A robust transmit CSI framework with applications in MIMO wireless precoding

Abstract

Transmit channel side information (CSIT) can improve performance of MIMO wireless systems by means of precoding. More complete and reliable CSIT provides more precoding gain. Instantaneous channel measurements provide the most potential gain, but suffer from delay-induced error due to channel temporal variation. Statistical channel knowledge provides less gain, but is reliable. In this paper, we propose a framework combining a possibly outdated channel measurement with the channel statistics - the mean and the covariance - to create a dynamic CSIT form, as a function of a correlation factor between the measurement and the current channel. The CSIT consists of an estimate of the current channel and the associated estimation error covariance, which function effectively as the channel mean and covariance. We apply a precoder design exploiting these channel statistics to illustrate the achievable gain and the robustness of the new CSIT framework