Online Collaborative Localization

Can fingerprinting-based localization benefit from simultaneous observations from different users? This article provides an affirmative answer to this question. Existing fingerprinting schemes work in a kind of stand-alone mode, as they localize a user merely based on her own observations about surrounding environments. However, we argue the existence of scenarios where multiple users request for localization services simultaneously. If they are not far apart, their observations are actually distributed versions of the same environment, which can be exploited to calibrate underlying localization systems. In this article, we propose an online collaborative localization (OCLoc) scheme to boost received signal strength (RSS) fingerprinting in scenarios with multiple requesting users. The basic idea of OCLoc is based on the fact of distance attenuated radio propagations. For two users, their received signals from the same access point (AP) can be used to infer who is closer to this AP. We exploit such information for online collaboration and propose a series of operations to calibrate reference points (RPs) from distributed versions of radio attenuation observations. By integrating user calibration and fingerprint distance, we define a new ranking measure for reselecting and weighting candidate RPs to output the final location. Experiments on field measurements validate the effectiveness of such online collaborations in terms of greatly reduced localization errors.

Automated summary

This article introduces an online collaborative localization (OCLoc) scheme that can enhance fingerprinting-based localization accuracy and reduce errors by leveraging simultaneous observations from multiple users, using received signal strength for localization, and optimizing the localization system through collaborative calibration.