ESTIMATION OF FREQUENCY-DELAY OF ARRIVAL (FDOA) USING 4TH-ORDER STATISTICS IN UNKNOWN CORRELATED GAUSSIAN-NOISE SOURCES

A new family of nonparametric and parametric methods based on fourth-o rder statistics for the estimation of the frequency-delay of arrival ( FDOA) between two sensor signal measurements, corrupted by correlated Gaussian noise sources in an unknown way, is presented. The new family of FDOA estimation methods utilizes the fourth-order cumulants or 1-D Fourier transforms of sliced fourth-order cumulants of the two signal measurements. It is demonstrated that the new family of FDOA estimati on methods suppresses the effect due to correlated Gaussian measuremen t noises and outperforms existing second-order statistics-based FDOA e stimation methods using either cross-ambiguity function or MUSIC algor ithm. Various simulation results are presented for different types of signals, different lengths of data, and different signal-to-noise rati os.