A Data Fusion Method of Indoor Location Based on Adaptive UKF

Abstract

Focus on the problem that indoor location accuracy is generally low and various indoor location technologies are not widely used because some factors such as cost and accuracy. A data fusion method based on adaptive unscented Kalman filter (UKF) indoor location is proposed by analyzing the limitations of signal strength value (RSSI) fingerprint location, geomagnetic localization and inertial navigation location. The algorithm uses six-position error calibration method and Kalman filter to compensate the MEMS-SINS data, and establishes the correlation between location data and RSSI/geomagnetic data based on feature sorting vector fingerprint matching method. Finally, it is proposed to combine the adaptive factor with the unscented Kalman filter for data fusion, which improves the data stability and indoor location accuracy. The experimental results show that the adaptive UKF data fusion using MEMS-SINS/RSSI/geomagnetic data in the indoor environment can combine various advantages and achieve high-precision indoor location with an average absolute position error of 0.563m under the premise of low cost.