Title: Trade-offs In Quasi-Decentralized Massive MIMO

Abstract: Typical massive multiple-input multiple-output (MIMO) architectures consider a centralized approach, in which all baseband data received by each antenna has to be sent to a central processing unit (CPU) to be processed. Due to the enormous amount of antennas expected in massive MIMO base stations (BSs), the number of connections to the CPU required in centralized massive MIMO is not scalable. In recent literature decentralized approaches have been proposed to reduce the number of connections between the antennas and the CPU. However, the reduction in the connections to the CPU requires more outputs per antenna to be generated. We study the trade-off between number of connections to the CPU and number of outputs per antenna. We propose a generalized architecture that allows exploitation of this trade-off, and we define a novel matrix decomposition that allows lossless linear equalization within our proposed architecture.