Novel Transmit Beamforming Schemes for Time-Varying Fading Channels

Transmit beamforming has been widely adopted for wireless systems with multiple transmit antennas. For an independent block fading channel, the Grassmannian beamformer has been shown to provide very good performance using limited amount of feedback. However, the original Grassmannian beamformer does not take the time domain correlation of the channel fading into consideration. In this work, based on a first order auto-regressive (AR1) dynamic fading model, we develop two new classes of beamforming algorithms that exploit the interframe correlations in the channel fading. The first algorithm is based on a predictive vector quantization (PVQ) approach, and the resulting PVQ beamformer accomplishes very good SNR performance. In addition, to simplify the implementation complexity, we also develop a successive beamforming (SBF) algorithm. The new SBF scheme uses the knowledge of the previous fading blocks to aid the beamforming codebook design of the current fading block. Through numerical simulations, we demonstrate that the proposed PVQ beamformer and successive beamformer outperform several other previously proposed beamformers at various fading scenarios.