Bayesian source localization with uncertain Green’s function

International audienceThe localization of an acoustic source in the oceanicwaveguide is a difficult task because the oceanic environmentis often poorly known. Uncertainty in the environment resultsin uncertainty in the source position and poor localizationresults. Hence, localization methods dealing with environmentaluncertainty are required. In this paper, a Bayesian approach tosource localization is introduced in order to improve robustnessand obtain quantitative measures of localization uncertainty.The Green’s function of the waveguide is considered as anuncertain random variable whose probability density accountsfor environmental uncertainty. The uncertain distribution overrange and depth is then obtained through the integration of theposterior probability density (PPD) over the Green’s functionprobability density. An efficient integration technique makes thewhole localization process computationally efficient. Some resultsare presented for a simple uncertain Green’s function model toshow the ability of the proposed method to give reliable PPDs