Robust downlink SINR balancing based on MMSE optimal CSI feedback

Abstract

A robust signal-to-interference-plus-noise ratio (SINR) design is proposed that is based on fed back channel state information (CSI). The channel estimator, the quantizer, and the rank-reduction basis are jointly optimized to get a minimum mean square error (MMSE) optimal CSI feedback. Based on the CSI feedback, the transmitter can compute the conditional covariance matrices of the different channels that enable a balancing of the different SINRs on average. The robust SINR balancing algorithm shows good balancing properties also for the instantaneous SINR and clearly outperforms the non-robust SINR balancing approach.