A new robust transmission technique for the multiuser MIMO downlink

We present a new robust feedback and transmit strategy for multiuser MIMO downlink communication systems, termed Rate Approximation (RA), and analyze its performance. The new scheme combines flexibility and robustness needed for reliable communications with the user terminal, under a limited feedback constraint. The scheme responds to two major developments in the recent literature: One observation is that it is not so important to approximate the channel but rather the rate itself. The second observation is that a fixed codebook at the transmitter is often better when simultaneously the channel state information is more accurate. Both observations are incorporated in the new scheme where the transmit and feedback codebook are strictly separated and user rates are brought to the base station subject to a controlled uniform error. The analysis provides two astonishing results. First, under perfect channel state information at the transmitter RA outperforms zeroforcing beamforming (ZFBF) for a large fraction of the practically relevant signal-to-noise ratio range for the considered operating point. Second, under a limited feedback constraint the quantization error scaling of RA is shown to be doubly exponentially in the number of feedback bits, whereas the quantization error of ZFBF with random vector quantization was shown by Jindal to scale only exponentially. Simulations sustain our analytic results showing the superior performance of the new scheme.