Matched field processing for underwater source localization

Summary form only given, as follows. Conventional direction-of-arrival (DOA) estimation assumes that the signals incident on the receiving array are plane waves, an incorrect assumption when the speed of wave propagation in the medium is a function of position. Matched field processing generalizes DOA estimation by using replica fields which are not plane waves, but which are solutions to the wave equation in the particular environment present. The strong dependence of the underwater pressure field upon source position allows estimation of both range and depth of the source. However, inaccurate knowledge of the physical characteristics of the propagation medium may result in significant estimation errors. Four methods for performing the matched field processing, the Bartlett, minimum-variance, autoregressive, and MUSIC methods, are investigated both theoretically and via computer simulation. In particular, the sensitivity of each method to signal-to-noise ratio and spatial noise structure as well as to errors in the sound speed profile, is investigated.