Time-delay estimation in underwater acoustics: A global optimization approach

Abstract

Arrival times and amplitudes of individual ray paths provide significant information for source localization and geoacoustic inversion in underwater acoustics. Considered in a statistical framework, estimates for such quantities can be obtained via maximum likelihood or maximum a posteriori methods. These approaches, however, require function maximization on a multidimensional space, demanding extensive computations depending on the number of distinct paths. Local search methods have been proposed as computationally low-cost alternatives; such techniques, however, often get trapped in local maxima. In this work, a global optimization approach is proposed for the maximization of the posterior probability distribution of arrival times and amplitudes. To avoid an exhaustive search, an estimate of the posterior distribution is obtained employing Gibbs sampling; it is, subsequently, straightforward to obtain values for time delays and amplitudes for which the distribution attains a maximum. Simulations show that the new method provides excellent estimates that are comparable to those of the analytical maximum a posteriori approach. Real data results from the Haro Strait Experiment further validate the method. [Work supported by ONR.]