LocaRDS: A Localization Reference Data Set

TL;DR

LocaRDS is an extensive open dataset of real-world wireless signals designed to facilitate the evaluation and comparison of localization techniques, especially for aircraft positioning in crowdsourced sensor networks.

Contribution

This work introduces LocaRDS, a large-scale, real-world dataset for localization research, along with a reference implementation and evaluation framework.

Findings

LocaRDS contains over 222 million measurements from 50 million transmissions.

The dataset enables direct comparison of different localization methods.

Demonstrated effectiveness in aircraft localization using crowdsourced sensors.

Abstract

The use of wireless signals for purposes of localization enables a host of applications relating to the determination and verification of the positions of network participants, ranging from radar to satellite navigation. Consequently, it has been a longstanding interest of theoretical and practical research in mobile networks and many solutions have been proposed in the scientific literature. However, it is hard to assess the performance of these in the real world and, more severely, to compare their advantages and disadvantages in a controlled scientific manner. With this work, we attempt to improve the current state of the art in localization research and put it on a solid scientific grounding for the future. Concretely, we develop LocaRDS, an open reference dataset of real-world crowdsourced flight data featuring more than 222 million measurements from over 50 million transmissions recorded by 323 sensors. We show how LocaRDS can be used to test, analyze and directly compare different localization techniques and further demonstrate its effectiveness by examining the open question of the aircraft localization problem in crowdsourced sensor networks. Finally, we provide a working reference implementation for the aircraft localization problem and a discussion of possible metrics for use with LocaRDS.