Uplink Performance And Scalability of Centralized MMSE in Cell-Free Massive MIMO Systems

Abstract

Cell-Free Massive MIMO (CF-mMIMO) is a cornerstone technology for beyond-5G and 6G networks, designed to eliminate cell boundaries and provide uniform spectral efficiency to all users. However, in dense deployments, the system's uplink performance is severely bottlenecked by pilot contamination and inter-user interference. This paper investigates the uplink performance of a CF-mMIMO system utilizing a Centralized Minimum Mean Square Error (MMSE) receiver combined with Linear MMSE (LMMSE) channel estimation. Under realistic simulation conditions—including log-normal shadow fading, fractional power control, and forced pilot sharing—the proposed architecture is evaluated against a standard Maximum Ratio Combining (MRC) baseline. The results demonstrate that Centralized MMSE successfully suppresses pilot contamination, eliminating the 10^{-2} Bit Error Rate (BER) floor inherent to MRC. Furthermore, the MMSE combiner more than doubles the median spectral efficiency from 4 bits/s/Hz to nearly 10 bits/s/Hz and demonstrates superior scalability, achieving over 70 bits/s/Hz sum spectral efficiency when equipped with 16-antenna Access Points (APs).