An Energy-Efficient Crowd-Sourcing-Based Indoor Automatic Localization System

The development of crowd sourcing and inertial sensors built into mobile devices provides a new opportunity for reducing the consumption of fingerprint-based indoor location. However, this development also brings a variety of new theoretical and practical problems. In this paper, we present an automatic room-level localization approach to resolve three key problems: the collection of fingerprints, determination of fingerprints’ location information, and energy consumption. Unlike previous works, we propose an active sampling algorithm to filter reliable and useful information from numerous users’ inputs. We also design a virtual room generation mechanism and floor plan mapping algorithm to construct a reliable room fingerprint database without manual annotations. Experimental results demonstrate that compared with the traditional algorithms the new approach supports room-level localization with higher than 95% accuracy and reduces energy consumption by more than 40%.