Research On Direction-of-Arrival Estimation Based On Sparse Reconstruction

With the development of sparse reconstruction theory,DOA estimation methods using sparse reconstruction are gradually emerging and have achieved a series of remarkable results.Sparse reconstruction based DOA estimation method has strong adaptability to small snapshots,low signal-to-noise ratio(SNR)and correlation signals,and has super high angle resolution.So it has been widely concerned and applied.However,the model mismatch caused by convex relaxation of sparse constraint function,the grid mismatch caused by grid generation,and the effect of mutual coupling of array elements make the construction of the sparse reconstruction model difficult,which seriously restricts the development of sparse based method and its scope of application.This paper aims to study and solve the problems of mode error,grid mismatch,and difficulty of constructing sparse based DOA estimation model with mutual coupling,enhance angle estimation performance and expand the application scope of sparse based DOA estimation methods.The main work of this article is summarized as follows:An off-grid DOA estimation method with successive nonconcex sparsity approximation via proximal splitting is proposed for the model error caused by convex relaxation of sparse constrained functions.In this method,an off-grid signal model of array output is firstly constructed by using the first-order Taylor interpolation technique,and the grid spacing constraint is introduced into the model.Then,a family of complex Gauss functions are used as the sparse inducing function,and a successive nonconcex sparsity approximation is used to make it approach 0? norm gradually,which effectively alleviates the model error caused by the convex relaxation of sparse constraint function,and at the same time avoids the risk of the algorithm falling into the local optimization.Finally,using the differentiable property of sparse promoting function,a new update strategy via proximal splitting alternative is designed to ensure the convergence of iterative algorithm,and realizes the parameters update by solving a series of proximal iterative problems,which greatly improves the computational efficiency of the proposed method.An off-gird DOA estimation method using parametric dictionary learning is proposed to solve the grid mismatch problem caused by spatial grid generation.Firstly,from the perspective of dictionary learning,the DOA estimation problem is regarded as a parametric dictionary learning problem,in which the array manifold dictionary is uniquely determined by the angle parameter of signal incidence,and the parameter ? is constantly learned through the observation datas.Next,a sparse Bayesian reconstruction framework is adopted in the parametric dictionary learning model,and a fast Bayesian inference framework is introduced to update unknown variables,which improves the running speed of the algorithm.Then,based on the analysis of hierarchical Bayesian model,a three-layer Bayesian model is proposed to better induce the sparsity of the solution.At last,the correlation between different snapshot sampling data for multi snapshot sampling data is considered.In the algorithm,the correlation matrix is introduced into Bayesian inference,which effectively utilizes the structural information of multi snapshot sampling data.A sparse on-grid DOA estimation method is proposed in the presence of unknown mutual coupling.Firstly,a new construction method of transformation matrix that can be applied to any array geometric structure is studied.Then,an improved array manifold matrix and array output signal model with mutual coupling are proposed.Using the spatial sparsity of array incident signal,a novel block sparse reconstruction method for DOA estimation with unknown mutual coupling is proposed,which does not have any compensation of mutual coupling parameters and can utilize the all output data of array elements.Finally,this block sparse reconstruction problem for DOA estimation is solved by parameter splitting technique,and the block sparse signal and auxiliary parameters are updated alternately.At the same time,in order to speed up the convergence of the iteration,a prediction-calibration acceleration technology is used,which can effectively improve the convergence speed of the proposed method.A framework of DOA and mutual coupling parameters estimation under unknown mutual coupling and an off-grid DOA estimation method based on sparse regularized least squares(SRLS)with unknown mutual coupling are proposed to overcome grid mismatch in sparse based DOA estimation method under unknown mutual coupling.Firstly,the DOA estimation problem with mutual coupling is transformed into a joint block sparse signal recovery problem,in which the sparsity of joint block sparse signals is described by Frobenious/1? norm.Secondly,in order to overcome the grid mismatch problem caused by grid generation,two offgrid sparse signal model with unknown mutual coupling is proposed by using the first-order Taylor interpolation,and the two models have the same optimal solution.Finally,an alternative iterative updating framework is proposed to update unknown parameters,including joint block sparse signals,mutual coupling coefficient vector and grid offset vector alternately.