The Research Of Robust DOA Estimation Based On Sparse Structure

Direction of arrival?DOA?estimation plays an important role in array signal processing and has widely application in many fields such as radar and sonar.The accuracy of the traditional direction of arrival estimation algorithms cannot be guaranteed due to the effect of the malfunction sensors and seriously contaminate signals by impulsive noise.Therefore,the robustness DOA estimation is an important research topic in array signal processing in order to achieve the accurate direction parameters.Moreover,the computational complexities of the state-of-art algorithms for two-dimensional parameter estimation is very high and these algorithms cannot be applied to the practical engineering environment.Thus,it is very meaningful to investigate the fast and robust two-dimensional DOA estimation algorithm.A series of new and effective DOA estimation methods have been provided to eliminate the effect malfunction sensors and impulsive noise.The main contributions and innovative points of this dissertation are listed as follows:Firstly,the investigation history,current situation and the mathematical models of the DOA estimation are detailed introduced in this paper.Then,several subspace-based DOA estimation algorithms and sparsity-based DOA estimation algorithms are introduced.In order to improve the DOA estimation performance under the condition of malfunction sensors and impulsive noise,a underdetermined DOA estimation method using the Khatri-Rao?KR?transformation is proposed to deal with the malfunction sensors and a robust DOA estimation using???_2,p-norm is proposed to suppress the effect of the impulsive noise,respectively.For the malfunctioning sensors in the received array,the proposed method using the quasi-stationary signal characteristic to eliminate the effect of the malfunctioning sensors.The quasi-stationary signals has the statistical property that it remains locally static over one frame but exhibit differences from one time frame to others.The special time domain property enables us to perform underdetermined direction-of-arrival estimation in space domain.By exploiting the temporary diversity of the quasi-stationary signals,the Khatri-Rao operation can be used to transform the local covariance matrix in each frame to a new matrix with large virtual array aperture.Since the Khatri-Rao operation is equivalent to repairing the spatial response of the failure sensors,the proposed algorithm can solve the DOA estimation problem in the presence of malfunction sensors.Theoretical analysis and simulation results demonstrate that effectiveness and superiority of the proposed method compared with the MUSIC algorithm in terms of spatial spectrum estimation performance and estimation accuracy.To address the DOA estimation in impulsive noise environment,inspired by the???_p-norm,the???_2,p-norm and the???_2,p correlation coefficient between matrices are proposed in this paper.The alternating convex optimization algorithm is adopted to achieve efficient solution for???_2,p correlation coefficient.Combining the???_2,p correlation coefficient with the orthogonal matching pursuit?OMP?theory,a more robust???_2,p-OMP algorithm under the impulsive noise environment is proposed for DOA estimation.The???_2,p-OMP algorithm can recover the position and energy of the observed signal by calculating the???_2,p correlation coefficient of observed signal and sense matrix,so the???_2,p-OMP algorithm can accurately reconstruct the sparse signals in impulsive noise environment.The proposed???_2,p-OMP algorithm can realize the robust DOA estimation for 1D-DOA estimation in multiple snapshots received data and 2D-DOA estimation in single snapshot data.The simulation results demonstrate that the proposed???_2,p-OMP method can suppress the outliers existed in the impulsive noise and avoid the spectrum peak search process,which further verified the robustness of the proposed method.Finally,the main contributions of this dissertation are summarized and the shortcomings are pointed out.And the future research topics are also briefly presented.