Original Article Published on phys.org
Accurate indoor localization has the potential to transform the way people navigate indoors in a similar way that GPS transformed the way people navigate outdoors. Over the last 15 years, several indoor localization technologies have been proposed and experimented by both academia and industry. A major problem is that the proposed solutions are evaluated in different conditions, which makes hard any comparison of their performance.
The Microsoft Indoor Localization Competition aims to bring together real-time or near real-time indoor location technologies and compare their performance in the same space, solving the problem of absence of a common environment to compare indoor localization technologies. The competition, now at the third edition, took place in Vienna, Austria, on Sunday, April 10th, and Monday April 11th, 2016. The evaluation environment was very challenging, spanning two floors with metal, with a total evaluation area of approximately 500 m2.
All submissions were classified into two categories: 3D and 2D localization. All teams in the 3D localization category were required to deploy customized hardware, and provide 3D locations during the evaluation. All teams in the 2D localization category did not require customized hardware (just Commercial Off-The-Shelf devices) and reported 2D locations during evaluation. During the first day the teams were given 7 hours to setup and calibrate their systems and during the second day each team provided the coordinates of 15 test points with known ground truth. At the end, the evaluation metric was the average localization error across these 15 test points.
A team of researchers from the Pervasive Wireless System group led by Dr. Domenico Giustiniano from IMDEA Networks Institute competed with a solution in the 2D category. At the end of the competition, the team of researchers from IMDEA Networks achieved an average error of 3.17 meters, which resulted in ranking 5th out of ten teams in the final. (Out of the fifteen teams that were initially represented at the competition in the 2D category, only ten made it to the final ranking).