Generalized Superimposed Channel Estimation for Uplink RIS-aided Cell-free Massive MIMO Systems

This paper proposes a generalized superimposed channel estimation scheme for an uplink cell-free massive multiple-input multiple-output (mMIMO) system, which is aided by several reconfigurable intelligent surfaces (RIS) to enhanced performance in terms of coverage and spectral efficiency. We consider that the system has both direct links (between access points (APs) and users) and indirect links through each RIS. The estimated channels are used to detect the data streams, and consequently, bit error rate (BER) and sum-rate performances are evaluated. Moreover, we optimize phase shift coefficients to minimize the channel estimation error statistics. Two levels of receiver cooperations (fully centralized processing and local processing) are considered in this work. Simulation results show that the RIS-aided cell-free mMIMO system with the optimal RIS phase coefficients can effectively improve the channel estimates as compared to those without RIS or with randomly phased RISs. It is also verified that fully centralized processing provides much lower channel estimation normalized mean-square error (NMSE) and BER than that for local processing, and confirmed that generalized superimposed training (GST) scheme shows the better performance in channel estimates compared with the standard superimposed training (ST) and the regular pilots (RP) scheme.