Autonomous 3D Indoor Localization Based on Crowdsourced Wi-Fi Fingerprinting and MEMS Sensors

Location-based services have become more and more important with the development of the Internet of Things (IoT) technology in recent years. Global Navigation Satellite System (GNSS) is widely used for positioning outdoors while it is still challenging to realize accurate and universal 3D indoor localization in complex indoor environments. Crowdsourcing-based positioning method is proposed aiming at autonomously constructing the navigation database based on the pedestrians' daily-life data. This paper proposes an autonomous 3D indoor localization algorithm using the combination of crowdsourced Wi-Fi fingerprinting and Micro-Electro-Mechanical System sensors (3D-CSWS). An enhanced complementary filter is applied to provide accurate attitude information by integrating multi-sensor data with the detection of external acceleration and quasi-static magnetic field. In addition, the gradient descent (GD) algorithm is proposed to optimize the forward pedestrian dead reckoning and the optimized trajectories are weighted fused to construct the final navigation database after quality evaluation. In the on-line phase, the adaptive particle filter is used to integrate the results of Wi-Fi fingerprinting and multiple sensors to provide accurate and concrete 3D indoor localization performance. The real-world experimental results demonstrate that the proposed 3D-CSWS is proved to achieve autonomous and precise 3D indoor localization performance among complex indoor environments.

Automated summary

Location-based services have become increasingly important with the development of IoT technology. While GNSS is widely used for outdoor positioning, achieving accurate and universal 3D indoor localization remains challenging. This paper proposes a crowdsourcing-based positioning method using crowdsourced WiFi fingerprinting and MEMS sensors to achieve autonomous and precise 3D indoor localization. The proposed algorithm combines multi-sensor data and applies an enhanced complementary filter and gradient descent algorithm for accurate attitude information and optimized pedestrian dead reckoning. The results show that the proposed method achieves autonomous and precise 3D indoor localization performance in complex indoor environments.