Towed array processing as a Bayesian problem

Abstract

The towed array range and bearing estimation problem is cast in the form of a Bayesian estimation problem. It is shown that by casting the problem in the form of a MAP estimator, the Kalman estimator naturally follows. There are two advantages to this. First, the problem becomes recursive, resulting in an adaptive processor. Second, once the problem is in the Kalman form, it becomes possible to include sophisticated models of the signals and noise in a natural way. Thus, performance can be enhanced, since the models essentially provide a priori information to the processor. In this work an algorithm, which explicitly contains the forward motion of the array, has been developed. It is capable of performing bearing and range (wavefront curvature) estimation with a short array. Here, short means that the ratio of the physical aperture of the array to the range of interest is small. An example using simulated data is given where an array with a length of 45 meters is capable of determining the range and bearing of a narrow band source. Results are shown for several array speeds, signal-to-noise ratios bearings and ranges.