On the Information Rate of MIMO Systems with Finite Rate Channel State Feedback Using Beamforming and Power On/Off Strategy

摘要

It is well known that Multiple-Input Multiple-Output (MIMO) systems have highspectral efficiency, especially when channel state information at thetransmitter (CSIT) is available. When CSIT is obtained by feedback, it ispractical to assume that the channel state feedback rate is finite and the CSITis not perfect. For such a system, we consider beamforming and power on/offstrategy for its simplicity and near optimality, where power on/off means thata beamforming vector (beam) is either turned on with a constant power or turnedoff. The main contribution of this paper is to accurately evaluate theinformation rate as a function of the channel state feedback rate. Name a beamturned on as an on-beam and the minimum number of the transmit and receiveantennas as the dimension of a MIMO system. We prove that the ratio of theoptimal number of on-beams and the system dimension converges to a constant fora given signal-to-noise ratio (SNR) when the numbers of transmit and receiveantennas approach infinity simultaneously and when beamforming is perfect.Asymptotic formulas are derived to evaluate this ratio and the correspondinginformation rate per dimension. The asymptotic results can be accuratelyapplied to finite dimensional systems and suggest a power on/off strategy witha constant number of on-beams. For this suboptimal strategy, we take a novelapproach to introduce power efficiency factor, which is a function of thefeedback rate, to quantify the effect of imperfect beamforming. By combiningpower efficiency factor and the asymptotic formulas for perfect beamformingcase, the information rate of the power on/off strategy with a constant numberof on-beams is accurately characterized.