WiFi-PDR indoor fusion positioning based on EKF

Abstract

In order to solve the problem that WiFi positioning technology is easily interfered by various factors in indoor positioning environment, and the cumulative error of pedestrian dead reckoning (PDR) technology based on inertial measurement unit (IMU) is large, this paper proposes a WiFi-PDR fusion positioning technology based on Extended Kalman Filter (EKF). In order to avoid the multipath effect of modeling with received signal strength (RSSI) data, offline database modeling is performed with channel state information (CSI) data. The collected CSI data are pre-processed with Hampel filtering and discrete wavelet transform (DWT) for noise reduction, and a WiFi location fingerprint database is established. On the basis of the peak value method, the double threshold method is added to detect the gait of pedestrians, and feedback adjustment is added to correct the heading angle online to improve the estimation accuracy of the pedestrian heading angle. The EKF algorithm is used to fuse the WiFi-PDR data, and a fusion positioning model is established to realize the real-time positioning of pedestrians. Experiments show that the positioning accuracy of the fusion positioning model proposed in this paper reaches 1.23m, which reduces the cumulative error of PDR positioning to a certain extent, and improves the accuracy and reliability of WiFi-PDR indoor positioning.