This paper suggests indoor positioning system using sensor fusion algorithm based on the Extended Kalman Filter (EKF). The filter of this system is comprised of Wi-Fi Positioning System (WPS) and Pedestrian Dead Reckoning (PDR). WPS estimates a rough position and PDR makes this position become smoother by using this filter. In this fusion system, the change of user’s heading (user’s rotation) is expressed in the local coordinate system to make it possible to obtain the change of user’s rotation more accurately. Therefore, the system dynamic model in the EKF can be designed more precisely. This paper also proposes the method to use pseudo velocity and WPS position as measurements. ‘Pseudo’ means that the velocity is not real quantity from the sensor. The combination of PDR speed and direction which is obtained by map direction, the magnetometer and rotation angle from the gyroscope. The map information can be used when the map data is available. When user is on the open space or on the place where the map does not exist, the output of sensors and WPS position are used to estimate the direction. This is determined by the environment. This augmented measurement can improve the estimation performance by correcting the state variables.

Keywords: indoor positioning, extended Kalman filter, hybrid positioning, pedestrian dead reckoning