Uplink Performance Analysis of Asynchronous Cell-Free mMIMO With Two-Layer Decoding

Abstract

In practical cell-free (CF) massive multiple-input multiple-output (mMIMO) networks, asynchronous reception occurs due to distributed and low-cost access points (APs), where the signals arrive at each AP at different time. In this paper, we investigate uplink (UL) spectral efficiency (SE) of asynchronous CF mMIMO with spatially correlated Rician fading channel. On the basis of the availability of prior information at APs, we derive the phase-aware minimum mean square error (MMSE) and non-perceptual linear MMSE (LMMSE) estimators. To mitigate the inter-user interference, we consider a two-layer decoding method in UL transmission. For the first-layer decoding, maximum ratio (MR) precoding is employed, while the large-scale fading decoding (LSFD) method is utilized in the second-layer decoding. Meanwhile, we consider the scenario in CF mMIMO where there is a large number of user equipment (UE), resulting in high computational complexity. To address this challenge, scalable CF mMIMO (SCF-mMIMO) architecture is proposed. On the basis of MMSE and LMMSE estimators, the novel low complexity partial MMSE (P-MMSE) detector and partial LMMSE (P-LMMSE) detector are proposed for centralized combining. For distributed combining, we also proposed the novel local partial MMSE (LP-MMSE) detector and local partial LMMSE (LP-LMMSE) detector. Numerical results demonstrate that LSFD method can enhance UL SE in CF mMIMO. Furthermore, the impact of performance loss resulting from the absence of phase information is contingent upon the length of pilot. It is minimal when pilot contamination is low. Finally, the simulation results demonstrate that the SE of the proposed detectors closely approximate the optimal combining technique for both distributed and centralized combing. It is important to note that the proposed detectors preserve performance while significantly lowering complexity.