Multiuser Massive MIMO Performance With Calibration Errors

Massive multiple-input multiple-output (MIMO) in time-division duplexing (TDD) systems relies on the channel reciprocity to obtain the downlink (DL) channel state information (CSI) with the uplink (UL) CSI. But in practice, the transmit branch is composed of different radio frequency (RF) circuits from the receive branch, which breaks the end-to-end DL/UL channel reciprocity. Antenna array calibration is thus necessary to restore this reciprocity. However, in the context of massive MIMO, most existing results are obtained assuming ideal calibration. In this paper, we first analyze the signal-to-interference-plus-noise ratio (SINR) performance and the system capacity in the case of nonideal antenna array calibration and spatial correlation among transmit antennas. Then, we derive the calibration requirements for different beamforming schemes to limit the performance degradation within ℓdB. Furthermore, we verify the accuracy of our analytical results with extensive numerical simulations. This work can be utilized to guide the designs of TDD massive MIMO communication systems.