FreeLoc: Calibration-Free Crowdsourced Indoor Localization

Many indoor localization techniques that rely on RF signals from wireless Access Points have been proposed in the last decade. In recent years, research on crowdsourced (also known as "Organic") Wi-Fi fingerprint positioning systems has been attracting much attention. This participatory approach introduces new challenges that no previously proposed techniques have taken into account. This paper proposes "FreeLoc", an efficient localization method addressing three major technical issues posed in crowdsourcing based systems. Our novel solution facilitates 1) extracting accurate fingerprint values from short RSS measurement times 2) calibration-free positioning across different devices and 3) maintaining a single fingerprint for each location in a radio map, irrespective of any number of uploaded data sets for a given location. Through experiments using four different smartphones, we evaluate our new indoor positioning method. The experimental results confirm that the proposed scheme provides consistent localization accuracy in an environment where the device heterogeneity and the multiple surveyor problems exist.