Bandwidth-Efficient Precoding in Cell-Free Massive MIMO Networks with Rician Fading Channels

Global precoding is an effective way to suppress interference in cell-free massive MIMO systems. However, it requires all access points (APs) to upload their local instantaneous channel state information (CSI) to a central processor via capacity-constrained fronthaul links, consuming significant bandwidth resources. Such overhead may become unaffordable in an ultra-dense network (UDN) in future 5G systems, due to the large number of APs and the frequent CSI uploads required to combat the fast-changing state of the high-frequency channels. In order to address this issue, we propose a novel bandwidth-efficient global zero-forcing precoding strategy for downlink transmission in cell-free massive MIMO systems. By exploiting the physical structure of Rician fading channels, we propose a novel model-based CSI compression mechanism, which decomposes a channel matrix into a line-of-sight (LoS) and a non-line-of-sight (NLoS) components, and then compresses them using a model-based method and a singular-value-decomposition (SVD)-based method, respectively. We also present two optimization-based algorithms to obtain the phase information of the LoS component of the channel, which is then used by the proposed channel matrix decomposition. The simulation results demonstrate the efficiency of the proposed precoding strategy on reducing the upload overhead and improving the bandwidth efficiency.