Research On Channel Estimation Of Massive MIMO System Based On PARAFAC Model

As one of the key technologies of the fifth generation mobile communication technology,large-scale multiple-input multiple-output(MIMO)technology solves the problems of low spectrum utilization,low power efficiency,and poor transmission reliability by increasing the antenna dimension.The increase in the number of transmit and receive antennas leads to an increase in the dimension of the channel matrix,which results in a large computational complexity for channel estimation.The pilot-based channel estimation algorithm can improve the channel estimation performance,but this method has the problem of pilot pollution,which will reduce the channel estimation accuracy and system performance.At the same time,pilots occupy system resources and have a significant impact on transmission performance.While reducing the computational complexity,reducing the influence of pilot pollution as much as possible has become a focus for scholars.This dissertation focuses on solving these problems and has developed a parallel cell factorization(PARAFAC)based single-cell MIMO system for channel estimation related research.Its main contents are summarized as follows:1.In order to improve the accuracy of large-scale MIMO channel estimation algorithms and reduce the pilot overhead,a channel estimation scheme dominated by the PARAFAC model is proposed.The proposed scheme constrains the constant modulus structural characteristics of the modulated signal and constructs a CM-PARAFAC model with a constant modulus constraint.The theoretical analysis proves that the proposed method can still be relaxed when the loading matrix contains a column correlation.Under unique conditions,channel identification and signal detection are realized.The simulation results show that the performance of the channel estimation is slightly better than that based on the PARAFAC model;but compared to the non-blind estimation method,the higher performance estimation effect can be achieved with fewer pilots.2.To reduce the complexity of channel estimation caused by high-dimensional data processing in massive MIMO systems,this paper combines the sparse theory with the PARAFAC model and proposes a PARAFAC model based on compressed sensing,which is applied to multiuser joint channel estimation.The idea of this method is described as a physical channel sparse channel,based on this configurationand low dimension parallel factor decomposition model,thereby reducing the computational complexity and accelerate the convergence.The simulation results verify that the proposed method makes full use of the space,time,and frequency multi-domain information in wireless communication systems.In the case of low computational complexity,only a few pilots are needed to obtain multiple users' signals.matrix,the channel matrix estimation value and the estimation accuracy of the channel near the uncompressed PARAFAC model estimation algorithm,the estimation method compared to the pilot has a substantial lifting.3.Projection-Regular alternating least squares(P-RALS)fitting algorithm and compression PARAFAC fitting algorithm are proposed and applied to the fitting of the constant modulus constrained PARAFAC and the compressed PARAFAC model.In this paper,the unique decomposition conditions of the new algorithm are deduced,and the influence of the relevant parameters on the normalized mean square error performance of the channel is simulated and the basis for the reasonable selection of algorithm parameters is provided.