Research On DOA Estimation Methods Based On Beamspace And Difference Coarray Domain For Circular Arrays

Direction of arrival(DOA)estimation is of great significance for theoretical research and practical applications in a variety of aspects such as radar and communications.Due to the special array structure,uniform circular array(UCA)can provide azimuth information from 0?to 360?,and can also provide elevation information,making it a main platform for passive detection techniques.Beamspace based root methods for UCAs avoid spectral peak searching,so they have low computational complexity.Moreover,they have high estimation accuracy.Due to the merits,such methods have attracted much attention.However,in practical applications,these methods face many difficulties.The ideal signal model assumption for such methods does not consider non-ideal conditions in practice,such as limited number of sensors,small sample size,and mutual coupling.Under non-ideal conditions,beamspace based root methods will have residual terms when applying beamspace transformation(BT),and then suffer serious DOA estimation performance degradation.In addition,many methods based on UCA can not provide effective DOA estimates in complex environments,such as underdetermined DOA estimation and nonuniform noise scenarios.This thesis researches on DOA estimation for practical applications and simultaneously contains the advantages of UCAs for two-dimensional DOA estimation from the perspective of algorithm and array design.Firstly,for the error caused by the BT under non-ideal factors,this thesis proposes modified beamspace based methods with higher estimation accuracy,in order to improve DOA estimation performance and meet the requirements of fast processing in practical environments.Furthermore,to systematically improve the underdetermined DOA estimation performance in practice,this thesis uses sparse circular array(SCA)as the passive detection platform,and utilizes the property and advantages of SCAs to improve underdetermined DOA estimation performance under non-ideal conditions.At the same time,to solve the problem that the existing SCAs can not provide the same estimation accuracy for different azimuths,a new SCA is designed and can further reduce mutual coupling.The main research results are summarized as follows:Firstly,for beamspace based one-dimensional DOA estimation for UCAs,a Root MUSIC algorithm based on beamspace matrix updating is proposed.In order to removethe error caused by the BT when the number of sensors is relatively small,we first calculate residual terms caused by the BT in beamspace covariance matrix and remove them,and then use Root MUSIC to estimate DOAs based on the updated covariance matrix.Performance analysis and simulation results show that the proposed algorithm effectively improves the DOA estimation performance.Furthermore,under small sample size and with a small number of sensors,a Root MUSIC algorithm based on beamspace matrix updating is proposed.The proposed algorithm considers both the residual terms caused by the BT and the correlation terms between signal and noise under small sample size in the beamspace covariance matrix,and removes these terms when updating the beamspace covariance matrix,thereby improving the DOA estimation performance.We also give the theoretical analysis of the proposed algorithm.Compared with MUSIC,the two proposed algorithms have lower computational complexity,and are more suitable for real-time data processing scenarios.Compared with Root MUSIC,the proposed algorithms have better estimation performance when the number of sensors or sample size is small.Simulation results verify the effectiveness of the two proposed algorithms.Secondly,for beamspace based two-dimensional DOA estimation for UCAs,a new searching algorithm for elevation estimation is proposed at first.The proposed algorithm avoids the error caused by estimating null subspace of the existing method,so the accuracy of elevation estimation is improved.Furthermore,a two-dimensional DOA estimation algorithm based on beamspace data updating is proposed in the presence of mutual coupling and with a relatively small number of sensors.The proposed algorithm improves the accuracy for both azimuth and elevation estimation by iteratively updating the beamspace data through elimating the residual terms caused by the BT.The theoretical estimation error of the proposed algorithm is analyzed.Generally,only one or two iterations are enough for the proposed algorithm to obtain reliable estimation performance.Compared with UCA Rank Reduction(RARE),the proposed algorithm significantly improves the estimation accuracy.Compared with two-dimensional MUSIC,the proposed algorithm has much lower computational complexity,and the estimation performance of the proposed algorithm is close to that of the two-dimensional MUSIC.Simulation results verify the effectiveness of the proposed algorithm.Finally,this thesis researches on underdetermined DOA estimation based on difference coarray domain by using SCAs.We first propose an DOA estimation algorithm for sparse linear arrays based on pseudo data set in difference coarray domain by using therepeated lags in the difference coarray to construct the pseudo data set for a virtual array with larger aperture,in order to improve DOA estimation performance.Then,for SCAs extended from sparse linear arrays,we propose a difference coarray domain matrix completion based underdetermined DOA estimation algorithm for nested sparse circular array(NSCA).The algorithm makes full use of the symmetry characteristics of circular arrays from the perspective of matrix completion to construct a noise-free covariance matrix with more degrees of freedom(DOFs)for a virtual UCA.By using the completed matrix,the DOA estimation performance improves compared with the existing methods.Both the above two proposed algorithms utilize all the information of the data covariance matrix,and work well in the presence of nonuniform noise.The theoretical analysis and simulations show that the two proposed algorithms have good estimation performance.At last,we propose a coprime sparse circular array(CSCA),which has a smaller number of sensors and retains the advantages of circular arrays for two-dimensional DOA estimation.Theoretical analysis shows that compared with the existing NSCA and super nested sparse circular array(S-NSCA),the CSCA can further reduce mutual coupling and provide approximately the same estimation accuracy for any azimuth,and simulations verify the effectiveness of the CSCA for one-dimensional and two-dimensional DOA estimation.