Research On Near-field Source Location Method Based On Fractional Cumulants In Mixed Noise Background

The problem of source localization is one of the three main problems in the field of array signal processing,and near-field source parameter estimation,as a branch of source localization,is a hot research topic for experts and scholars.When the received signal of the array contains α noise with strong impact characteristics and more practical α and Gaussian mixture noise,traditional near-field source parameter estimation methods based on Second Order Moment,High order Cumulant,or Fractional Lower Order Statistics have the problem of significant errors or even failure.In recent years,the breakthrough and development of fractional calculus theory provide a possible way to solve a series of problems existing in the above methods.To solve the problem of existing methods being unable to effectively locate nearfield sources in α and Gaussian mixture noise backgrounds,this paper combines the Fractional Order Cumulant operator with classical subspace class parameter estimation methods and conducts research from different perspectives.The main work is as follows:1.A two-dimensional multiple signal classification algorithm based on fractional order cumulants is proposed to address the parameter estimation problem in the twodimensional model of near-field sources in the background of mixed α and Gaussian noise.This method overcomes the drawbacks of non zero low order moments of noise fraction and nonlinearity of fractional low order statistics operators.A multi signal classification algorithm based on fractional order cumulants and decoupling operation is proposed,which avoids two-dimensional spectral peak search without losing accuracy.It has the advantages of low complexity and high accuracy,and has high engineering application value.2.Based on the 3D near-field source model,a 3D subspace-like estimation signal parameters via rotational invariance techniques is proposed for the near-field source parameter estimation problem of 3D models in the background of α and Gaussian mixed colored noise.To reduce algorithm complexity,this article defines a pseudo covariance matrix based on fractional order cumulant operator,and then uses the Joint Schure Decomposition method for parameter pairing and estimation of near-field sources.The proposed pseudo covariance based algorithm can not only handle Gaussian noise but also suppress α noise.Compared with the proposed 3D class subspace rotation invariant algorithm based on fractional order cumulants,this method greatly reduces the complexity of the algorithm and has more practical application value.3.Propose a sparse near-field source localization method based on fractional order cumulants.This method constructs a matrix vector observation model by utilizing fractional order cumulants output from different array elements,transforming the multi vector solving problem into a virtual single vector solving problem.Then,according to the Lagrange multiplier method,the near-field source parameters are obtained by alternating iteration.Simulation experiments show that the proposed algorithm can effectively suppress α and Gaussian mixed color noise.