On Hybrid Localization in Half-Open Areas

Given the enormous demands for location based applications and service, localization technologies, which estimate accurate locations of the mobile users, have been developed rapidly outdoors (e.g., GPS) and indoors (e.g., WiFi fingerprinting). However, in half-open areas where existing studies are rarely conducted, either the outdoor or indoor localization technique fails to achieve a satisfactory performance due to the short of signals. In this paper, we propose a hybrid localization method tailored for the use in half-open areas. Specifically, we present two individual probabilistic localization techniques, which are extended from conventional GPS and WiFi fingerprinting. Moreover, our hybrid approach combines the locations estimated by the two techniques, and modifies the estimation by an enhanced Kalman filter leveraging the user movement capacity as well as the location probability. The experiment results show our hybrid localization method outperforms the conventional GPS and WiFi fingerprinting based methods in half-open areas.