Feedback Reduction based on Clustering in MIMO-OFDM Beamforming Systems

For transmit bearnforming in multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems, channel state information (CSI) need to be available to the transmitter. Clustering is an effective approach to reduce the data rate of CSI feedback. In this paper, different clustering schemes are investigated, and a new clustering method based on mean channel response is proposed. Further, we present an effective feedback reduction method, where the residual frequency correlation of adjacent clusters is exploited to reduce the feedback overhead. The simulation results show that the proposed clustering method has the similar BER performance with low computation complexity relative to the SNR-maximizing clustering, and the feedback reduction approach can reduce remarkably the feedback rate with some performance loss.