Linear Precoding for Downlink Massive MIMO with Delayed CSIT and Channel Prediction

We consider a multi-cell multi-user downlink channel of a time-division duplex (TDD) MIMO system, where the base stations (BSs) employ the concept of massive MIMO, i.e., they are equipped with a large number of antennas. In addition, the number of users increases with the same speed. Focusing on the practical impairments of the channel such as pilot contamination and, in particular, delayed channel state information at the transmitter (CSIT), we derive an approximation of the sum rate with regularized zero-forcing (RZF) precoding, which provides a quantification of the capacity loss. As a result, it is deemed necessary to obtain the deterministic equivalent sum rate by incorporating in our analysis channel prediction circumventing the degradation due to delayed CSIT. The proposed results are accurate for realistic system dimensions, as simulations testify. Finally, we show the benefits of applying RZF in the sum rate against using eigenbeamforming (BF) for the same Doppler shift with no extra computational complexity.