Localization of wideband signals using least-squares and total least-squares approaches (Q2723880)

The purpose of this paper is two-fold:NEWLINENEWLINENEWLINE(1) First, to introduce a new focusing technique for the coherent signal-subspace processing method. The least-squares and the total least-squares coherent signal-subspace methods are formulated in a general structure; under certain constraints it is shown that the two techniques provide an identical solution. The method can also be viewed as an extension of the coherent signal-subspace algorithm to non-unitary focusing techniques.NEWLINENEWLINENEWLINE(2) The second important purpose of the paper is to use this formulation of the focusing technique for deriving a new focusing algorithm. The focusing matrices in the new algorithm are determined based on a modification of the total least-squares method. The computational complexity of the new focusing algorithm is significantly lower than the one for the rotational signal-subspace (RSS) method, while the focusing gain of the new technique is larger than the focusing gain of the same RSS algorithm. The study of the computer simulations shows that the new method has a smaller bias compared to the RSS method.