A segment-based fusion algorithm of WiFi fingerprinting and pedestrian dead reckoning

Abstract

Wireless indoor positioning systems have been extensively investigated and have enabled a great number of pervasive applications. Traditional indoor localization systems mostly rely on received signal strength (RSS) fingerprints to distinguish locations, which may fail to achieve rich performance in a real environment. For example, in the complicated indoor environments, RSSs fluctuate severely, so that it causes inaccurate positioning results. Pedestrian dead reckoning (PDR) systems, which leverage inertial sensors to obtain the user's motion information and estimate the user's current location, have also been widely adopted for real-time indoor pedestrian location tracking. In light of the recursive characteristic of PDR positioning, if the parameters of the sensors are estimated erroneously, it may bring accumulated tracking errors. In this paper, a hybrid algorithm which integrates the fingerprinting positioning and the PDR positioning is proposed to attain a robust and continuous positioning trajectory. Both the WiFi fingerprinting positioning system and the PDR system are expected to be complementary to each other. In our environment, experimental results indicate that the mean error of our proposed hybrid algorithm is 2.4255m which is reduced by 36.3% and 54.3% compared with PDR approach and WiFi fingerprinting approach alone, respectively.