Research On Nested Array Design And DOA Estimation Algorithms

Direction of arrival(DOA)estimation is a key technology in the field of electronic reconnaissance and electronic countermeasures,which is important to the performance of relevant systems.DOA estimation techniques based on multi-sensor arrays can realize simultaneous estimation of multiple signals with the advantages of high precision and super resolution,which has received wide attention from researchers.Most of traditional DOA estimation techniques are proposed for uniform arrays.And,their DOA estimation performance is restricted as the inter-element spacing and array aperture of uniform arrays are limited by half wavelength of signals.Sparse arrays,in contrast,are not limited by half wavelength of signals and have larger inter-element spacing and array apertures,which offers a possibility to improve the DOA estimation performance.Nested arrays(NAs),as one type of sparse arrays,have simple array structures,large array apertures,and high degrees of freedom(DOFs).DOA estimation techniques based on NAs can construct continuous virtual arrays with significantly increased DOFs,which enables underdetermined DOA estimation.Accordingly,based on the current research status of NAs and DOA estimation,this thesis investigates the NA design and DOA estimation algorithms,and proposes several optimized NAs and DOA estimation algorithms to improve the DOA estimation performance.The main research work in this thesis can be summarized as follows:Firstly,the structural optimization techniques for one-dimensional(1-D)NAs are investigated.To address the problem of limited DOFs and array apertures of 1-D NAs,several improved NAs are proposed by utilizing the spatial and temporal characteristics of received data as well as the noncircular characteristics of signals.By exploiting the spatial and temporal information of received data as well as the noncircular information of signals,these improved NAs can construct virtual arrays consisting of sum sets and difference sets,which enable dramatically increased DOFs.Based on the relationship between NAs and their sum sets as well as difference sets,array structures are systematically optimized to effectively increase the array apertures.In addition,mathematical relationship between reference sensors and non-reference sensors is also given in the design of NAs based on the spatial and temporal characteristics of received data.By improving the sensor utilization,DOFs and array aperture are effectively increased.The above improved NAs can effectively improve the maximum number of identifiable sources and DOA estimation accuracy,which makes it possible to complete accurate DOA estimation of multiple signals using fewer sensors.Simulation experiments verify the superiority of these improved NAs.Secondly,the two-dimensional(2-D)three-parallel NA design and reduced-dimension DOA estimation technique via sparse representation are investigated.To address the problem of DOA estimation for 2-D parallel NAs,two kinds of three-parallel NAs and a reduced-dimension DOA estimation algorithm via sparse representation are proposed.The proposed arrays consist of three parallel nested subarrays,which increase the DOFs by systematically optimizing sensor positions and provide possibility for the improvement of DOA estimation performance.Two cross-covariance matrices can be constructed by keeping the parallelism of three nested subarrays.Then,vectorization operation can be performed to obtain reduced-dimension virtual received data,whose corresponding virtual array is fully continuous.This makes all virtual sensors available for increasing the maximum number of identifiable sources.Finally,reduced-dimension DOA estimation is achieved by joint sparse representation and total least-squares estimation technique.Simulation experiments demonstrate the effectiveness and superiority of the proposed arrays and algorithm.Finally,the 2-D DOA estimation technique for L-shaped NA is investigated.To solve the problems of angle pairing,angle ambiguity,and limited estimation capability in DOA estimation of L-shaped sparse array,a 2-D DOA estimation algorithm,named as the joint 2-D vectorized conjugate augmented processes and unitary ESPRIT,is proposed based on the spatial and temporal characteristics of received data.To enable the 2-D DOA estimation of L-shaped NA without angle pairing process and angle ambiguity,generalized planar NA is proposed first,followed by analyzing the relationship between L-shaped NA and generalized planar NA.Combining the spatial and temporal characteristics of received data,transformation principle between L-shaped NA and generalized planar NA in virtual domain is presented,which can extend properties of 1-D NA to 2-D domain without changing the characteristics of 1-D one.Subsequently,covariance matrix is estimated utilizing the received data of virtual planar NA,the vectorization of which can generate a virtual planar array with dramatically increased DOFs.Finally,the estimation of 2-D angles is completed simultaneously by using unitary ESPRIT algorithm,which effectively avoids the angle pairing and angle ambiguity problems in L-shaped array DOA estimation and enhances the 2-D DOA estimation performance.Particularly,the proposed algorithm can be applied to other any type of L-shaped sparse arrays.In the end,simulation experiments verify the superiority of the proposed algorithm.