FDD massive MIMO with analog csi feedback

Massive multiple-input multiple-output (MIMO) is a breakthrough in cellular networking, through using a large number of antennas to serve simultaneously many users. Most prior work on massive MIMO assumes time-division duplex (TDD) operation to take advantage of channel reciprocity. Typical cellular systems, however, make use of spectrum dedicated for frequency-division duplex (FDD) operation. In this paper, we consider an FDD massive MIMO network where the base stations obtain downlink channel estimates via analog feedback from the users. Assuming an uncorrelated channel model, we derive closed-form expressions of the downlink achievable signal-to-interference-plus-noise ratio (SINR) and the corresponding rates. These expressions provide a means for studying rate with and without feedback overhead penalties. A surprising result is that depending on system parameters like the Doppler shift, it is desirable to shut off a subset of antennas at the base stations to maximize achievable downlink sum-rates.