Beyond 5G: Exploiting learning aided precoder for downlink Cell-Free networks

Abstract

Cell-free massive multiple-input multiple-output (CFMM) networks with its ubiquitous coverage at high spectral efficiency (SE), is one of the promising technology for 5G and beyond system. In this study, We propose a new framework for downlink (DL) CFMM system operating under Rayleigh fading channel model. We introduce new deep learning-based precoding scheme that improve the performance of the proposed system by reducing run time and computational complexity as compared to conventional linear precoding schemes. We also introduce an improved version of basic scalable pilot assignment algorithm which further enhances system performance. We derive closed- form expression for average DL spectral efficiency (SE) for the proposed scheme considering channel estimation error and pilot contamination(PC), which is then compared with Minimum Mean Square Error(MMSE), Regularised Zero Forcing (RZF) and Maximum Ratio (MR) combining techniques. We analyse the proposed scheme with perfect channel state information(CSI), instantaneous CSI, coherent transmission, non-coherent transmission, different pilot configuration, non-linear and linear precoding techniques. Numerical results shows that the proposed deep learning based precoding scheme outperforms other conventional techniques. endabstract