Robust tensor beamforming for polarization sensitive arrays

Robustness is of great importance in array beamforming. With the purpose of improving the robustness of the array beamforming, methods using tensor operations are explored in this paper. Specifically, a higher-dimension tensor decomposition method to construct minimum variance distortionless response model (TD-MVDR) is proposed under the assumption that the polarization sensitive array enjoys the multilinear translation invariant property. Whereafter, the proposed TD-MVDR algorithm is incorporated into the improved conjugate gradient least squares method called TD-ICGLS to obtain a better robustness. Considering that the degradation caused by the presence of the random steering vector mismatches, we derive a diagonal loading model for TD-ICGLS to improve the robustness of it. Moreover, a method for determining the loading level is put forward as the key step for the proposed robust tensor beamformer. Results demonstrate that the proposed diagonal loading TD-ICGLS beamformer yields more robust performance than existing matrix-based solutions, such as global beamforming, while operating in a challenging scenario where the signal-of-interest power approaches the jamming power. Meanwhile, an improvement of the computational complexity in terms of TD-ICGLS is noteworthy.