spatial parameters of multipath components (MCPs) is not challenging due to the advanced Multiple-input-multiple-output technique. In this paper, we propose a robust LOS/NLOS propagation identification system by using Wi-Fi signal. We first resolve the angle-of-arrival and time-of-flight of MCPs, and then creatively introduce Hopkins statistic to measure the spatial randomness of LOS and NLOS propagation, respectively. We finally employ the non-parametric method to calculate the optimal threshold of Hopkins statistic to differ the two cases. The experimental results show that the proposed system achieves LOS and NLOS identification rates over 91% and 87%, respectively under different system configuration. Furthermore, the upper-layer indoor localization applications can be benefited by cooperating with the proposed algorithm, and the case study of an indoor localization removes large localization errors effectively.