Design of Generalized Superimposed Training for Uplink Cell-free Massive MIMO Systems

In this paper, a generalized superimposed training (GST) scheme is used for uplink cell-free massive multiple-input multiple-output (mMIMO) systems to mitigate the pilot contamination effect, where multiple users are served by various access points (APs). In the proposed scheme, the pilots and data sequences are superimposed entire coherence time using a novel precoding technique. We consider variable data length in the Rician fading environment and estimate the channel using the least-squares (LS) estimator, and subsequently, the data are estimated. Moreover, we derive an expression for signal-to-interference-plus-noise ratio (SINR) for the proposed GST scheme. The results are given in terms of bit error rate (BER), sum-rate, and normalized mean-squared error (NMSE) of the channel estimation. We compare our proposed GST scheme, the conventional superimposed training (ST), and the regular pilots (RP) scheme, and showed the benefit of the proposed approach. Finally, two receiver cooperation levels are considered, including fully centralized processing and localized processing. We realize lower NMSE and BER but the higher value of sum-rate in fully centralized processing with the GST scheme.