Research On Efficient Algorithm And Array Error Eliminating Technique In DOA Estimation

Array signal processing is an important branch in signal processing,it is widely used in military and civil fields such as radar,sonar,communication,biomedical engineering,etc.Direction-of-arrival(DOA)estimation is one of the key problems in array signal processing.DOA estimation problem in the following two situations are studied in this thesis:(1)With the increasing requirement for the performance of antenna arrays,the sensor elements in an antenna array become greater in number.In 2D arrays,the 2D DOA estimation algorithm would bear heavy computational burdens due to the use of a large number of sensor elements.So how to improve the computational efficiency of 2D DOA estimation algorithm become a focus of the thesis.For this problem,some low-complexity 2D DOA estimation algorithms have been proposed,but they focus on reducing the computational cost from the spectral search.The calculation of sample covariance matrix and its eigenvalue decomposition still bring heavy computational burdens.Therefore,the thesis proposes an efficient 2D DOA estimation algorithm based on Nystrom approximation.Unlike the conventional eigenstructure-based 2D DOA estimation algorithm,the proposed algorithm estimate signal and noise subspaces with Nystrom approximation,which only need to calculate two sub-matrices of the whole sample covariance matrix and avoid the need to directly calculate the eigenvalue decomposition of the sample covariance matrix.The proposed algorithm can improve the computational efficiency greatly.Complexity analyses and simulation results verify that the proposed algorithm can provide a comparable estimating performance with the conventional algorithm while improve the computational efficiency greatly.The main innovation of the proposed algorithm:it avoid the calculation of the whole sample covariance matrix,only need to calculate the low-dimension sub-covariance matrices,the eigenvalue decomposition is also based on the low-dimension matrices,the proposed algorithm is efficient especially in the case of large scale arrays.(2)Conventional subspace-based DOA estimation algorithms are based on the assumption of ideal array manifold,and they can't be suitable for the case where array errors(sensor gain and phase errors)exist.The thesis will study the array errors eliminating technique in DOA estimation.For this problem,the current array error calibration approaches are sensitive to phase errors,the recently proposed magnitude-only measurements based DOA estimation algorithm can completely eliminate the influence of phase errors,but it become degraded when gain errors exist.Hence,the thesis proposes a DOA estimation algorithm based on phase retrieval in the presence of array errors.At first,the proposed algorithm estimate the gain errors and noise power from the covariance matrix,and then a compensated covariance matrix is obtained.We take the magnitude information of the elements in the first column of the compensated covariance matrix and form a phase retrieval problem.The problem is solved by a sparse phase retrieval approach and the estimating arriving angles are obtained.Numerical results demonstrate that the proposed algorithm outperforms the current algorithms in the presence of array errors.The main innovation of the proposed algorithm:it provide a new approach to convert the DOA estimation problem into a phase retrieval optimization problem,the approach eliminate the influence of gain errors and noise from compensating the covariance matrix,and the effect of phase errors is removed by taking the magnitude squared of the elements in the first column of the compensated covariance matrix.