DOA Estimation Performance Analyses Of Coprime Arrays Based On Deconvolution Method

Direction-of-Arrival(DOA)estimation is a basic problem in array signal processing.How to improve the performance of DOA estimation with a limited number of sensors is a very challenging and significant topic.Coprime arrays based on a combination of two subarrays of coprime numbers(M,N)of sensors have received a lot of attentions due to that a small number of M+N-1 sensors can achieve the same angular resolution of a uniform array of many more(MN)sensors and its co-array achieves the same degree of freedom as that uniform array.However,the proposed product and min processing for coprime arrays have high sidelobes and can produce false targets depending on the source bearings.The high sidelobe problems can be solved using the MUSIC algorithm,but it assume high SNR and a known number of targets which are not the case for many applications such as sonar.This paper studies the DOA estimation of the coprime array and semi-coprime arrays,and discusses the performance of various methods including multiple targets.We propose to apply the deconvolved conventional beamforming(DCBF)algorithm to the coprime array.The advantages and disadvantages of the existing methods and the proposed deconvolution method are evaluated based on the main lobe widths,side lobe levels,grating lobes,array gain,degree of freedom and compared to verify the effectiveness of each method for target detection.Array gain is analyzed for the product and deconvolution methods;the non-linear methods are excluded.The main content of this paper can be summarized as follows:1.We give a comprehensive review of the coprime array theory,the basic signal processing methods,the derivations of the associated signal models,and elaborate the advantages and disadvantages of various methods.2.For the frequently-used product processing in coprime array,a coherent processing scheme is proposed,which can greatly improve the array gain of this method,and the reason is explained by theoretical derivation.3.DCBF is applied to the coprime array,which greatly improves the resolution and accuracy of DOA estimation and improves the array gain.In the single target and multi-target scenarios,the results are compared with other methods,and the processing performance of different methods in different scenarios is analyzed.4.The virtual array or co-array methods which were introduced to detect multiple targets(with a number larger than the number of array elements)are reviewed.To further increase of number of detectable targets,we compare two methods for hole filling in the virtual array using the nuclear norm optimization and IFT method.The performance analysis are new.The performance of DCBF and virtual array methods are compared.5.The array structure of the recently proposed semi-coprime array is reviewed.We point out the shortcomings of its original processing method in multi-target and other scenes,and propose to use the virtual domain MUSIC method and deconvolution method,yielding an improved resolution and ability to detect multiple targets.