Robust Adaptive Beamforming Based On Covariance Matrix Reconstruction

Adaptive beamforming is a spatial filtering technique that aims to receive the signal from the desired direction while suppressing interferences and noise,and it is widely applied in various fields such as radar,sonar and wireless communication.When there are steering vector(SV)mismatches caused by non-ideal factors,the robust adaptive beamformers based on sample covariance matrix(SCM)achieve robustness at a cost of degraded capability of interference suppression.Consequently,how to improve the interference suppression capability while keeping the main lobe response distortionless becomes a practical problem,which needs to be solved.The recently developed adaptive beamforming based on covariance matrix reconstruction technique can effectively avoid the phenomenon of desired signal “selfcancellation”,and it can greatly improve the interference suppression capability of adaptive beamformers.However,the existing covariance matrix reconstruction-based adaptive beamformers rely on the prior information of interferences,and they are also highly sensitive to sensor position errors.To tackle these problems,this dissertation presents in-depth researches on the technique of covariance matrix reconstruction,and develops a series of robust adaptive beamforming algorithms based on covariance matrix reconstruction.The main contributions of this dissertation are listed as follows:1.Considering the coexistence of SV mismatches and coherent signals may degrade the interference suppression capability of existing reconstruction-based beamformers,a spatialsector-reconstruction-based adaptive beamforming algorithm using interference uncertainty set is presented,and this algorithm is effective for all types of array geometry.Firstly,this beamforming algorithm obtains a de-correlated SCM using the IAA-APES algorithm.By using the traditional spatial-sector-reconstruction method,the proposed algorithm reconstructs the interference-plus-noise covariance matrix(INCM)by integrating Capon spatial spectrum over the interference SV uncertainty set.Finally,the desired signal covariance matrix(DSCM)is reconstructed in a similar manner of INCM reconstruction,and the SV of the desired signal is estimated as the eigenvector corresponding to the largest eigenvalue of the DSCM.Numerical simulations and experimental results show that the proposed algorithm can effectively suppress coherent interferences in the presence of SV error.2.Considering that the existing reconstruction-based adaptive beamformers are sensitive to sensor position errors,a spatial-sector-reconstruction-based adaptive beamforming algorithm using compensations of sensor position errors for linear arrays is proposed,and this algorithm is effective for linear arrays.Firstly,the proposed algorithm roughly estimates the direction of the desired signal and interferences using low resolution direction of arrival(DOA)finding algorithms.Then the weighted subspace fitting method is used to construct the optimization problem of estimating the sensor position errors.By exploiting the independence of sensor position errors,an iterative algorithm is proposed to solve the position error of each array element in turn.Finally,the estimated sensor position errors are compensated in the procedure of covariance matrix reconstruction using spatial-sector approach,and a more accurate INCM is obtained,and the SV of the desired signal is corrected.Numerical simulations and experimental results show that the proposed beamformer can form nulls at the direction of interferences while keeping the main-lobe distortionless,which increases the beamformer's output SINR.3.Considering that the interference suppression capability of the existing reconstructionbased beamformers may be degraded when sensor position errors are coupled with DOA errors,an adaptive beamforming algorithm based on subspace-bases-transition for linear arrays is proposed.The proposed algorithm utilizes the orthogonality of subspaces and the weighted subspace fitting technique to construct a multivariable nonlinear optimization problem,and uses the genetic algorithm and a quasi-Newton method to obtain a set of non-orthogonal bases of the signal subspace.By using the subspace bases transition technique,a highly accurate INCM is then extracted from the signal subspace,and the and the SV of the desired signal is corrected.Simulations and experimental results show that output SINR of the proposed beamformer approaches the optimal SINR when sensor position errors and DOA errors both exist.Considering the problem of non-uniform Gaussian noise and linear array deformation in practical applications,the subspace bases transition technique is applied to flexible linear arrays,and an adaptive beamforming algorithm for flexible linear arrays is presented.Numerical simulations and experimental results show that in a non-uniform Gaussian noise environment,the output SINR of the proposed beamformer is less than 1 d B from the optimal SINR under the array deformation conditions.4.Considering the insufficient interference suppression capability of existing wideband adaptive beamforming method,the spatial-sector-reconstruction approach is extended to timedomain wideband adaptive beamforming,and a reconstruction-based time-domain wideband adaptive beamforming algorithm is proposed.Firstly,uncertainty sets of stacked SVs are constructed,and then the spatial spectrum in the uncertainty set is double-integrated over the interference sector and the bandwidth of interest,thus reconstructing a wideband stacked INCM.Then the main-lobe response variation inequality constraint is transformed into a weighted part in the objective function,which results in a closed form of weighting vectors of the adaptive beamformer.The stacked SV of the desired signal is then estimated using two different methods.Finally,the performance of the proposed algorithm is analyzed by numerical simulations,and the capability of the beamformer to suppress wideband interferences is verified by an underwater acoustic experiment.