Map assisted PDR/Wi-Fi fusion for indoor positioning using smartphone

In this paper we present a map-assisted pedestrian navigation system for smartphone user which combines map information, IMU-based Pedestrian Dead Reckoning (PDR) and Wi-Fi localization using fingerprinting method. PDR (Pedestrian Dead Reckoning) using smartphone consist with step detection, step length estimation and heading estimation. However, these algorithms have errors caused by various reasons such as step length error at uncertain user, magnetic disturbance in indoor situation and unstable position of smartphone. To increase accuracy of the PDR, Wi-Fi fusion or map matching method has been proposed. However, previous methods could not solve fault matching or creating map in hall area. Especially in hall, pedestrian could make various trajectories that accurate map structures are required. For solving the structure of map database in hall problem and accurate link selection, we propose a Virtual Link (VL) algorithm with a Virtual Track (VT). Furthermore, an Extended Kalman Filter (EKF) is used for estimating pedestrian position and IMU sensor errors. With map information, step length estimation error, heading error at pedestrian dead reckoning and some IMU sensor errors are estimated. Real world experiments are conducted at building, and it shows less than 3m of CEP (Circular Error Probability) after 200m walk.