Research On The Algorithms Of Estimation Of Direction Of Arrival Based On Compressive Sensing

Direction of Arrival(DOA)estimation is a very important topic in the context of array signal processing,and has attracted significant interests due to its wide applications in radar,communication and radio astronomy.DOA estimation uses data obtained from sensor arrays to get the angles of arrival and is an essential step before beamforming.At the same time,the field of Compressive Sensing(CS)grows rapidly as an effective means of dimension reduction and has been gaining momentum in the field of array signal processing.Algorithms based on CS have advantages over conventional algorithms and have wide applications in the future.This paper aims to construct DOA algorithms with excellent performance and wide applicability based on the new results of CS.Firstly,this paper introduces the background of the research,and summarizes the research progress of DOA estimation algorithms and the current research status of DOA estimation.And the MUSIC and ESPRIT algorithm,the two most important kinds of subspace methods,is presented.Then the basis of the CS theory is studied.After that two kinds of algorithms based on CS is specifically introduced and analyzed.And the general steps of applying CS theory to DOA estimation are summarized afterwards.Secondly,steps to achieve gridless estimation in CS-based algorithms are studied,and the grid refinement DOA estimation based on Least Square Residual Compressive Sensing(LS-CS)is proposed.The grid effect is unavoidable due to the discretization of the angle space when applying CS to DOA estimation.Researches show that gridless CS and Bayesian CS can be utilized to alleviate the grid effect and have better performance than traditional methods especially in high SNR case.Simulations illustrate the effectiveness of the gridless methods comparing to the methods with grid effect.Simulations also show that the proposed LS-CS based algorithm owns a better accuracy and smaller computational complexity comparing to the conventional algorithm.Finally,the DOA estimation based on the one-bit quantization is studied,and a gridless one-bit DOA estimation algorithm based on the classification is proposed.The one-bit quantization,which only reserves the sign information of the received measurements,has been gaining momentum in future massive MIMO systems.Traditional one-bit DOA estimation algorithm needs a sufficient number of snapshots to obtain the sample covariance matrix while one-bit DOA estimation based on CS is valid in single measurement vector(SMV)case.Based on the sparsity of the incident signals,modeling the one-bit estimation as the classificationproblem is feasible.A refinement process utilizing consistence reconstruction is also presented in the proposed method.Simulations are conducted for validation.