Cell-Free Massive MIMO With Finite Fronthaul Capacity: A Stochastic Geometry Perspective

In this work, we analyze the downlink performance of a cell-free massive multiple-input-multiple-output system with finite capacity fronthaul links between centralized baseband unit and access point (APs). Conditioned on user AP locations, first derive an achievable rate for randomly selected in network that captures effect as compression error. From expression, establish traditional architecture where each serves all users network, becomes zero size grows. Hence, to have meaningful analysis, architecture, model locations two independent binomial processes over region provide accurate theoretical result determine coverage. contrast, asymptotically large consider user-centric is served by specified number nearest APs limits load. For Poisson (PPPs). Since expression function AP, statistically characterize load terms per AP. As exact derivation probability mass intractable, present expressions moments Next, approximate negative random variable through moment matching method. Using results along appropriate distance distributions PPP, coverage typical user. analyses, conclude when transmit power relatively high, more collocated antenna deployment preferred. Further, energy efficiency concave antennas