Sequential bayesian filtering for spatial arrival time estimation
Department of Mathematical Sciences
Doctor of Philosophy
Michalopoulou, Eliza Zoi-Heleni
Bhattacharjee, Manish Chandra
Dhar, Sunil Kumar
Luke, Jonathan H.C.
Arrival time estimation
Monte Carlo Markov chain
Locating and tracking a source in an ocean environment as well as estimating environmental parameters of a sound propagation medium is of utmost importance in underwater acoustics. Matched field processing is often the method of choice for the estimation of such parameters. This approach, based on full field calculations, is computationally intensive and sensitive to assumptions on the structure of the environment. As an alternative, methods that use only select features of the acoustic field for source localization and environmental inversion have been proposed. The focus here is on inversion using arrival times of identified paths within recorded time-series. After a short study of a linearization techniques employing such features and numerical issues on their implementation, we turn our attention to the need for accurate extraction of arrival times for accurate estimation. We develop a particle filtering approach that treats arrival times as "targets", dynamically modeling their "location" at arrays of spatially separated receivers. Using Monte Carlo simulations, we perform an evaluation of our method and compare it to conventional Maximum Likelihood (ML) estimation. The comparison demonstrates an advantage in using the proposed approach, which can be employed as a pre-inversion tool for minimization and quantification of uncertainty in arrival time estimation.
njit-etd2011-097 (96 pages ~ 699 KB pdf)
Please complete this Feedback Form to inform us about your experience using this website. It will assist us in better serving your information needs in the future. Thank You!
Created March 9, 2012