Research On Tensor Decomposition-Based Channel Estimation For Millimeter-Wave MIMO-OFDM Systems

The low-end frequency of radio spectrum has become saturated,and spectrum resource is becoming scarcer and scarcer.The development of high-end spectrum has become an inevitable trend for future communications.Millimeter-wave massive MIMO(multiple input multiple output)technology is regard as one of the key technologies to realize the 5G(5th Generation Mobile Communication),and also the research hotspot of the 6G(6th Generation Mobile Communication).The channel estimation in millimeter-wave massive MIMO systems with OFDM(orthogonal-frequency division multiplexing)techniques is one of the hot topics at present.This thesis combines the millimeter-wave communication scenarios with the tensor decomposition techniques,and proposes various tensor-decomposition based channel estimation schemes for different millimeter-wave massive MIMO-OFDM systems.Above all,this thesis investigates the point-to-point millimeter-wave massive MIMO-OFDM system,and puts forward a PARAFAC decomposition-based scheme for joint multiuser uplink channel estimation.In this scheme,the received signal at the base station is formulated as a third-order low-rank PARAFAC tensor.Firstly,three factor matrices including channel parameters are uniquely decomposed from the constructed tensor by the proposed ATALS(Accelerated Trilinear Alternating Least Squares)algorithm,and then channel parameters are extracted from these factor matrices through the one-dimensional search method.Moreover,the CRB(Cramér-Rao Bound)results are derived for evaluation in the situation of multiple users.Theoretical analysis and simulation results reveal that the scheme performs well with a few training sequences,and the mean square error of estimated channel parameters is close to their corresponding CRB results.Compared with other existing algorithms,our proposed scheme outperforms its counterparts both in estimation accuracy and computational complexity.Considering the millimeter-wave massive MIMO-OFDM systems aided by passive RIS(Reconfigurable Intelligent Surface),this thesis utilizes the sparsity of millimeter-wave channels,and proposes a PARAFAC decomposition-based scheme for millimeter-wave channel estimation.The scheme consists of three stages.In the first stage,the ATALS algorithm is proposed for low-rank PARAFAC decomposition.In the second stage,the KRF(Khatri-Rao Factorization)and the KF(Kronecker Factorization)algorithms based on SVD(Singular Value Decomposition)are proposed for estimation of the related matrices including angles variables,from the factor matrices obtained by the first stage.In the last stage,channel parameters are extracted from these related matrices.The proposed scheme not only realizes an accurate estimation of the cascade channel,but also estimates separately the transmitter-to-RIS channel and the RIS-to-receiver channel,up to a trivial scaling factor.Theoretical analysis and simulation results reveal that the scheme performs well with a few training sequences.Compared with other existing algorithms,our proposed scheme outperforms its counterparts both in estimation accuracy and computational complexity.