Investigation of the Impact of Zero-Forcing Precoding on the Variation of Massive MIMO Transmitters’ Performance With Channel Conditions

Abstract

In this letter, the impact of two variants of zero-forcing (ZF) precoding (conventional and minimax) on the performance of massive multiple-input-multiple-output (MIMO) transmitters is investigated through simulations and experiments. In massive MIMO transmitters, the average-power levels across the radio frequency (RF) chains depend on the choice of precoder and the channel conditions. Compared to the conventional ZF precoder, the minimax variant results in significantly less variations in the average-power levels across the RF chains and, consequently, in less disparity between their operating characteristics. Thus, the minimax precoder, albeit requiring more computational resources, reduces the variation of the transmitter's performance with the channel conditions. Experiments performed using the two ZF variants, in conjunction with digital predistortion, on a two-user four-chain MIMO transmitter under 200 different channel realizations revealed that the minimax variant allows for up to 10 dB and 8% reductions in the adjacent-channel power ratio and root normalized mean-square error, respectively, in each chain.