Research On Source Localization And Numerical Quiescent Pattern Synthesis Based On Adaptive Array Theory In Harmonic Domain

Array signal parameters estimation techniques have played a fundamental role inmany applications involving radar, sonar and communications and so on. With thedevelopment of array signal estimation algorithms in practical applications, theinvestigation on robustness and high accurate array signal parameter estimationalgorithm have received great interest. Direction-of-estimation of multiple sources incolor noise environment, and beamforming problems of the spherical microphonearray are investigated in this dissertation. This research of the dissertation mainlyconsists of the following parts:A novel method is presented to locate near-field sound sources in the presence ofnon-uniform sensor noise. Firstly, we derive the estimated noise covariance matrixfrom the data of array outputs. Next, the received data for parameter estimation is thenprewhitened using the estimated noise covariance. Therefore, highly biased estimationcaused by nonuniform sensor noise is enhanced. Furthermore, the performanceimprovement of standard MUSIC method is confirmed by computer simulations.An approach to adaptive quiet numerical pattern synthesis based on sphericalharmonics theory is presented. By applying the adaptive quiet numerical patternsynthesis method in space domain to spherical harmonic domain, by virtue ofexploiting the property of geometric structure of spherical microphone array, insidelobe region, a mass of virtual interferences are introduced, forming abundant nullareas to suppress amplitudes of interferences and noise. By applying theabove-mentioned method, directive property and robustness of spherical microphonearray are improved largely.