A calibration-free indoor localization system using pseudo-distances in WLAN environments

To avoid substantial manual calibration effort, some WiFi-based indoor localization methods attempt to achieve calibration-free feature. However, many of these methods have additional requirements which limit their practical applicabilities, e.g. lateration-based methods require the target locations to be covered by at least three APs, some methods require specific relationships between the training locations and APs. In this paper, we propose an indoor localization system named SimpLoc, which does not require any manual calibration effort nor any strict condition of use. The system applies the multi-dimensional scaling technique by converting the crowd-sourced RSS data to pseudo pairwise distances between the unknown locations. In a 36 × 23m2 area, SimpLoc achieves a mean distance error of 1.93m. Experimental results show that the error can still be maintained at 2.14m when only 10 percent of the training data are used.