Ocean Acoustic Signal Processing—A Bayesian Approach
James V. Candy, IEEE OES distinguished lecturer
The first IEEE OES distinguished lecture for this year was delivered by James V. Candy in collaboration with the Depart of Physics, University of New Orleans (UNO), on 3 March 2021. Title of the Talk: Ocean Acoustic Signal Processing – A Bayesian Approach. The lecture was given online, and is now available on YouTube: www.youtube.com/channel/UC6wjVnDY2-BmzdS8LzxrdHQ
Please view and enjoy this distinguished lecture.
James V. Candy
James V. Candy is the Chief Scientist for Engineering, a Distinguished Member of the Technical Staff and founder/former Director of the Center for Advanced Signal & Image Sciences (CASIS) at the Lawrence Livermore National Laboratory. He is also an Adjunct Professor, Electrical & Computer Engineering Department, University of California, Santa Barbara.
He is a Fellow of both the IEEE and the Acoustical Society of America (ASA). A Distinguished Alumnus of the University of Cincinnati, Dr. Candy is a recipient of many awards such as the IEEE OES Distinguished Technical Achievement Award, and Interdisciplinary Helmholtz-Rayleigh Silver Medal in Signal Processing/Underwater Acoustics by the Acoustical Society of America. He has published over 225 journal articles, book chapters, and technical reports as well as written six texts in signal processing. He has presented a variety of short courses and tutorials sponsored by the IEEE and ASA in Applied Signal Processing, Spectral Estimation, Advanced Digital Signal Processing, Applied Model-Based Signal Processing, Applied Acoustical Signal Processing, Model-Based Ocean Acoustic Signal Processing and most recently Bayesian Signal Processing for IEEE Oceanic Engineering Society/ASA. His research interests include Bayesian learning, estimation, identification, spatial estimation, signal and image processing, array signal processing, nonlinear signal processing, tomography, sonar/radar processing and biomedical applications. For a full CV please visit this page:
The application of Bayesian methods to complex ocean acoustic processing problems, especially in shallow water, has evolved from well-known probability distributions like Gaussian leading to model- based, Kalman filtering solutions to nonparametric representations driven by the uncertain ocean environment leading to sequential Monte Carlo or equivalently particle filtering solutions. In this lecture, an overview of particle filtering methods coupled to a shallow ocean modal tracking application motivated by the nonlinear nature of underlying ocean acoustic phenomenology is presented. Beginning with a brief overview of Bayesian inference leading to sequential processors, the Bayesian paradigm is established. Simulation-based methods using sampling theory and sequential Monte Carlo realizations are discussed. Here the usual limitations of nonlinear approximations and non-gaussian processes prevalent in classical algorithms (e.g. Kalman filters) are no longer a restriction to perform Bayesian processing. It is shown how the underlying state variables are easily assimilated into this sequential Bayesian construct. With this in mind, the idea of a particle filter, which is a discrete nonparametric representation of a probability distribution, is developed and shown how it can be implemented using sequential methods. Finally, an oceanic application of this approach is discussed comparing the performance of the particle filter designs with that of the classical unscented Kalman filter.