Multi-parameter Estimation For Wideband Signals Based On Sparse Reconstruction

Source parameter estimation is an indispensable part in the field of array signal processing,which is widely penetrated in the fields of radar,biomedicine,wireless communication and robotics.According to the difference of signal bandwidth,a signal can be divided into narrowband signal or wideband signal.Wideband signal received more and more attention due to carrying large amount of information and strong anti-jamming capability.The existing localization parameter estimation algorithms mostly based on subspace theory,but the inherent drawbacks of the subspace theory limits its practical application.The emergence of sparse signal reconstruction theory provide a possible way to circumvent the problems existed in the subspace methods and can bring high resolution and good robustness to noise.The existing signal parameter estimation algorithms based on sparse signal reconstruction were mainly focused on the narrowband signal of one-dimensional DOA estimation,multi-parameter joint estimation algorithms for wideband signal are less,and most of them suffer from the deficiencies such as low parameter estimation accuracy,low maximum resolving power of array DOA estimation and the additional parameters matching process.This paper concentrates on researching high resolution wideband signal multi-parameter estimation problems utilizing sparse signal reconstruction.Based on the analysis of classical sparse signal reconstruction methods and wideband signal processing methods by the numbers,we in-depth study 1-D and2-D DOA estimation of far-field wideband signals and DOA and range joint estimation of near-field wideband signals step and step.Our aim is to provide a series of new ideas for wideband signal multi-parameter estimation problems in sparse signal reconstruction framework.The main contributions and innovative points of this dissertation are listed as follows:?1?A new estimation algorithm based on sparse reconstruction is proposed for the one-dimensional DOA estimation problem of far-field wideband signals.By preprocessing the broadband signals,we avoid constructing multiple sparse matrices and reduce the algorithm complexity.Based on the sum average transformation,a vector sparse representation method based on second order statistics is presented to reduce the data dimension and concentrate the noise energy.With the help of TLP penalty function,the weighted constraint condition is constructed to enhance the signal sparsity,preferably close to ₀?-norm to reconstruct signals and solve the problem of unfair constraint of traditional penalty function.The algorithm can not only realize the joint estimation of the frequency data of broadband signals,but also improve the precision and angle resolution of the parameters,especially in the cases of low SNR,little snapshots and small source spacing.?2?A new two-dimension DOA estimation algorithm based on the concept of co-prime is proposed under the sparse framework for the estimation of elevation angle and azimuth angle of the far-field wideband signals.The DFT transform of the broadband signals is carried out,then the different frequency pairs are selected according to the co-prime of each band to obtain more additional different positions of the difference coordination matrices virtual matrix elements and increase the degree of freedom of DOA estimation.Based on the sparse consistency of the vector sparse model of the covariance matrices of different frequency components,a broadband signal parameter estimation model based on joint sparse reconstruction is presented to improve the positioning accuracy and reduce the computational complexity.By means of the obtained elevation angle estimate to construct azimuth angle estimation model,and put it under the sparse reconstruction framework to achieve joint estimation to avoid parameter matching.The algorithm takes into account the same information between the bands of broadband signal,which can effectively increase the degree of freedom,estimate more information sources,and do not need parameter matching,has high estimation performance.?3?According to the problem of DOA and range joint estimation of near-field wideband signals,a new algorithm based on sparse reconstruction is proposed referring to the spatial-only model.Constructing array output which is not related to the frequency components of the broadband signal,but only to the DOA and range parameters of the signal based on spatial-only model.By the symmetry of the array,extract some elements of the second-order statistics of the array outputs to construct the observation model only related to signal DOA information,and put it under the sparse reconstruction framework to transform into a LASSO optimization problem to solve.Finally,the observation vector is constructed only related to range parameter,and the range parameter is solved by the sparse reconstruction algorithm.The algorithm does not depend on any prior parameter estimation results and need any frequency decomposition or focus process,so it is not affected by forecast angle and can improve the parameter estimation performance,and does not need additional parameters matching process.