Model-Based Processing is essentially a way of incorporating physics into the processing scheme in a self-consistent manner. This work presents some of the techniques that have been applied to acoustic array problems. Three situations are addressed. The first is the bearing estimation problem. It is shown that if the forward motion of a towed array is incorporated into the signal model, the performance, as measured by the variance of the estimate, is significantly improved. The second problem treated is that of range estimation. Here it is shown that, by modeling the signal as a cylindrical wavefront, and including the forward motion of the array, the range of an acoustic source can be estimated with an array whose physical aperture is short as compared to the range of the source. The third problem addressed is that of model-based localization of a source using a fixed vertical array. In this case, the signal is represented by a normal-mode propagation model. This differs from matched field processing in that it includes the propagation model parameters themselves in the scheme, thereby dealing with the so-called mismatch problem, i.e., the problem that arises when the model parameters are not well-known. It also differs from the matched-field approach in that it does not require an exhaustive search over the parameters of interest to obtain a solution. The performance improvements that Model-Based Processing is capable of are demonstrated using experimental results.
PDF
download
878987797
028-85220240
