WiFi Fingerprint-Based Multi-user Localization for Smart Home by WiFi Sensing

Device-free and passive indoor localization based on WiFi channel state information (CSI) has attracted a great deal of research interest and yielded a broad range of related applications. However, most of existing approaches are influenced heavily by random noises, and also they fail in the presence of multiple users in the same area. In this paper, we present an accurate indoor localization scheme for multiple users using BP neural network deployed on commodity WiFi. To this end, this proposed approach mainly exploits three key techniques. Firstly, we carry out data preprocessing to obtain informative signals and then eliminate the random noises present in CSI measurements. Secondly, feature extraction is conducted to characterize multi-user profiles through combining data fusion and singular value decomposition (SVD) methods. Finally, BP neural networks is introduced to learn these features extracted and realizes the goal of multi-user localization. Furthermore, we implement the proposed approach on a set of WiFi devices and further evaluate it in typical indoor scenario. The experimental results relying on real-world data show that this approach can achieve a satisfactory performance in multi-user localization compared with existing approaches.