Studies On Robust Beamforming And High Resolution DOA Estimation Techniques Of Underwater Targets

The performance of passive acoustic signal processing techniques would becomeseverely degradedwhen the acoustic source of interest is obscured by stronginterferences. To improve the performance of passive acoustic signal processingtechniques in practical acoustic situations, this dissertation systematically deploystheoretical and experimental studies on beamforming and direction-of-arrival (DOA)estimation techniques in aspects of enhancing the ability of suppressing interferencesand their robustness as well. This dissertation is supported by the Fund of the KeyLaboratory of Underwater Acoustics and Key Laboratory of Underwater SignalProcessing and Control. The main contributions are as follows:1. In the case of passive sonar, a novel approach to adaptive beamformer isproposed, in which firstly the desired signal component is attenuated from incomingdata snapshots by the spatial matrix filter, and then the weight of the adaptivebeamformer is computed using filtered data snapshots. The approach overcomes thedesired signal cancellation problem, and thus the robustness against system errors isgreatly promoted. Computer simulation results demonstrate its validity.2. In order to alleviate the performance degradation of adaptive beamformingmethods in the presence of fast-moving strong interferers, a robust adaptivebeamforming algorithm is presented. Null-broadening technique is used to suppressinterferences, and diagonal loading is employed to enhance its robustness meantime.This robust adaptive beamformer can also be used to provide high quality beam outputdata in beamspace methods. Computer simulation results show that the proposedalgorithm has a good ability of detecting and estimating the DOAs of weak stationarysignals under circumstances of strong interferences.3. An efficient method to alleviate the influence of array mismatch errors isproposed, in which calibrated array manifold is employed in the low-sidelobebeampattern synthesis procedure in the order that the effect of residual interferences isweakened. Anechoic tank experimental results show that calibrated array manifold isable to abate the distortion of beampattern caused by system errors, which guaranteesan expected low-level sidelobe, and therefore the algorithm can get better performanceof DOA estimation.4. The design approach of spatial matrix filter and its application in the DOAestimation are studied profoundly. Due to the insufficiency of conventional matrixfilters, this contribution develops a design scheme based on the second-order cone programming (SOCP). The approach can form nulls in stopband to suppress stronginterferences while guarantee the design accuracy of the passband. Furthermore, twodimensionality reduced adaptive spatial matrix filter are developed in order to reducecomputation significantly at the cost of minor degradation of estimation performance.5. A design approach of adaptive interpolation spatial matrix filter is presented.To the output data of arrays with arbitrary geometry, this approach is able to do somevirtual interpolation transforms, which lead to characteristics of the output data ofuniform linear array (ULA), and ultimately to the employment of methods suitable toULA. The approach can form deep nulls in the directions of interferers, and get betterperformance of DOA estimation. Computer simulation results prove its validity.6. The broadband coherent high resolution DOA estimation methods based onthe constant beamwidth beamforming and on the virtual interpolation technique arestudied, respectively. Both of them are designed through the cone optimizationtechnique, which consequently promote the design accuracy, and hence improveperformances of both methods in high resolution DOA estimation significantly. BothComputer simulation results and lake experiments show that the proposed algorithmscan enhance the performance of DOA estimation.7. In the case of incoherent signals, the proposed algorithm employs theadaptive spatial matrix filter to suppress interferences outside the area of interest; andin case of coherent signals, the proposed algorithm combines the adaptive virtualinterpolation matrix and the spatial smoothing technique, thus forms nulls in thedirections of interferers to suppress them in the meantime of dealing with the coherentproblem. Computer simulation results show that the proposed scheme has a desirableperformance in suppressing interferences thus enhance the ability of estimating theDOAs of weak stationary signals under circumstances of strong interferences,...