CSI reduction of MIMO-OFDM systems by parameterization

Multiple antenna systems have been attracted for a long time due to the outstanding capacity and error performance. Corresponding to the space-time coding, the transmitter could apply the precoding or prefiltering based on the fed-back channel state information (CSI). The feedback of beamforming vectors, which are the essential CSI, have been discussed by many authors and some efficient schemes have been proposed. However, the power allocation among antennas, sub-channels, and users are also important issues for increasing the channel capacity and improving the transmission performance. To implement efficient power allocation, the receiver needs to feed back the estimated channel responses (CRs) to the transmitter. In this paper, we propose several efficient CRs feedback algorithms based on the parameterization of CRs. Our development is first by the regressional polynomial model of the CRs across the sub-channels of the orthogonal frequency-division multiplexing (OFDM) transmission. The polynomial coefficients are extracted as the parameters of CRs. To efficiently feed back the large amount of the polynomial coefficients for the wideband systems of multiple-input multiple-output (MIMO) with OFDM transmission, we propose efficient quantization algorithms based on the Gaussian quantization (GQ), Delta modulation (DM), and the recursive least-squares (RLS) method. The performance of our algorithms are evaluated based on the mean-square error (MSE) of the fed-back regressional polynomial coefficients, beamforming system, and the channel capacity. We also consider the channel phase rotation and prediction approaches for further reduction of feedback load. The simulation results show that some of the proposed algorithms could attain the upper bound of channel capacity of the MIMO-OFDM system with low feedback load in the time- and frequency- selective channel.