FM Fingerprint Database Online Construction and Calibration based on Propagation Model and PDR Fusion Indoor Positioning

The demand for indoor positioning-based services grows rapidly. Fingerprinting technology has been widely applied in indoor positioning since its high accuracy and low dependence on external devices. However, the main dilemma limiting its development is the time-consuming and laborious site survey in the offline stage. In this paper, we use FM radio signal as the navigation source, and present a completely online FM indoor fingerprint database construction and calibration method based on the propagation model. With the floorplan obtained only, FM data initially collected inside the building is utilized to estimate the model parameters and build the fingerprint database. In the stage of online calibration and positioning, PDR is used to calibrate the database and constrain the fingerprint matching range, then an EKF-based fusion method is developed to improve the positioning performance and ensure the accuracy of calibration. Considering that the little data collected at the initial walking stage may cause inaccurate estimation of model parameters, different calibration strategies are designed using the least square and characteristics of FM fingerprint distribution to increase the stability and accelerate the convergence speed of database error. Field experiments are conducted in the campus office building, and the results indicate the effectiveness of the proposed method.