Practical Differential Quantization for Spatially and Temporally Correlated Massive MISO Channels

In this paper, an implementable channel quantization scheme that can effectively exploit both spatial and temporal channel correlation for massive MIMO systems is proposed. In limited feedback systems, differential quantization conducted by applying skewing and rotations on a differential codebook is effective on improving the overhead efficiency practically. To apply these techniques in massive MIMO systems, we adopt noncoherent trellis coded quantization (NTCQ) for Rayleigh channel as a foundation. The inherent codebook of NTCQ is defined and investigated thoroughly. Then we propose a scheme that can produce a differential inherent codebook, which makes adaptive skewing and rotations applicable. In numerical simulations, compared to previous approaches on differential NTCQ, superiority of the proposed scheme is significant. It needs no prior statistical knowledge of channel correlation, while high overhead efficiency can also be achieved. The results reveal that in massive MIMO systems, if ideal channel state information at user terminals is assumed to be available, precisely feeding them back to base stations is practical within affordable overhead and computations.