Precoder Quantization for MIMO-OFDM Systems over Frequency Selective Channels

Transmit beamforming can improve the performance of multi-input multi-output (MIMO) systems. For FDD systems, the beamforming matrix must be quantized and fed back to the transmitter side. Different compact quantization codebooks have been designed to reduce the overhead of the feedback channel. In this paper, we compare the performance of two different quantization methods for a fixed codebook: vector- wise quantization based on householder reflection and the joint matrix quantization. The vector-wise quantization decomposes the matrix codeword into vectors using the Household transform. We find that vector-wise quantization can outperform joint matrix quantization due to the unbalanced gain of spatial eigen- mode of the MIMO channel. We also proposed a recursive sphere chordal distance quantizer (SCDQ) to reduce the computational cost of the matrix quantization. Simulation results are provided to verify our findings.