A Fairer Comparison of Processing Implementations in User-Centric Distributed Massive MIMO Systems

This paper performs a new comparison between distributed and centralized processing implementations of user-centric distributed massive multiple-input multiple-output (D-mMIMO) systems, also known as cell-free mMIMO. The impacts of processing capacity limitations and inter-central processing unit (CPU) coordination are considered to perform a fairer comparison. The analysis includes two radio units (RUs) selection methods, varying levels of inter-CPU coordination, and different processing capacity restrictions. Simulation results indicate that processing capacity limitations affect spectral efficiency (SE) more in centralized implementations than in distributed ones. Moreover, centralized implementations may require more inter-CPU coordination to improve SE.