Semi-blind receivers based on a coupled nested Tucker-PARAFAC model for dual-polarized MIMO systems using combined TST and MSMKron codings.

Over the past two decades, tensor-based approaches have generated ever-increasing interest for designing multiple-input multiple-output (MIMO) communication systems. In this paper, we propose a new double directional dual-polarized (DD-DP) MIMO system, equipped with uniform rectangular arrays (URAs) at both ends of the link, which combines a multiple Kronecker product of symbol matrices with a tensor space-time (TST) coding, called TST-MSMKron coding. It is first shown that the channel tensor is represented by a fifth-order PARAFAC decomposition, separated into two parts associated with vertically and horizontally polarized receive antennas, respectively. Then, it is established that the tensor of received signals satisfies a new coupled nested Tucker-PARAFAC model whose core tensor is the coding tensor. Capitalizing on this tensor modeling and assuming the tensor coding is known at the reception, two semi-blind receivers composed of two stages are derived for estimating the transmitted symbols and channel parameters (directions of departure (DoD) and arrival (DoA) angles, path loss coefficients). A new rectification method is proposed to ensure the Vandermonde structure of the estimated steering matrices. Parameter identifiability and computational complexity are analyzed for each receiver. Monte Carlo simulation results are provided to illustrate the effectiveness of the proposed TST-MSMKron coding and semi-blind receivers, in terms of symbol error rate (SER) and channel parameters estimation.