Scheduling for Multiuser MIMO Broadcast Systems: Transmit or Receive Beamforming?

In this paper, we present an approximation formula and the close-form expression for the sum rate of the transmit and the receive zero-forcing (ZF) multiple-input multiple-output (MIMO) broadcast systems with user selection, respectively. Instead of assuming a large number of users to obtain a scaling law as most current work, we derive the sum rate formulas of the ZF MIMO broadcast systems with a small number of scheduled users. By analysis and simulations, we find that when taking the variations of feedback channel into account, the receive ZF MIMO broadcast system is more robust to feedback errors and can deliver equal or even higher sum rate than the transmit ZF MIMO broadcast system. We discuss whether a feedback channel is suitable to send channel state information (CSI) for calculating transmit antenna beamforming weights, or suitable to send CSI for selecting users in the receive ZF MIMO broadcast system. Our results show that as the variation of feedback channel errors increases from 0.5 to 1.5, the receive ZF 3 × 3 MIMO broadcast system can provide 36% to 116% higher sum rate than the transmit ZF 3 × 3 MIMO broadcast system in the case of 20 users at signal to noise ratio (SNR) equal to 20 dB. Providing that more feedback bandwidth and an error-free feedback channel are available, the transmit ZF MIMO broadcast system can achieve higher sum rate than the receive ZF MIMO broadcast system.