Bayesian localization and tracking with environmental and array-element uncertainties.

Abstract

This paper considers matched-field source localization and tracking when environmental parameters and/or array-element positions are not well known. A Bayesian formulation is applied in which source, array, and environmental parameters are considered unknown random variables constrained by noisy acoustic data and by prior information on parameter values (e.g., physical limits for environmental properties and element positions) and on interparameter relationships (e.g., limits on source speed and interelement spacing). The goal then is to extract source information from the posterior probability density (PPD). One approach is based on maximizing the PPD over all parameters to obtain optimal source locations. A key to solving this problem efficiently is that the VITERBI algorithm is applied to compute the highest-probability source track for each environment/array realization: this provides the optimal track, while requiring the optimization is applied only over the nuisance parameters. A second approach involves integrating the PPD over unknown environmental and array parameters to represent source-location information as a series of joint marginal probability surfaces over range and depth. Given the strong nonlinearity of this problem, marginal PPDs are computed numerically using efficient Markov-chain Monte Carlo importance sampling methods. The approaches are illustrated with examples based on simulated and measured acoustic data.