Geomagnetic/PDR fusion localization method based on smartphones

Abstract Indoor pedestrian positioning serves as a crucial foundation for location-based services. Due to the ease of sensing geomagnetic signals, geomagnetic-based positioning methods have become a focal point in indoor pedestrian positioning research. To address the challenges of significant cumulative errors and low positioning accuracy in existing particle filter (PF)-based fusion methods, this paper proposes a geomagnetic/pedestrian dead reckoning (PDR) fusion localization method for smartphone-based indoor pedestrian positioning. This method integrates the particle swarm optimization algorithm with the traditional PF to achieve optimal position estimation and enhance real-time positioning accuracy. Furthermore, the dynamic time warping A* algorithm is introduced to generate variable-length geomagnetic sequences, enabling the correction of cumulative errors over time and mitigating the limitations of PF-based methods. Through experimental comparisons with mainstream positioning methods, the proposed approach demonstrates superior performance, achieving average errors of 0.90 m and 0.72 m in two typical scenarios, significantly outperforming PDR, MaLoc, and Magicol. Additionally, experimental validations conducted across different smartphone models and users of varying heights confirm the method’s applicability and stability. This robustness indicates the potential of the proposed method to support reliable indoor positioning across diverse devices and user conditions.