Research on High Robust Device-Free Localization Model for Multiple Applications
Xiao-Jiang CHEN,Li-Li CHEN,Bo-Hang LI,Zhan-Yong TANG,Bin-Bin XIE,Wei WANG,An-Wen WANG,Ding-Yi FANG
DOI: https://doi.org/10.11897/SP.J.1016.2019.00403
2019-01-01
Abstract:Device-free localization is the core technology that enables a wide variety of applications such as elderly care, smart home, and wildlife monitoring.Recently, wireless signal based technologies attract more interest.Their key advantages over the camera and sensor-based approaches are that they do not need users to carry any devices as they rely on the wireless signals reflected off users while providing better coverage as they can operate through walls.To our best knowledge, however, existing wireless based device-free localization methods either require large scale training effort or require dedicated devices which have poor universality, otherwise, apoor localization performance will be achieved.In this paper, we present MaLoc, a robust device-freelocalization model for multiple applications, which does not require extensive prior learning and can work effectively with any devices regardless of whether they are cheap commercial or dedicated.The intuition underlying our design is that as long as the object blocks the wireless signal coming from a certain direction, the corresponding direction of arrival will experience a sharp drop in signal strength.By leveraging the direction of arrival information at the receiver, we are able to estimate the target's location.The main innovation of this paper involves two aspects. (1) In order to remove phase error of the commodity WiFi device caused by multipath and hardware noise, a linear transformation is employed to the raw phase values, then the stable phase differences are derived for the antennas.By leveraging the stable phase differences among the antenna array as well as the reference antennas, we are able to localize the target. (2) To solve the arrival angle offset problem caused by background noises fluctuations, we design an angle error compensation strategy to calibrate the arrival angle from the target, further improving the robustness and localization accuracy.Finally, we have conducted comprehensive experiments to verify the effectiveness and robustness of MaLoc.The experimental results show that MaLoc achieves the median error of around 0.75 mand 1.5 min the poor multipath environment and in the rich multipath environment.We believe that MaLoc will be apromising candidate in a wider range of wireless sensing applications, and the proposed model can be extended to other topics such as human gesture recognition and target trajectory tracking.