A Stochastic-Geometry Analysis on Scaling Up Cell-Free Massive MIMO Systems With a Cost-Efficient Topology

Abstract

Cell-free massive multiple-input multiple-output (CF-mMIMO) has emerged as a prominent technology in the wireless communication due to its outstanding performance. However, the common star topology requires all access points (APs) connect to the central processing unit (CPU) through individual fronthauls, leading to high cost and difficult scalability. In this letter, we use a cost-efficient topology to scale up the CF-mMIMO system, which allows the new AP (NAP) to connect to the CPU through existing APs (EAPs), called chain topology. As the EAP needs to forward signals from NAPs to the CPU, we assume each EAP can connect only a limited number of NAPs. Utilizing stochastic geometry tools, closed-form expressions of fronthaul length for star and chain topology are derived, respectively. Based on them, we find that the chain topology can significantly reduce the system cost. Moreover, the benefits of the chain topology become increasingly evident as the system scales up.