The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLVI-3/W1-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-3/W1-2022, 177–184, 2022
https://doi.org/10.5194/isprs-archives-XLVI-3-W1-2022-177-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVI-3/W1-2022, 177–184, 2022
https://doi.org/10.5194/isprs-archives-XLVI-3-W1-2022-177-2022
 
22 Apr 2022
22 Apr 2022

CROWDSOURCED WIFI FINGERPRINT LOCALIZATION IN URBAN CANYON

Y. Su, L. Chen, and X. Liu Y. Su et al.
  • State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China

Keywords: Crowdsourcing, Localization, WiFi, Fingerprint, Urban Canyon, RSSI (Received Signal Strength Indication), KWNN (K Weighted Nearest Neighbor)

Abstract. Although Global Navigation Satellite System (GNSS) has achieved success in outdoor localization, it does not often work well in urban canyon, which is due to the weak signals and the loss of satellites. WiFi technology is widely used at present, and the crowdsourced WiFi data has the advantages of rich sources and low cost. Therefore, utilizing the crowdsourced WiFi data for localization may effectively improve the deficiency of GNSS in the urban canyon. In this paper, we propose a novel method of crowdsourced WiFi fingerprint localization in urban canyon. Considering that the crowdsourced data is noisy, discontinuous and unstable, we carry out pre-processes for data refining, and grid-based statistical method for noise smoothing. Then in order to quickly locate the terminals in large-scale area, the AP coverage intersection method is proposed, in which the coverage range, centers and density of all APs are inferred, and the personal hotspots as well as the mobile APs are removed. To further enhance the positioning accuracy, the fine localization is carried out, which is based on the iterative KWNN algorithm. Extensive field tests are carried out in a typical urban canyon, results show that the average positioning error of our method is 16.82 m, which shows the effectiveness of the proposed method for crowdsourced positioning in urban canyon.