AmbiLoc: A year-long dataset of FM, TV and GSM fingerprints for ambient indoor localization

Ambient indoor localization - an approach that leverages ambient radio signals - has been previously shown to provide promising positioning performance using the globally available infrastructure of FM, TV and cellular stations. However, the need for specialized equipment and laborious data collection constitute a high entry barrier for follow-up studies.

This paper presents AmbiLoc - a dataset of radio signals for ambient indoor localization research. The dataset has been systematically collected in multiple testbeds, including large-scale and multi-floor buildings, over the course of one year. Due to the use of a software-defined radio receiver, raw signal samples in AmbiLoc allow extraction of arbitrary fingerprinting features. The first edition of AmbiLoc, introduced in this paper, includes received signals strength (RSS) fingerprints of FM, TV and GSM signals, along with the relevant metadata (such as weather conditions). The dataset is available online at AmbiLoc.org.

As the first public dataset of ambient localization signals, AmbiLoc provides an easy entry and a common reference for researchers exploring novel indoor localization methods.