Rate Analysis in Massive MIMO with Covariance Estimation

Massive MIMO is a key enabler of 5G and beyond, which achieves higher data rates, improves spectral efficiency, and supports more simultaneous users. It has been shown that massive MIMO has an unlimited capacity that increases with the number of base station antennas while the base station perfectly knows the users' covariance matrices. In this paper, we study the performance of the uplink massive MIMO system while estimating the covariance matrices and the users suffering from pilot contamination. We demonstrate that the capacity remains unlimited even when the covariance matrices are entirely unknown (i.e., not estimated) at the base station using multi-cell MMSE for (small-scale) channel estimation and data detection. However, covariance estimation is advantageous in terms of improving the uplink capacity. Our results indicate that the capacity continues to grow without bounds, even in fast-fading channel conditions, as modeled by a time-varying Autoregressive channel model. Monte Carlo simulations verify our results.