A Study on Cluster-Centric Cell-Free Massive MIMO System

Cell-free massive multiple-input multiple-output (CF-mMIMO) systems have been attracting a great attention as a promising architecture for the beyond fifth-generation (B5G) systems. CF-mMIMO is designed to deploy a large number of base station (BS) antennas to accommodate a relatively small number of mobile users. Considering that the computational complexity and the fronthaul capacity are limited in any practical system, scalability is the key to making CF-mMIMO practical. Thus, a user-centric (UC) approach is taken, i.e., each user is served by a certain number of antennas with high channel gain (called the antenna-set). Such a CF-mMIMO is called UC CF-mMIMO in this paper. However, as the user density becomes higher, the antenna-sets tend to overlap at a higher probability, and accordingly, severer interference will occur among the neighborhood users, thereby degrading the transmission performance of UC CF-mMIMO. To mitigate this severer interference problem, in this paper, we propose to group the neighborhood users as the user-cluster and to apply multiuser MIMO (MU-MIMO) technology to each user-cluster. This CF-mMIMO is called cluster- centric (CC) CF-mMIMO. In CC CF-mMIMO, user-clusters are first formed and a set of antennas is associated with each user-cluster for performing MU-MIMO communication in parallel. We evaluate the link capacity and the user fairness by computer simulation to reveal that the proposed CC CF-mMIMO outperforms the UC CFmMIMO.