Abstract: In view of the characteristics of WiFi positioning, PDR (pedestrian track calculation) positioning and geomagnetic positioning, an indoor positioning method based on fusion of WiFi, PDR and geomagnetism is proposed. Firstly, the proposed fusion localization approach fuses the results of WiFi localization and PDR localization by UKF (Unscented Kalman Filtering) algorithm, and obtains an improved localization result, then, with the improved positioning result as the center, the size of the range is dynamically controlled by the standard deviation of the error in the improved positioning result and the WiFi positioning result of this time, finally, second positions are made by geomagnetic matching in this range, and the final positioning results are obtained. Through experiments and simulation, the average positioning error of the proposed fusion positioning method is as low as 1.43M, and the positioning accuracy is improved greatly.