Lattice Quantization for Centralized and Scalable Cell-Free Massive MIMO with Realistic Fronthaul

Abstract

To implement a practical large-scale cell-free massive multiple-input multiple-output (MIMO) network, the finite fronthaul capacity must be taken into consideration. This paper proposes two scalable lattice-based vector quantization schemes for the fronthaul compression of the fully centralized cell-free massive MIMO network. Both schemes require only the statistical information of the channel and achieve close-to-perfect fronthaul performance with few-bit lattice quantizers under different channel models. With the correlated Rayleigh channel model, adjusting the shape of the codebook according to the eigenvalue distribution of the channel correlation matrix at each access point (AP) improves the achievable spectral efficiency (SE) compared to simply applying the near-circular lattice codebook. Maximum ratio (MR) and minimum mean-square error (MMSE) combining schemes are utilized to evaluate the performance of the proposed quantization methods.