Acrux: Indoor Localization Without Strings - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Acrux: Indoor Localization Without Strings

Résumé

We present Acrux, the first indoor localization system to achieve meter level accuracy while relying exclusively on a single fix and the sensors commonly found in off-the-shelf smartphones. Acrux uses dead-reckoning, the approach that gives probably the best chance at a completely autonomous indoor localization system. Unfortunately, it has not been mastered on smartphones beyond a few dozen meters due to its inherent integration drift. As a result, all dead-reckoning based solutions in literature require periodic recalibration using input from outside -- attaching strings preventing indoor localization from becoming mainstream. While it is virtually impossible to completely eliminate integration drift, Acrux is the first solution to succeed in dead-reckoning with meter level accuracy for several hundred meters, enough to relax the requirement for periodic recalibration in most indoor scenarios. To accomplish this, Acrux replaces step-counting, the standard approach for measuring distance using sensors, with an approach that measures the speed of locomotion. Although a straightforward accurate estimation of motion speed using the erroneous sensors found on smartphones is infeasible, Acrux combines a novel approach with measurement based analysis to achieve that. Leveraging its excellent dead-reckoning capability, Acrux is shown to provide indoor localization with median error between SI0.7 meter and SI1.2 meter and 98% percentile error of SI3 meter in a dozen of scenarios in 4 different buildings -- without any recalibration.
Fichier non déposé

Dates et versions

hal-03116298 , version 1 (20-01-2021)

Identifiants

Citer

Jean-Gabriel Krieg, Gentian Jakllari, Hadrien Toma, André-Luc Beylot. Acrux: Indoor Localization Without Strings. 20th ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM 2017), ACM; IEEE, Nov 2017, Miami, United States. pp.187--196, ⟨10.1145/3127540.3127545⟩. ⟨hal-03116298⟩
29 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More