A Bayesian Model Based on Link Distribution Features for Multitarget Passive Localization in Visible Light Sensing

Passive localization using visible light (VL) sensing has been considered as a promising solution for indoor human detection. A major challenge is to avoid false target positioning in multitarget positioning scenarios. Besides, a reasonable probable target area model is also critical to the accuracy of positioning and counting targets. In this article, a novel passive localization scheme is proposed to locate multiple targets. This scheme introduces a rectangular probable target area model, which is used to calculate a rectangular area containing the potential location of the target to be located. Compared with existing models, it is more suitable for positioning scenarios under low-density deployment of sensing nodes. Furthermore, we present an improved successive cancellation (SC) algorithm to excluding false target localization. To determine the authenticity of targets by the SC algorithm, a Bayesian model is introduced to optimize the SC algorithm according to the multidimensional shadowed link information of candidate targets. Numerous simulation results show that the proposed multitarget passive localization scheme can improve the problem of false target localization in multitarget localization scenarios. And it also can achieve outstanding performance in localization accuracy.