Asymptotic analysis of cooperative massive MIMO networks with user centric clustering

We derive a closed-form expression for the throughput in the uplink of a cooperative massive MIMO system. We consider a network where a cluster of base stations co-operate to detect signals from mobiles using a linear minimum-mean-square error receiver. A user-centric clustering approach is assumed, where the cluster that detects the signals from each mobile is composed of the K base stations that are closest to the mobile. The analysis combines stochastic geometry and infinite random matrix theory. We assume that base stations and mobiles are distributed as independent Poisson Point Processes on the plane. Using an asymptotic analysis, a closed-form expression for the spectral efficiency is derived, which is verified by simulations to be accurate even for a moderate number of antennas. This result helps us understand the influence of tangible system parameters such as mobile and base-station densities, number of antennas per base station, and number of co-operating base stations on spectral efficiencies. Our findings show that for a given number of cooperating antennas, it is optimal to have all cooperating antennas in a single base station. But, if the area density of antennas is also limited, then the number of co-operating base stations should be increased instead.