Multi-User Multi-Input Multi-Output (MU-MIMO) Downlink Beamforming Systems with Limited Feedback

This paper considers zero-forcing beamforming (ZFBF) MU-MIMO systems in which one stream is transmitted per user. Assuming no knowledge of other users' channel matrices, user terminals (UT) feedback the effective vector channel (not the full MIMO channel matrix) using channel vector quantization and a channel quality index (CQI) indicating its estimated receive signal-to-noise-ratio (SINR). Limited feedback causes residual multi-user interference at the UT, which otherwise would be removed completely by ZFBF. Consequently, neither the UT nor the base station has sufficient information to calculate SINR accurately making MU-MIMO design quite challenging. A blind CQI estimation at the UT is proposed which averages expected interference over all possible co-scheduled transmit beamformers, which is then corrected at the BS based on the actual transmit beamformers used. A joint method for receive beamformer design and channel codeword selection which optimizes received SINR is also presented. Link and system level results demonstrate performance gains compared to prior work.