| The estimation of parameters such as direction-of-arrival(DOA)and range of a source signal is an important part of array signal processing and it is widely used in many fields such as radar and wireless communications.However,in practical scenarios,farfield and near-field sources often coexist,and a simple parameter estimation method for far-field source or near-field source is no longer applicable.The existing mixed source parameter estimation methods are generally divided into two categories:(1)subspacebased methods;(2)sparse reconstruction-based methods.However,the subspace-based methods require a large number of snapshots,and it is difficult to ensure high estimation accuracy when there are fewer snapshots,while the sparse reconstruction-based methods need to mesh the required parameters,which will lead to the grid mismatch problem.In recent years,off-grid algorithms have appeared in DOA estimation for far-field sources,which can effectively solve the problems of traditional sparse reconstruction algorithms.On this basis,an in-depth study is conducted in this thesis on the off-grid parameter estimation problem of mixed sources in the far and near fields,and new algorithms are proposed.The main work of this paper includes:Aiming at the problem that the estimation accuracy of existing subspace algorithms is not high in the case of limited snapshot,a mixed-source parameter estimation method based on low-rank matrix reconstruction(LRMR)is proposed.It uses LRMR to recover the covariance matrix of the signal and suppress the noise term.At first,according to the characteristics of the symmetric array covariance matrix,the range parameter is eliminated by extracting the inverse diagonal elements of the reconstructed matrix.Then the extracted elements are reconstructed into the Toeplitz matrix and the DOA of the mixed source is obtained by the rotational invariance technique method.On the basis of the estimated DOA,the polynomial rooting algorithm is used to obtain the range parameter of the nearfield source.Compared with traditional subspace-based algorithms,it has higher estimation accuracy in limited snapshots.A mixed source parameter estimation method based on off-grid sparse Bayesian learning(OGSBL)is proposed to solve the grid mismatch problem in sparse reconstruction algorithms.By using the parameter separation technology to eliminate the range parameter of the near-field source,the dimensionality reduction operation is realized.Then the OGSBL algorithm is used to estimate the angle from the model containing only the DOA parameter,and finally the polynomial rooting is used to obtain the range parameter of the near-field source.The error of the algorithm caused by grid mismatch can be reduced,without two-dimensional search,the problem of too large dictionary dimension can be avoided,and the automatic pairing of DOA and range parameters of the mixed sources can be obtained.The Newtonized orthogonal matching pursuit(NOMP)algorithm is applied to the parameter estimation of the mixed source.First,an off-grid two-dimensional parameter estimation method based on NOMP is proposed for the mixed far-near field sources under the uniform linear array.Gradient updates are performed on the DOA and range to solve the potential grid mismatch problem by discretizing the continuous space.On this basis,the model is further extended,and a NOMP-based mixed target off-grid three-dimensional parameter estimation method is proposed for monostatic MIMO radar in the far and near fields.The parameter separation technology is used to separate the target's direction-ofdeparture(DOD),direction-of-arrival(DOA)parameters and range parameters,the DOD and DOA estimates of the mixed target are obtained with the help of the 2D-NOMP algorithm,and then the 1D-NOMP algorithm is used to obtain the range estimate of the near-field target.In fewer snapshots,the algorithm can improve the estimation accuracy of target parameters,achieve automatic parameter pairing,and have low computational complexity. |
PDF Full Text Request