A novel fusion methodology for indoor positioning in IoT-based mobile applications

Abstract

We present a novel fusion-based Indoor Positioning System (IPS) that fuses the sensory data of three popular smartphone sensor technologies i.e. WiFi, Bluetooth Low Energy (BLE) and Pedestrian Dead Reckoning (PDR) to achieve a superior indoor localization accuracy. The proposed system can be used for a broad base of Internet of Things (IoT) related applications that involve mobility and localization. In our approach, we first compare the indoor positioning accuracy of fingerprinting schemes using WiFi and BLE through their Received Signal Strength (RSS) values. In this comparison, we use these two technologies both separately and in combination, and show that the combined fingerprints of WiFi/BLE sensors give better results than individual fingerprints. We then fuse the combined fingerprinting results of WiFi/BLE sensors with the PDR using two methods: (i) segment-based fusion and (ii) the Kalman Filter (KF). In this experiment, we show that the fusion of WiFi, BLE and PDR using the KF gives 23% improvement in average error over WiFi/PDR fusion and 25% improvement in average error over BLE/PDR fusion while all these different techniques are superior than simple PDR-based localization. We, therefore, conclude that WiFi/BLE/PDR fusion is a promising approach to be considered for future IPSs being designed for personalized IoT-based smart systems.