A Kalman Filter Based Indoor Tracking System via Joint Wi-Fi/PDR Localization

Abstract

The indoor localization has attracted great attention in both academia and industry with the growing demands on indoor location-based services. However, existing Wi-Fi fingerprint-based localization solutions are not sufficient for tracing moving target due to the unexpected noises of the Wi-Fi measurements. In view of this, this study is dedicated to implementing a robust indoor tracking system via the joint of Wi-Fi fingerprint and the PDR (pedestrian dead reckoning) techniques. Specifically, we first propose a two-steps joint localization approach, in which K-Nearest-Neighbor (KNN) is adopted to estimate the initial location based on Wi-Fi fingerprint, and then the PDR is adopted to refine the location based on the acceleration and the forwarding direction, which are detected by the embedded sensors in mobile devices. Specifically, the displacement is estimated via the direction, the stride length and the acceleration. Then, the Wi-Fi fingerprintbased estimated position and the PDR estimated position are fused based on Kalman filter. The main idea is to set the Wi-Fi fingerprint-based position as observation vectors, which are further used to update the PDR estimated positions. Finally, we conduct a series of experiments in a real-world environment, which demonstrate the effectiveness and the robustness of the developed indoor tracking system. In addition, our filed testing also demonstrates that the proposed algorithm can effectively improve the indoor localization accuracy.