Capacity of MIMO systems: Impact of spatial correlation with channel estimation errors

We study the impact of channel spatial correlation on the ergodic capacity of multiple-input multiple-output (MIMO) systems with channel estimation errors in block-fading channels. We consider transmit antenna correlation and use an accurate capacity lower bound to carry out the theoretical analysis for MIMO systems with and without covariance feedback. We show that for non-feedback systems, the capacity increases with channel correlation at low signal-to-noise ratio (SNR). This finding is in contrast to the existing result for non-feedback systems with perfect channel estimation. For covariance feedback systems, we find that the capacity increases with correlation. We also show the robustness of equal power transmission and the optimality of beamforming at low SNR for non-feedback and covariance feedback systems, respectively. Our numerical results validate our analysis and show that the theoretical low SNR analysis can extend to moderate SNR values.