Toward Reliable Indoor Tracking Based on Motion Features and Multipath-Assisted

Recently, indoor target localization and tracking with only one access point (AP) have attracted significant attention in wireless sensor networks (WSNs). Despite much research in this field, a universal and accurate solution is still rare in reality. To this end, we propose a system that achieves reliable decimeter-level indoor tracking with only one AP in this article. We first use the 3-D orthogonal matching pursuit (OMP) to recover the angle of arrival (AoA), angle of departure (AoD), and time of flight (ToF) of the multipath signals. We construct a fusion tracking model based on these channel parameters, which can adapt to different situations. Concretely, when there are available reflection paths, we combine the target motion features and the geometric features of multipath reflections to achieve improved multipath-assisted tracking. In addition, we provide an auxiliary tracking method with only the direct path in the absence of available reflective paths. After that, we use a Bayesian framework to analyze the fusion tracking model and formulate it as a hidden Markov model (HMM). Then, we construct a particle filter to address the HMM instead of the Kalman filter because of the nonlinear model. Finally, we validate the proposed system in various indoor environments on a
$4\times $
4 multiple input multiple output (MIMO) system implemented by software-defined radio (SDR) X310 devices. The experimental results show that the proposed system has good tracking performance and a median trajectory error of 0.44 m, comparable to the prior state-of-the-art systems.