Tensor beamforming for multilinear translation invariant arrays

摘要: In the past few years, multidimensional array processing emerged as generalization of classic signal processing. Tensor methods exploiting multidimensionality provided more accurate parameter estimation and consistent modeling. this paper, multilinear translation invariant arrays are studied. An M-dimensional admits a separable representation in terms reference subarray set M — 1 translations, which is equivalent to rank-1 decomposition an Mth order manifold tensor. We show that such property can be exploited design tensor beamformers operate multilin-early on level instead global level, usually case with linear beamforming. important reduction computational complexity achieved proposed beamformer negligible loss performance compared classical minimum mean square error (MMSE) beamforming solution.