Unnormalized Large-Scale Fading Precoding and Joint Power Control for User-Centric Cell-Free Massive MIMO

Abstract

Cell-free massive MIMO (mMIMO) stands out as a promising technology for delivering uniformly good wireless services. Prevailing solutions for downlink (DL) precoding are often derived from the uplink-downlink (UL-DL) duality, which correlates DL precoding with UL transmission parameters. Additionally, normalizing the DL precoding vector is a common practice for the sake of power control convenience. In this paper, we introduce a two-layer large-scale fading precoding (LSFP) purely from the perspective of DL transmission. We demonstrate that the optimal two-layer LSFP can be derived from two subsequent signal-to-leakage-interference ratio (SLIR) maximization problems, which is independent of UL transmission parameters. Furthermore, we reveal that normalizing the precoding vector diminishes spectral efficiency (SE) for worse-case UEs, whereas unnormalized precoding schemes, following specifically designed power control policies, ensure much higher SE for worst-case UEs by exploiting the channel diversity at both layers of precoding. We derive closed-form expressions for the proposed unnormalized LSFP scheme and achievable SE under local maximum ratio (MR) precoding and the proposed power control policies. Through analytical and simulation results, we demonstrate that the proposed unnormalized LSFP schemes significantly improve worst-case SE and ensure better fairness among users.