A Practical Indoor Positioning System Based on Collaborative PDR and Wi-Fi Fingerprinting

An Indoor Positioning System (IPS) is vital in pro-viding location-based service (LBS) indoors. The shortcomings of the Global Positioning System (GPS) indoors have led to the development of many alternative solutions for IPS. Wi-Fi finger-printing is a promising solution due to its ability to accurately predict indoor locations. Moreover, the widespread deployment of Wi-Fi infrastructures makes the Wi-Fi fingerprinting-based IPS technique readily applicable. However, this technique re-quires high overhead due to the need to define many reference points (RPs) within the subject area. Multiple Wi-Fi received signal strength (RSS) samples shall be collected from each RP, introducing extensive manual work. This fact makes the Wi-Fi fingerprinting-based IPS is less practical. This issue is addressed by employing pedestrian dead reckoning (PDR) as part of the proposed technique. Realizing a practical IPS based on collaborative use of PDR and Wi-Fi fingerprinting. In the proposed method, most of the time, the location prediction will be based on PDR. When the step count reaches a threshold, a switch from PDR to Wi-Fi fingerprinting is made to decide the user's location. Then, the system resets the step count and uses the PDR again. This approach significantly reduces the total number of RPs needed by the system. The numerical results show that the proposed technique has 99% lower overhead than the conventional Wi-Fi fingerprinting technique and achieves 55.6% better prediction accuracy than the standalone PDR technique. The results justify the benefits of the proposed IPS based on collaborative PDR and Wi-Fi fingerprinting.