Continuous Indoor Positioning Fusing WiFi, Smartphone Sensors and Landmarks

Area of Science:

• Computer Science
• Electrical Engineering
• Geomatics Engineering

Background:

• Accurate indoor positioning remains a challenge, with existing methods like WiFi positioning and pedestrian dead reckoning (PDR) having limitations.
• Fusion approaches offer potential but often rely on empirical noise parameter settings and struggle with error accumulation.

Purpose of the Study:

• To propose a novel fusion approach for indoor positioning by integrating WiFi positioning, PDR, and landmarks.
• To enhance the accuracy and reduce the computational cost of WiFi positioning and PDR through adaptive noise modeling and landmark recalibration.

Main Methods:

• A novel fusion approach based on an extended Kalman filter (EKF) integrating WiFi positioning, PDR, and landmarks.
• Adaptive measurement noise statistics for WiFi positioning using kernel density estimation.
• Trusted area definition and WiFi signal outlier detection to improve WiFi positioning accuracy and reduce computational load.
• Integration of gyroscope, accelerometer, and magnetometer for PDR heading determination using an EKF.
• Recalibration of PDR positioning and heading estimations using indoor landmarks to mitigate error accumulation.

Main Results:

• The proposed fusion approach demonstrated substantial positioning accuracy improvement in a realistic indoor environment.
• Significant enhancements were observed compared to individual positioning approaches, including PDR and WiFi positioning.

Conclusions:

• The novel fusion approach effectively leverages the complementary strengths of WiFi positioning, PDR, and landmarks.
• The adaptive noise modeling and landmark recalibration contribute to improved accuracy and robustness for continuous indoor positioning.