Bluetooth-Based WKNNPF and WKNNEKF Indoor Positioning Algorithm

Abstract

Indoor Positioning System (IPS) in generally perform as a network of devices that always located the objects or people inside a building wirelessly. An IPS has direction relies nearby anchors and also can be entirely local to your smartphone. With the rapid growth and sharp increase in Indoor Positioning System (IPS) demand in the world, there are a lot of researchers trying to invent new algorithm to develop IPS. This paper proposed the Bluetooth-Base Indoor Positioning Algorithm. The RF characteristics such as RSSI and WLAN RSSI fingerprinting system normally formed by two phases, fist is offline phase and second is online phase. Fingerprinting system handling both off-line and online data and estimate the user’s location. Our algorithm design is a collection of Weighted K-Nearest Neighbors (WKNN) and Filtering algorithms by KALMAN Filter. Finally, to avoid the problems of IPS and get a better accurate we proposed two algorithms: Weighted K-Nearest Neighbors Particle Filter (WKNNPF) and Weighted K-Nearest Neighbors Extended Kalman Filter (WKNNEKF) compare to KNN and WKNN result. After comparing we found that the result of WKNNPF and WKNNEKF is better result than KNN and WKNN. The Probability in 3M of WKNN is about 79%, WKNNEKF is about 89%, and WKNNPF is about 95.1%. Among one of the proposed algorithms WKNNPF is better than WKNNEKF on accuracy 1.7-2 meters with 42.2m/s response time.