Antenna Combining and Codebook Design for the MIMO Broadcast Channel with Limited Feedback

We consider a MIMO broadcast channel where both the transmitter and receivers are equipped with multiple antennas and where channel state information at the transmitter (CSIT) is provided through limited (i.e., finite-bandwidth) feedback from the receivers. Adopting a simple zero-forcing type of beamformer, we show how additional degrees of freedom provided by multiple antennas at the receiver can be exploited to improve system performance. We propose a new solution where the linear receiver and the channel quantization for CSIT feedback are jointly designed in order to maximize the expected SINR for each user. Based on a modified Lloyd-Max algorithm, the proposed channel quantization technique uses empirical channel realizations drawn from channel statistics, creating tailored codebooks suited for any antenna configuration or environment. The resulting codebook also has a tree structure that can be utilized in time-correlated MIMO channels to reduce feedback overhead. Simulation results show the effectiveness of the overall transceiver design strategy compared to prior techniques in a variety of correlation environments.