LOS Path Identification in Multipath Environments with an Unknown Number of Passive Targets
Yifan Liang, Cengcang Zeng, Hongbin Li · 2024
This work extends a recently introduced type-based clustering algorithm (TCA) [1] for identifying line-of-sight (LOS) paths in multipath environments with multiple passive targets. In particular, while [1] assumes that the number of targets is known, we consider herein a more practical scenario without the assumption. The system consists of several spatially dispersed sensors, each emitting a unique waveform. These sensors exploit the returned echoes to measure both LOS and non-line-of-sight (NLOS) time delays (or equivalently, ranges) of targets within the observation area. For ease of exposition, we consider a 2-dimensional (2D) localization scenario. In this context, every range estimate represents a circle, and combining measurements from different sensors produces intersection points on the plane. By applying TCA and examining the structural properties of points created by LOS paths, we categorize the clustered point sets into six possible cases. To estimate the number of targets in presence, we analyze these cases and formulate a discriminant function to identify the sets associated with the targets, thereby enabling us to jointly determine the number of targets and the associated LOS measurements. We present numerical results to showcase the performance of the proposed scheme under various configurations.