Maximum Likelihood Estimate of the Coordinates of Wideband Noise Sources in the Case of Unknown Spectral Densities of the Signal and the Noise

We propose a method of obtaining the maximum likelihood estimate of the coordinates (the bearing, in a special case) of a wideband Gaussian noise source in the presence of background interference which is uncorrelated at the elements of the antenna array. The method is adapted to unknown spectral densities of the noise and the signal produced by the source. The Cramer-Rao bound is constructed, which characterize the variance of the coordinate estimate. Empirical estimates of the variance are compared with the Cramer-Rao boundary and the empirical variance estimates by the method of numerical simulation in the case of routine summation of the power values in the frequency channels. It is shown that when the spectral densities of the noise and the signal differ considerably, the proposed method ensures a much better accuracy. It is also shows that the influence of weak correlation of the noise at the receiver elements is insignificant.