Multidimensional Parameter Estimation Of Array Based On Tensor Decomposition

Array signal processing is an important component module in the field of modern signal processing.It has been widely used in military and civil fields.The problem of parameter estimation is the focus of research in array signal processing.Many classical estimation methods have been developed in this field,but the traditional matrix-based estimation methods can not effectively utilize the multi-dimensional structure information inherent in the received data.Different from this,tensor algebra is suitable for representing the data structure of high-dimensional data.This thesis will focus on the application of tensor decomposition in multi-dimensional parameter estimation of source,and consider a new data representation method to solve the traditional parameter estimation problem.The main work and innovation of this thesis are as follows:(1)A distributed coherent source parameter estimation method based on tensor decomposition is proposed.In the case of small angle diffusion,the integral form of the steering vector is approximated to Hadamard product form.The received data is transformed into tensor model.The factor matrix is solved by PARAFAC decomposition.Because the factor matrix is the steering vector,the central angle information is solved by least squares according to the known steering vector structure.The method achieves the best estimation performance among all the methods based on steering vector approximation,and the computational complexity is less than that of other methods in the case of large array number.(2)A tensor decomposition estimation method for distributed coherent sources based on signal noncircularity is proposed.In order to further improve the performance of parameter estimation,a new third-order tensor model with more parameter information is constructed by considering the non-circularity of the signal to expand the received data.Similarly,the PARAFAC decomposition method is used to solve the angle parameter information.The simulation results show that the performance of the PARAFAC decomposition method combined with non-circularity is better than that of the traditional PARAFAC decomposition method.When the angle diffusion increases,the non-circular PARAFAC decomposition method can restrain the performance deterioration of parameter estimation.(3)A method for estimating moving target parameters of FDA-MIMO radar based on tensor decomposition is proposed.The radar received data is transformed into a thirdorder tensor model,and the factor matrix is estimated.In order to solve the problem of coupling between range and angle in FDA radar parameter estimation,the angle and Doppler frequency shift information are solved separately,and then the range is estimated separately by constructing one-dimensional MUSIC spectral peak search based on the obtained angle of arrival.