Design Of Sparse Electromagnetic Vector Sensor Arrays And DOA Estimation

In signal processing,besides the information such as amplitude,frequency,phase and waveform,the polarization information is another important feature available.Compared with the conventional scalar array,electromagnetic-vector-sensor(EMVS)can perceive all the electric and magnetic field components of the incident waves.It can not only provide the DOA estimations,also be able to get the polarization information of the signals.And the performance of multi-dimensional parameters estimation and signal detection can be further improved by combining polarization domain information with spatial domain information.Therefore,EMVS has gradually become a research hotspot in signal processing in recent decades.On the other hand,the estimation accuracy is proportional to the array aperture in the DOA estimation.However,in order to avoid the angle ambiguity,the spacing between adjacent antennas usually should not be greater than the half wavelength of the incident signal.Thus,a larger aperture requires more antennas,but the mutual coupling between antennas and the cost of the whole array also increases.By distributing the array elements using an inter-sensor spacing larger than the half wavelength,sparse array can enlarge the array aperture at the same array cost,also the performance of angle estimation.And the mutual coupling between adjacent antennas can be reduced to a certain extent.Therefore,in this thesis,we focus on the combination of EMVS and sparse array.Aiming at expanding the aperture of the array and reducing the design cost of the array,three sparse array expansion methods for two kinds of EMVS,as well as the corresponding two-dimensional DOA estimation algorithm are proposed,including a multiscale sparse EMVS linear array and two kinds of L-shaped sparse array of EMVS.Comparing with existing arrays,our proposed arrays can resolves more sources,and has better performance for DOA estimation.The specific works are summarized below:1.The multiscale sparse EMVS linear array.Based on the dual parallel line EMVS,the same EMVSs are used as the array element to expand in the orthogonal direction of its own parallel direction.The array is divided into multiple sub-arrays with multiple inter-sensor spacings which greater than half wavelength.And the larger array aperture can be obtained under the condition of same number of array elements.The corresponding DOA estimation algorithm combine the multiple ESPRIT algorithm and the vector-cross algorithm peculiar to EMVS to obtain the estimates of two dimensional DOA and polarization parameters by a multiscale disambiguation.Simulations demonstrate that the proposed multiscale sparse EMVS linear array has a better estimation performance compared with the uniform EMVS sparse linear arrays with same cost.And the algorithm is also applicable to coherent sources.2.The L-shaped sparse arrays of EMVSs.The linear array only has one-dimensional aperture extension,and the array cost will be high by using the whole EMVS as the array unit.Therefore,on the basis of triangular EMVS,combining with the expansion of L-shaped array in two-dimensional direction,and taking a component of EMVS as the array element to reduce the cost,we proposed two L-shaped sparse arrays of EMVSs.The hybrid L-shaped co-prime array sets the inter-sensor spacings as a special co-prime relationship.And the Chinese Remainder Theorem is utilized to resolve the ambiguations and estimate two-dimensional DOA.Simulations indicate that the estimation performance of this array is better than the typical L-shaped array and the multiscale sparse EMVS linear array under same array cost.And the number of sources that can be estimated is no longer limited by the number of components in the EMVS.Soon afterwards,by generalizing the relationship of inter-sensor spacings,the multiscale L-shaped sparse array is proposed which composed of multiple uniform sparse sub-arrays with different inter-sensor spacings.The multi-order disambiguation is utilized to estimate the two-dimensional DOA of targets.Since the selection of inter-sensor spacings has a great impact on the estimation performance,we analyze the threshold problem of inter-sensor spacings in this array,and give the method of threshold calculation.Comparing with the 2-D nested array and the multiscale sparse linear EMVS array,simulations indicate that the array performs better in parameters estimation and robustness to noise.