Abstract

Abstract. Indoor localization is an open problem. In this paper, we explore the possibility of fingerprinting rooms based on the intensity of light incident on a light sensor worn by the user, under static lighting conditions. We present three algorithms for fingerprinting, namely Bayesian, Range-max and Spatial and present experimental results for the first two. Bayesian performs better, achieving an accuracy (i.e success probability) of more than 90 % when the light sensor is worn on top of a hat, and more than 80 % when the light sensor is worn as a pendant. This approach does not require any infrastructure; existing light sources in rooms are used for localization. 1

Citations