A hybrid indoor positioning algorithm based on WiFi fingerprinting and pedestrian dead reckoning

WiFi fingerprinting method is an attractive indoor positioning method due to widely deployed WiFi access points (APs) and easily measured received signal strength (RSS). However, WiFi fingerprinting positioning results are unstable because of the fluctuation of RSS. Besides, pedestrian dead reckoning (PDR) method relying on inertial sensors has been widely used in real-time tracking. Since PDR has accumulated errors in long distances tracking, this paper proposes a hybrid algorithm that integrates PDR approach with WiFi fingerprinting approach to further improve positioning accuracy. There are two key points in our algorithm. The first is utilizing dynamic subarea to restrict the searching region of WiFi fingerprinting method. The second is determining particular weights to fuse the positioning results of the above two approaches according to the distances between the current positioning results and the previous hybrid location. Further, we improve our hybrid algorithm which is based on the adjacent estimated positions of WiFi fingerprinting method. Experiment results showed that the average errors of our hybrid algorithm and improved hybrid algorithm were 2.22m and 1.64m respectively, which were reduced by 42% and 57% compared with the pure PDR method. Therefore, the proposed algorithms can provide stable and high positioning accuracy in real environment.