Gain Without Pain: Accurate WiFi-based Localization using Fingerprint Spatial Gradient

Among numerous indoor localization systems proposed during the past decades, WiFi €ngerprint-based localization has been one of the most aŠractive solutions, which is known to be free of extra infrastructure and specialized hardware. However, current WiFi €ngerprinting su‚ers from a pivotal problem of RSS ƒuctuations caused by unpredictable environmental dynamics.Œe RSS variations lead to severe spatial ambiguity and temporal instability in RSS €ngerprinting, both impairing the location accuracy. To overcome such drawbacks, we propose €ngerprint spatial gradient (FSG), a more stable and distinctive form than RSS €ngerprints, which exploits the spatial relationships among the RSS €ngerprints of multiple neighbouring locations.As a spatially relative form, FSG is more resistant to RSS uncertainties. Based on the concept of FSG, we design novel algorithms to construct FSG on top of a general RSS €ngerprint database and then propose e‚ective FSG matching methods for location estimation. Unlike previous works, the resulting system, named ViVi, yields performance gain without the pains of introducing extra information or additional service restrictions or assuming impractical RSS models. Extensive experiments in di‚erent buildings demonstrate that ViVi achieves great performance, outperforming the best among four comparative start-of-the-art approaches by 29% in mean accuracy and 19% in 95th percentile accuracy and outweighing the worst one by 39% and 24% respectively. We envision FSG as a promising supplement and alternative to existing RSS €ngerprinting for future WiFi localization.