Analysis of Analog and Digital MRC for Distributed and Centralized MU-MIMO Systems

We present analytical expressions for the expected per-user signal power and interference in a multi-user multiple input multiple output system with analog maximal ratio combining (MRC) on the uplink and with matched filtering precoding on the downlink. This leads to a closed-form approximation of expected per-user signal-to-interference-plus-noise ratio (SINR). The analysis is carried out for different base station layouts, including a centralized collocated array and a distributed system with arbitrarily correlated Rayleigh fading. Unlike the digital MRC system, where the performance improves for a more distributed system, we show that analog MRC performs best in a centralized configuration. Furthermore, numerical results show that both analog and digital MRC benefit from a strong line-of-sight (LoS) presence in Ricean fading channels as the Rice K factor reduces the interference power. We show that in a centralized system, the performance of analog and digital MRC converge with increasing K factor, because strong LoS reduces the diversity of incoming signal power utilized by digital MRC. On the other hand, in a distributed system, the performance gap between these two techniques will remain due to the diversity of signal powers. Our closed-form approximation of the expected per-user SINR is general and appropriate for moderate to large numbers of antennas and arbitrary correlation models.