High Accuracy Parameter Estimation Technology For Mixed Far-field And Near-field Sources

At present,sensor arrays are widely used in high-speed wireless communication,radar networking system,underwater communications and target detection.However,with the enlargement of array aperture and diversity of signals,and the complex and changeable transmission channels,the scene where far-field and near-field signal sources coexist is more realistic.Therefore,it is of practical significance and practical value to study the parameter estimation algorithm of mixed far-field and near-field signal sources.Estimation of mixed far-field and near-field sources is a new research hotspot in the field of array signal processing,which deals with the problems of mixed sources classification,far-field sources direction of arrival estimation and near-field sources localization.Although the existing algorithms of mixed far-field and near-field sources classification and localization have made some achievements,there are still many shortcomings and deficiencies in the existing algorithms,such as reducing alculating time cost,improving the localization accuracy of near-field signal source and improving algorithm performance with array error.In order to avoiding spectral search processes of traditional algorithms of mixed far-field and near-field signal sources in uniform linear array,improve estimation accuracy of near-field sources and improve the performance of the algorithm with array error,the main research work of this paper includes research on algebraic algorithm of mixed far-near field sources classification and localization,research on improvement algorithm of near-field sources,research on algorithm of mixed far-near field sources in uniform linear array and uniform circular array under mutual coupling.(1)According to the characteristics of mixed far-near field sources receiving model in uniform linear array,a algebraic algorithm of mixed far-near field sources classification and localization is proposed in this paper,which can reduces the alculating time cost effectively without any spectral search process.In this algorithm,a fourth-order cumulant matrix with special structure is constructed by using fourth-order cumulant method.Then,eigenvalue decomposition of the fourth-order cumulant matrix and appropriate methods are used to obtain two groups of rotation factors.Finally,the classification and localization of mixed far-field and near-field sources are realized by comparing the two groups of rotation factors and the corresponding closed-form solutions.The simulation results show that the proposed algorithm can avoid spectral search and grid error,and reduce the alculating time cost effectively and improve the localization accuracy of mixed sources.(2)In order to deal with the problem that the localization errors of near-field signal source increase as the near-field effect increasing,a algorithm of mixed far-field and near-field sources classification and localization based on array partition is proposed,which improves the localization accuracy of near-field source significantly.The second order Taylor approximation error of time delay parameter is proportional to the aperture of array and inversely proportional to the distance parameter of signal source.The proposed algorithm divides the original array into several sub-arrays based on specific rules,and then obtains the directions of arrival of each signal source in different sub-arrays by using multi-sub-arrays cooperative estimation,and finally solves the non-homogeneous equations to get the least squares solution of source location.The simulation results show that the proposed algorithm can avoid spectral search and grid error,and reduce the computational complexity effectively and improve the localization accuracy of near-field signal sources.(3)In order to deal with performance degradation of traditional mixed far-field and near-field sources classification and localization algorithm under mutual coupling due to the mismatch of receiving samples and theoretical receiving models in uniform linear array,and the shortcomings of existing algorithms,a algorithm of mixed far-field and near-field sources classification and localization based on spatial spectral gradient is proposed.By constructing a special spatial spectrum,the far-field sources directions of arrival and the number of non-zero mutual coupling coefficients are estimated.Then the fourth-order cumulant method is used to effectively eliminate the spatial color noise after mutual coupling compensation.Finally,the parameters of the near-field sources are estimated by closed-form solution.The simulation results show that the algorithm can effectively guarantee the correctness of mutual coupling matrix reconstruction under mutual coupling,and reduce the computational alculating time cost and improve the localization accuracy.(4)In order to deal with performance degradation of traditional mixed far-field and near-field sources classification and localization algorithm under mutual coupling due to the mismatch of receiving samples and theoretical receiving models in uniform circular array,and the technology blank,a algorithm of joint space and time processing for unknown mutual coupling blind calibration and mixed sources identification using uniform circular array is proposed.In the proposed algorithm,a uniform circular array composed of even array elements is considered,and the directions of arrival estimation of far-field sources and the mutual coupling matrix reconstruction are realized by combining space-time joint processing method and matrix rank reduction method.Then the special properties of mutual coupling matrix in uniform circular array and matrix differencing method are used to eliminate the spatial color noise after mutual coupling compensation.Finally,the rank reduction estimation function are used to estimate the directions of arrival of near-field sources,and the second-order subspace searching function are used estimate the range of the near-field sources respectively.