Robust Multitarget Device-Free Localization and Tracking via Visible Light Sensing

Abstract

Device-free localization (DFL) is a promising technique that can locate and track targets carrying no devices via detecting shadowed links caused by targets. There are two main challenging problems needed to be solved for the multitarget DFL. One is that pseudotargets can be produced by the intersection points of the shadowed links caused by different targets in the localization stage. The other is that targets’ identities may be wrongfully swapped when targets come close to each other in the tracking stage. To solve the two problems, a robust multitarget DFL framework using visible light sensing is proposed. The proposed framework introduces a positioning label to verify the authenticity of the estimated target. Moreover, a successive cancellation algorithm is utilized to obtain coarse target areas, which are used as the constraint information for calculating locations and features of targets. With the prior information related to the locations and features of targets, a multitarget tracking method based on the Hungarian algorithm and the spatial filter is proposed to solve the target assignment problem. Extensive simulation results show that the proposed multitarget DFL framework achieves outstanding performance both in localization and tracking accuracy.