Research On DOA Estimation And Interference Suppression Based On Underwater Array

The propagation of acoustic signals under water is affected by the underwater acoustic channel(UAC),which makes a challenge for signal reception,detection and estimation.The UAC is considered as a complex channel,which exhibits severe transmission loss,significant diffractive,refractive and multipath effects,as well as ambient noise and interference from other sources.Furthermore,the effects introduced by UAC would become more severe in shallow water.In such conditions,array signal processing exploiting spatial coherence may offer improved performance.This dissertation focuses on array signal processing to develop robust algorithms for signal reception in the complex UAC.The main work is listed as follows:Precise acoustic signal parameter estimation is important for diverse marine geodesy surveys and several other applications.However,the received signal from a far-field passive or active target characterized by planar wavefront propagation is frequently affected by strong nearby interfering reflections.Their presence deteriorates the performance of direction-of-arrival(DOA)parameters estimation for the far-field target.In order to enhance the estimate,a novel beamformer is proposed.The beamformer optimizes the receiving beam pattern for far-field detection by maximizing beamformer sensitivity in the direction of the far-field target,with the imposed condition to eliminate interfering signals generated in near-field locations.As the interference suppression only occurs at the position of near-field interferences,a possible blind zone for far-field beampattern present conventional methods is not created.This beamformer is applicable for coherent signals and the scenario with multi interferences.For stationary situations where interferers’ locations are fixed,the proposed beamformer does not require periodic time updates with associated computational load.The proposed method can be extended to several new situations such as acoustic monitoring performed from a stationary platform and subjected to currents,waves,winds and other variables,all of them generating nearby interferences as well as diverse array configurations including 2D and 3D arrays.Zero-mean Gaussian random signals can be classified into three types:super-Gaussian,Gaussian,and sub-Gaussian,based on their kurtosis.This dissertation proposes a beamformer for non-stationary sub-Gaussian interferences based on the minimum dispersion(MD)criterion.A scenario often encountered in practice is that a strong sub-Gaussian interference crosses the observation region of an array with dynamic angle-of-arrival(AOA)such that the performance of MVDR beamfomer and its variants deteriorates.To suppress such interference,a continuous deep null sector over a pre-defined range of dynamic AOA is desired.The beamformer proposed in this dissertation optimizes the output based on the MD criterion with constraints on the average power over the dynamic AOA.In order to reduce the computational load,a linear constraint is applied instead of using the quadratic one.Three algorithms,the Quasi-MVDR,the Descent algorithm based on Newton’s method,and the Modified Gradient Projection algorithm,are used for the minimization of the problem.Simulations demonstrate that the proposed beamformer can suppress non-stationary sub-Gaussian interferences effectively in comparison with existing methods.Exploiting the framework of compressed sensing,the DOA estimation problem is formulated as one of sparse reconstruction.Compressive beamforming utilizing the sparse representation of received data shows superior performance compared to conventional methods under challenging scenarios with coherent sources and single snapshot.Offset and resolution analysis are conducted to investigate the limitations.To achieve the DOA estimation with high resolution,the structure of array is designed for the optimal Gram Matrix of Sensing Matrix.Alternatively,sparse representation of array covariance vectors is possible for DOA estimation with low computation complexity.Based on the sparse model of array covariance vectors,a recovery algorithm with re-weighted concept to enhance the solution is studied.Generally speaking,the larger the array aperture,the higher the resolution is achieved.However,the large array aperture presents difficulty of deployment and associated cost.The virtual array technique exploiting spatial coherence is used to address this problem.Motivated by this approach,virtual hydrophone is studied to solve the Single Channel Blind Source Separation problem that belongs to the underdetermined blind separation problem.The method can be applied to separate marine mammals’ signals from the data received by a single hydrophone in the presence of the vessel’s emitted noise.Further study indicates that using this procedure could improve the performance for blind sources separation.