Semi-Blind Channel Estimation for FDD Massive MIMO Systems based on Analog Feedback

We propose a joint uplink/downlink channel estimation scheme for frequency-division duplex (FDD) massive MIMO systems, that is based on the analog feedback of the downlink channel state information (CSI) and a blind estimation of the uplink channels. The latter operation is enabled by the correlative coding of the uplink signals. Our scheme reduces the resource overhead that is needed to acquire full CSI for FDD Massive MIMO systems. The proposed technique requires a minimum of $2M$ channel uses in contrast to the start-of-the-art algorithm that requires $2M+K$, where $M$ denotes the number of antennas at the base station, and $K$ denotes the number of single-antenna users. Moreover, the performance of the uplink channel estimation can be strongly improved for low SNR scenarios because $M$ uplink channel uses are utilized (in contrast to $K$ channel uses of the start-of-the-art algorithm Echo-MIMO).