Toward Accurate Indoor Positioning: An RSS-Based Fusion of UWB and Machine-Learning-Enhanced WiFi

A wide variety of sensors and devices are used in indoor positioning scenarios to improve localization accuracy and overcome harsh radio propagation conditions. The availability of these individual sensors suggests the idea of sensor fusion to achieve a more accurate solution. This work aims to address, with the goal of improving localization accuracy, the fusion of two conventional candidates for indoor positioning scenarios: Ultra Wide Band (UWB) and Wireless Fidelity (WiFi). The proposed method consists of a Machine Learning (ML)-based enhancement of WiFi measurements, environment observation, and sensor fusion. In particular, the proposed algorithm takes advantage of Received Signal Strength (RSS) values to fuse range measurements utilizing a Gaussian Process (GP). The range values are calculated using the WiFi Round Trip Time (RTT) and UWB Two Way Ranging (TWR) methods. To evaluate the performance of the proposed method, trilateration is used for positioning. Furthermore, empirical range measurements are obtained to investigate and validate the proposed approach. The results prove that UWB and WiFi, working together, can compensate for each other's limitations and, consequently, provide a more accurate position solution.