Channel Correlation Maps for Rate-Adaptive MIMO-OFDMA Systems

Channel frequency and spatial correlation are essential for adaptive channel prediction, which enables flexible scheduling. This correlation information is determined by the environment and can be learned and stored in a map to avoid online tracking. In this paper we propose a method to generate a channel frequency-spatial correlation map from the measurement campaign. Only two position-dependent parameters are stored in the map and the spatial atuocorrelations of the parameters are examined to interpolate channel correlation for the unmeasured positions. Adaptive unbiased prediction of channel gain based on the correlation map is used to compensate the outdated channel state information. The performance of flexible modulation and coding schemes using the distribution of predicted channel gain is shown through numerical simulation.