High Resolution Processing Technique Of Wideband Array Signals

Array signal processing is an important means of analysing and processing spatial signals. In various application domains, such as radio, sonar, seismology and expand-frequency communication etc, multiply wideband signals DOA (Direction of Arrival) estimation becomes more and more important. This dissertation is devoted to the study of several aspects of the wideband signal processing, concerning wideband signals number estimation, DOA estimation, array noise suppression, array signals fast processing, influences of array geometry structure and array errors on high resolution arithmetic.Wideband signals number estimation is studied in Chapter 2. Two methods of signals number estimation are developed. One is based on Bootstrap resampling. The other is based on sub-band averaging. Both have merits respectively. The Bootstrap resampling method can estimate wideband signal number under non-Gaussian noise condition. The sub-band averaging method can process coherent signals. By these methods, signals number can be estimated prior to focusing transformation, which is contrary to their former. These can avoid accuracy of signal number estimation affecting by focusing transformation.Rules of forming focusing matrixes are studied in Chapter 3. The performance of DOA estimation affecting by the focusing matrix and focusing frequency has been studied. Rules of selecting the optimal focusing matrix and focusing frequency are investigated. A way of forming focusing matrix has been proposed, which is independent of DOA. Because this way doesn't need preliminary DOA estimate, it can improve the accuracy of DOA estimation.Methods of wideband DOA estimation based on high order cumulant are studied in Charter 4. Measures of array aperture extension and noise suppression have been investigated, especially on non-Gaussian noise. A novel algorithm of wideband signal 2D-DOA estimation based on the virtual cross-correlation computation is developed, which can extent array aperture, suppress the noise (including Gaussian noise and non-Gaussian noise) and estimate 2D-DOA at the same time. Computer simulation confirms the methods availability.Technique of fast estimating wideband signals DOA is studied in Charter 5. Principle of reduce rank based on Multi-Stage Wiener Filter has been construed. And it has been used for wideband array signal processing. It has been proved that the orthogonal bases of Krylov subspace equal the signal subspace under some conditions. Conditions of equivalence have been presented also. A method of fast wideband signal DOA estimating has been proposed based on Krylov subspace, which need less computational cost than other methods and doesn't affect the estimation accuracy. The principles of selecting and methods of getting reference vector have been studied also.Influences of array geometry structure and array errors on high resolution arithmetic are studied in Charter 6. Effects on direction finding of array geometry structure have been analyzed. Model of wideband array error is introduced. A new algorithm is developed, which can estimate wideband signal DOA under many kinds model errors coexisting.