Robust Adaptive Beamforming With Vector-Sensor Arrays For Completely Polarized Signal

Adaptive beamforming has a pivotal position array in the signal processing technology field.Compared with scalar sensor array,the electromagnetic vector sensor(VS)array can effetively obtain and process the polarization information of the signal.So it can improve the robustness.Under the conditions of the completely polarized signals,quaternions and tensor are the two main areas of research.Quaternion is a high-dimensional algebraic tools and can make the electromagnetic vector sensor array output as the coefficient of quaternion's one real and three imaginary part.Therefore,quaternions can more accurately describe the structure of an electromagnetic vector sensor array's output signal.Tensor can intuitively show the output data that we want to process.It also can restruct the array output data with the index of space,polarization and sampling snapshots to achieve a muti-dimensional matching.Thus,this paper analyzes the shortcomings of the robust adaptive beamformer in the quaternion and tensor field over the past few years and then the following two methods are proposed.First,interference and noise covariance matrix reconstruction method was proposed,because Q-Capon has a bad performance and may cause a signal cancellation phenomenon in the case of the steering vector error and the polarization mismatch.We reconstruct the interference and noise covariance matrix within a certain range to overcome performance degradation caused by the steering vector error and the polarization mismatch.It has a better performance compared to other methods quaternion field.Second,the enhanced quaternion–valued principal eigenspace projection robust estimates the desired signal steering vector by maximum beamformer output power.It ensures the robust performance while improving output beamformer.It also does not require to set parameters and can be widely used.The simulation results demonstrate the effectiveness and superiority of this method.Third,by exploiting the inherent multidimensional smoothing processes involved in the matrix contractions of the covariance tensor and introducing the diagonal loading method,we propose a way that not only can tackle the coherent interference but also is robust to the steering vector mismatch errors.The DL-Bartlett double-level structure makes it more robust in the case of high SNR.Computer simulations show the good performance of the proposed method.