Mitigating Large Errors in WiFi-Based Indoor Localization for Smartphones
Chenshu Wu,Zheng Yang,Zimu Zhou,Yunhao Liu,Mingyan Liu
DOI: https://doi.org/10.1109/tvt.2016.2630713
IF: 6.8
2017-07-01
IEEE Transactions on Vehicular Technology
Abstract:Although WiFi fingerprint-based indoor localization is attractive, its accuracy remains a primary challenge, especially in mobile environments. Existing approaches either appeal to physical layer information or rely on extra wireless signals for high accuracy. In this paper, we revisit the received signal strength (RSS) fingerprint-based localization scheme and reveal crucial observations that act as the root causes of localization errors, yet are surprisingly overlooked or not adequately addressed in previous works. Specifically, we recognize access points (APs) diverse discrimination for fingerprinting a specific location, observe the RSS inconsistency caused by signal fluctuations and human body blockages, and uncover the transitional fingerprint problem on commodity smartphones. Inspired by these insights, we devise a discrimination factor to quantify different APs discrimination, incorporate robust regression to tolerate outlier measurements, and reassemble different normal fingerprints to cope with transitional fingerprints. Integrating these techniques in a unified system, we propose DorFin, i.e., a novel scheme of fingerprint generation, representation, and matching, which yields remarkable accuracy without incurring extra cost. Extensive experiments in three campus buildings demonstrate that DorFin achieves a mean error of 2.5 m and, more importantly, decreases the 95th percentile error under 6.2 m, both significantly outperforming existing approaches.
telecommunications,engineering, electrical & electronic,transportation science & technology