Research On Robust Beamforming Algorithms

Beamforming algorithm is an important research content in array signal processing field.It is widely used in radar,sonar,target detection,speech signal processing,medical imaging,parameter estimation,space radio and other fields.Traditional beamforming algorithms are mostly based on the case where the desired signal information is accurately known and the training data does not contain the desired signal.However,many application scenarios cannot meet the ideal conditions,therefore,the application of traditional beamforming algorithm is limited.The researchers have paid great attention to Capon beamformer due to its excellent performance and flexible expression.However,the beamformer relies on accurate experimental data and ideal experimental environment.In recent years,many improved algorithms based on Capon beamformer have been proposed to improve the robustness and adaptability of the beamformer.Based on the theoretical research of beamforming algorithm,three improved algorithms are proposed to make the better performance of traditional beamforming based on Capon beamforming in this dissertation.In the traditional beamforming algorithm,the error of steering vector will degrade the performance of beamforming greatly,a kind of beamforming algorithm based on the combination of steering vector optimization and diagonal loading is proposed in this dissertation.In this method,the cosine distance is used to determine the basis vector with the greatest correlation with the desired signal,then,the steering vector is modified by the correlation operation of basis vector.Combined with diagonal loading,the beamforming is realized.The selection of diagonal loading is determined by matrix decomposition and mean calculation.The proposed algorithm can deal with the problem of steering vector mismatch,and it has relatively low computational complexity,good adaptability and robustness.Simulation results show that the performance of the proposed algorithm is with advantages compared with many other improved beamforming algorithms.The traditional beamforming algorithm is prone to subspace ambiguity under low signal to noise ratio.In consideration of the problem,a kind of beamforming algorithm based on the combination of steering vector projection and covariance matrix reconstruction is proposed in this dissertation.The expected signal components are removed from the covariance matrix,and the steering vector is projected to deal with the error of the steering vector in the proposed algorithm.Compared with the reference algorithm,the proposed algorithm has obvious advantages and is also useful for various mismatches.In order to solve the problem that the traditional diagonal loading algorithm is difficult to determine the diagonal loading amount,an improved diagonal loading algorithm is proposed to select the loading amount dynamically.Through mathematical operation,the diagonal of covariance matrix is loaded differently and variable loading is realized.The proposed algorithm is explicit,convenient and flexible.The simulation results show that the performance is superior compared with traditional diagonal loading.It has a good prospect in engineering application.