Channel prediction for massive MIMO with channel compression based on principal component analysis

Abstract

Massive MIMO (multiple-input multiple-output) is one of the key technologies to realize 5G (5th Generation). Massive MIMO can be implemented with many antennas at a transmitter and receiver sides, and it can improve transmission quality at high frequency band by transmitting with superposing shift of radio wave toward the direction of the receiver. However, there exists an issue such as the increase of the amount of feedback of channel state information (CSI) from the receiver to the transmitter, due to the enormous number of antennas. For the purpose of solving this issue, there exists the technique to compress CSI to a lower dimension matrix and decrease the amount of feedback, by using principal component analysis (PCA). In the conventional method, the compression matrix to compress a channel matrix is calculated on the basis of PCA, and the compressed channel is fed back from the receiver to the base station (BS). In this method, the compression matrix used in PCA is generated based on the past CSI at the receiver, which leads to the degradation of transmission rate. This is because there is a mismatch between the CSI acquired at the transmitter and that when the transmitter transmits a signal, due to the channel variation during the feedback from the receiver to the transmitter. In this paper, to solve this problem, we propose the method based on PCA with the channel prediction. As the channel prediction, the forward-backward AR (Auto Regressive) model is used, and the compression matrix in PCA is generated from the predicted channels. By the computer simulation, it is shown that the system capacity is increased by generating the compression matrix from the predicted channel that improves the accuracy of channel restoration.