Accession Number : ADP002615

Title :   High Resolution Spectrum Analysis by Dominant Mode Enhancement,

Corporate Author : NAVAL UNDERWATER SYSTEMS CENTER NEW LONDON CT NEW LONDON LAB

Personal Author(s) : Owsley,N. L.

Report Date : 1983

Pagination or Media Count : 5

Abstract : High-resolution, data adaptive spectrum analysis techniques can exploit prior knowledge of the signal dimensionality and noise coherence to obtain even higher resolution capabilities than previously envisioned. The classical minimum variance (maximum likelihood) and maximum entropy (ME) spectrum estimators form the basis for corresponding enhanced estimators which require only a prior knowledge of signal bandwidth to provide resolution which is limited only by observation time. The basic idea is to identify only that portion of the estimated data covariance matrix which corresponds to the coherent signal portion of the analysis data. This can be accomplished alternatively by eigensystem analysis, singular value decomposition and Cholesky factorization of the estimated signal only covariance matrix. In this paper, the dominant mode form of the analysis data covariance matrix is presented along with expression for the readily derived inverse of this matrix. Next, the minimum variance (MV) and ME spectrum estimators in modal decomposition form are developed along with 3 dB down resolution expressions. Finally, the modal MV and ME estimators are applied to passive sensor array processing for source range estimation.

Descriptors :   *Mathematical models, *Spectrum analysis, *High resolution, Signal processing, Underwater sound signals, Matrices(Mathematics), Power spectra, Covariance

Distribution Statement : APPROVED FOR PUBLIC RELEASE