Characterization of marine noise using Beaulieu series

This paper proposes a computational technique for estimating parameters of probability density functions (PDFs) governing marine noise, using Beaulieu series. The PDFs are assumed to be mixtures-of-Gaussians from the widely used Middleton´s Class A model. The Beaulieu series for such PDFs are derived. Such an approach is orders of magnitude more efficient than approaches based on convolution integrals, and can be done in real-time. The computational complexity for the technique is then derived. Numerical simulations of these estimates show the method to be robust with respect to perturbations prevailing in the ocean. These results make it obvious that such a procedure can be used in conjunction with existing detectors on sonar platforms.