WiFi CSI Fingerprinting Positioning Based on User Rotation

Abstract

This paper studies the indoor fingerprinting positioning using Channel State Information (CSI) in commercial WiFi network environment. In this paper, we improve the existing indoor fingerprinting positioning method by considering the influence of human body absorption on CSI signal amplitude and collecting CSI data with user rotation at each reference location. The whole positioning process includes two stage: offline stage and online stage. In the offline stage, we extract features from the filtered CSI data of three APs at each reference location to construct CSI fingerprints. In the online stage, we first compare the feature vectors of filtered CSI data with fingerprints, and then calculate the Euclidean distance between the online CSI feature vector and fingerprints. Finally, user location will be obtained by the K Nearest Neighbor (KNN) algorithm. Experiments proved the performance improvement of the proposed CSI fingerprinting positioning method.