Localization of broadband ocean acoustic sources.

Abstract

Underwater acoustic propagation is characterized by multipath or multimode propagation. Signal processors can be designed to take advantage of the channel complexity if the environment is known. The proposed technique, channel matched filtering (CMF), synthetically backpropagates the wave front to a hypothesized source location. Passive estimates of source location can be made without knowledge of signal characteristics. CMF is demonstrated to be a good performer in the wideband case, as well as the narrowband case where matched field processing, a localization technique appropriate for narrowband sources, has previously been employed. Several ad hoc variants on CMF are also developed. Extensions of matched field processing for broadband signals are made as well. CMF and its variants are developed in the time domain which leads to computational efficiencies over the broadband matched field processor. Comparisons of the various techniques are made with various signal bandwidths and with channels perturbed by internal waves. Accurate and efficiently calculated impulse responses of the ocean acoustic channel are necessary for the successful application of the CMF technique. Gaussian beam theory (GBT) is developed for the propagation model since it provides impulse responses which are shown to approach the accuracy of mode based models, while the computational complexity of GBT is of the same order as ray based models. A procedure is developed here for the specification of the initial conditions of the GB model corresponding to an ideal point source which compares favorably with the accuracy of mode based models. Impulse responses are shown to be easily obtainable from GBT providing an efficient characterization of the channel in the time domain. Time front visualization of a propagating spatial impulse together with amplitude and phase information given by GBT provide great insight into the nature of ocean acoustic propagation.