Efficient Wi-Fi Fingerprint Crowdsourcing for Indoor Localization
- 9 June 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 22 (6) , 5055-5062
Abstract
Wi-Fi Received Signal Strength (RSS, fingerprints) based indoor localization is promising and widely investigated with the pervasive deployment of Wi-Fi Access Points. However, the process to collect RSS, also known as site survey, is labor-intensive. Thus, we propose and demonstrate an efficient fingerprint crowdsourcing method in this paper. Specifically, RSS measurements are obtained and annotated with location tags while a participant is walking along a chosen path with a smartphone at hand. In the localization stage, we adopt the Gaussian Process based solution and propose a novel mean function selection method. Extensive experiments show that the path-based site survey can achieve a comparable localization performance to the point-based site survey, but takes less survey time. We find that fingerprints collected while walking are more suitable for localizing moving pedestrians. In addition, due to the sparsity of fingerprints collected through crowdsourcing, the proposed mean function selection strategy is advantageous and can reduce localization errors significantly compared to a baseline solution.Keywords
This publication has 25 references indexed in Scilit:
- TuRF: Fast data collection for fingerprint-based indoor localization2017
- Fusing Similarity-Based Sequence and Dead Reckoning for Indoor Positioning Without TrainingIEEE Sensors Journal, 2017
- Recent Advances in Indoor Localization: A Survey on Theoretical Approaches and ApplicationsIEEE Communications Surveys & Tutorials, 2016
- Gaussian Process Regression for Fingerprinting based LocalizationAd Hoc Networks, 2016
- WicLoc: An indoor localization system based on WiFi fingerprints and crowdsourcing2015
- FreeLoc: Calibration-free crowdsourced indoor localization2013
- Normative spatiotemporal gait parameters in older adultsGait & Posture, 2011
- Gaussian Processes for Signal Strength-Based Location Estimation2006
- Properties of indoor received signal strength for WLAN location fingerprinting2004
- RADAR: an in-building RF-based user location and tracking system2002