Application of robust filtering techniques to time delay estimation in the Arctic environment

This paper is concerned with the application of robust Wiener filtering theory to the removal of narrowband signal components from broadband signal components. It is the intent that these filters be inserted in line prior to cross-correlation. It is shown that when measurements made are the incoming signals auto correlation function, this method reduces to estimating the spectrum using the Maximum Entropy Method, and the resulting filter is determined as if the estimated spectrum were the actual spectrum.