Research On Co-Prime Planar Vector-Sensor Array Signal Processing Based On Tensor

Vector-sensors has more excellent information acquisition ability than scalar-sensors,so it has been widely used in the field of array signal processing.As a kind of nonuniform array with large aperture and high degrees of freedom,coprime planar arrays(CPPAs)can obtain higher direction-finding accuracy and resolution than conventional two-dimensional planar arrays(such as uniform retangular arrays(URAs)).Therefore,the combination of vensor-sensors and CPPAs,has the potential to improve the array signal processing performance.However,it should be noted that the data dimension of vector-sensor CPPA signals is higher and the processing difficulty is larger,so the existing signal processing methods based on matrix algebra theory are no longer suitable.The theory of tensor algebra can effectively process the high-dimensional data.Therefore,this paper utilizes the tensor algebra theory to conduct tensor modeling and processing research on the vector-sensor CPPA signals.The main research contents are as follows:(1)research on tensor modeling and processing of uniform retangular array signals based on scalar-sensors,and give a method to estimate the direction of arrival(DOA)of array signals by using Canonical polyadic(CP)decomposition.Then it is extended to the field of vector-sensors,and a method is given to estimate DOA and polarization parameters of array signals by CP decomposition.Simulation results show that the method is effective and the performance of vector-sensor arrays is better than scalar-sensor arrays.(2)in order to improve degrees of freedom of the traditional scalar-sensor CPPAs,a new tensor-based array signal processing method is proposed.Firstly,the two sparse arrays in the CPPA are split,and the corresponding two received signal matrices are processed into two fourth-order tensors.Then the two fourth-order tensors are cross-correlated.And process the result into an equivalent virtual array to receive signal tensor.The analysis shows that the received signal of a CPPA with 2L~2-1 physical sensors can be processed into a nonuniform virtual URA with(L+1)~4/16 sensors(called sum-difference co-array)by the proposed signal processing method.Finally,the signal DOA is estimated by CP decomposition and the existence of coprime property of the adjacent array elements of the nonuniform virtual URA.Subsequently,it was extended to the field of vector-sensors,which further improved the signal parameter estimation performance.(3)in order to solve the problem of element repetition in the sum-difference co-array corresponding to the CPPA,a new CPPA structure is proposed.In addition,a new method of signal modeling and processing based on tensor is proposed for the new array.The analysis shows that a new CPPA with 4M~2+N~2-1 physical scalar-sensors(where M and N are coprime integers)can be processed into a virtual array with0.34(MN+M+N)~2 sensors by the proposed method.The method is extended to the field of vector-sensor,by using a single vector sensor that the target can be orientated without ambiguity in the whole space.The new CPPA structure is improved,and the spacing between adjacent elements is increased to improve the aperture of the array.Therefore,the new CPPA based on vector-sensors has the characteristics of both high degrees of freedom and high parameter estimation performance.