Communications and Networking. 11th EAI international Conference, ChinaCom 2016 Chongqing, China, September 24-26, 2016, Proceedings, Part II

Research Article

Crowdsourcing-Based Indoor Propagation Model Localization Using Wi-Fi

Download
183 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-66628-0_56,
        author={Yongliang Sun and Jian Wang and Wenfeng Li and Rui Jiang and Naitong Zhang},
        title={Crowdsourcing-Based Indoor Propagation Model Localization Using Wi-Fi},
        proceedings={Communications and Networking. 11th EAI international Conference, ChinaCom 2016 Chongqing, China, September 24-26, 2016, Proceedings, Part II},
        proceedings_a={CHINACOM},
        year={2017},
        month={10},
        keywords={Crowdsourcing Wi-Fi localization Outlier processing Propagation model Coordinate correction},
        doi={10.1007/978-3-319-66628-0_56}
    }
    
  • Yongliang Sun
    Jian Wang
    Wenfeng Li
    Rui Jiang
    Naitong Zhang
    Year: 2017
    Crowdsourcing-Based Indoor Propagation Model Localization Using Wi-Fi
    CHINACOM
    Springer
    DOI: 10.1007/978-3-319-66628-0_56
Yongliang Sun,*, Jian Wang1, Wenfeng Li1, Rui Jiang2, Naitong Zhang
  • 1: Nanjing University
  • 2: Nanjing Tech University
*Contact email: syl_peter@163.com

Abstract

To save labor and time costs, crowdsourcing has been used to collect received signal strength (RSS) for building radio-map of Wi-Fi fingerprinting localization with common users’ mobile devices. However, usually a great number of crowdsourcing data should be collected to calculate a satisfactory localization result. Therefore, we proposed a crowdsourcing-based indoor propagation model (PM) localization system in this paper. Our system only needs to collect crowdsourcing data at a few locations called crowdsourcing points, which can be easily finished in a short time. The system first eliminates RSS outliers in crowdsourcing data and then optimizes PM parameters using the processed data. Furthermore, the processed data is also used to estimate a distance between a user and the nearest crowdsourcing point for coordinate correction. Experimental results show that our system is able to achieve a comparable performance and the mean error of PM localization method is reduced from 7.12 m to 3.78 m.