Fusion algorithm of improved fingerprinting/PDR/Map based on Extended Kalman Filter (EKF)/Particle Filter(PF)

摘要

The indoor location service has become the hotspot in both academia and industry due to the increasing demands of navigation for complex indoor environments. In order to improve the performance of positioning systems in indoor dense cluttered environments, an improved fingerprinting algorithm which fuses Received Signal Strength Indication (RSSI) and Time of Arrival (TOA) (named FP-RT) is proposed to mitigate the fluctuation of RSSI measurements caused by indoor multi-path fading and building obstacles. Meanwhile, a novel indoor positioning system integrating FP-RT, Pedestrian Dead Reckon (PDR) and map information based on second-order cascaded Extended Kalman Filter (EKF) and Particle Filter (PF) (FPM-EP) is proposed to decrease fluctuation of fingerprinting and accumulated error pf PDR. In FPM-EP system, the estimated results of FP-RT are used to decrease the cumulative drifts of Pedestrian Dead Reckon (PDR) based on Extended Kalman Filter (EKF). Moreover, the map information is introduced to constrain the FP-RT/PDR-derived position via Particle Filter (PF). Experiments are conducted in indoors, and the performance of the FP-RT, the fingerprint-based positioning without TOA-aided (named FPWT), the FPM-EP algorithm, EKF-based positioning and PF-based positioning were evaluated. The experimental results show that the mean error of FP-RT is 1.67m, which is improved by 22% in terms of positioning accuracy. Moreover, the mean accuracy of the proposed fusion system is 0.84m, which is an improvement by 49.7% and 64.2%, 34.8% and 33.3%, compared with FP-RT, PDR, EKF-based and PF-based algorithm respectively.