ACE: Accurate and Automatic CSI Error Calibration for Wireless Localization System
Feiyu Han, Chengchen Wan, Panlong Yang, Hao Zhang, Yubo Yan, Xiang Cui
Abstract
Wireless localization based on commodity WiFi infrastructures has become one of the most important functions provided by ubiquitous mobile devices. Unfortunately, it heavily relies on the accuracy of the CSI measurements, which contain several kinds of phase distortion. Specifically in this work, we explore the source of CSI errors for WiFi APs with multiple antennas, which are the phase offset between RF chains and have a π phase shift. In order to mitigate these errors in an accurate and automatic way, we devise a customized hardware design with a carefully designed algorithm, which could ensure accurate and automatic error calibration. We implement and validate our system, and conduct extensive evaluation tests. The results show that our system can achieve a median AoA error of 8 degrees in LOS scenarios. The state-of-the-art indoor localization system SpotFi can benefit from our system, which achieves an improvement of nearly 15cm in median localization error and 40 cm in 90 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> percentile tail localization error.