STAR-RIS-Assisted Cell-Free Massive MIMO with Multi-antenna Users and Hardware Impairments Over Correlated Rayleigh Fading Channels

TL;DR

This work analyzes uplink performance of STAR-RIS–assisted cell-free massive MIMO systems with multi-antenna users under hardware impairments and spatially correlated Rayleigh fading. It develops a comprehensive system model and stud ies two uplink processing levels: Level 1 (local AP processing with centralized LSFD decoding) and Level 2 (fully centralized processing), deriving closed-form SE expressions for Level 1 with MR combining and LSFD, and evaluating Level 2 with MR and global MMSE combining. The results show hardware impairments significantly degrade SE, especially at the user side, but gains can be offset by increasing the numbers of user antennas ($N_u$), APs ($M$), and STAR-RIS elements ($L$); moreover, Level 2 consistently outperforms Level 1, particularly when direct links are blocked. The findings offer practical design guidance for deploying STAR-RIS in CF-MIMO, highlighting the value of centralized processing and multi-antenna users in mitigating non-ideal hardware effects.

Abstract

Integrating cell-free massive multiple-input multiple-output (MIMO) with simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RISs) can provide ubiquitous connectivity and enhance coverage. This paper explores a STAR-RIS-assisted cell-free massive MIMO system featuring multi-antenna users, multi-antenna access points (APs), and multi-element STAR-RISs, accounting for transceiver hardware impairments. We first establish the system model of STAR-RIS-assisted cell-free massive MIMO systems with multi-antenna users. Subsequently, we analyze two uplink implementations: local processing and centralized decoding (Level 1), and fully centralized processing (Level 2), both implementations incorporating hardware impairments. We study the local and global minimum mean square error (MMSE) combining schemes to maximize the uplink spectral efficiency (SE) for Level 1 and Level 2, respectively. The MMSE-based successive interference cancellation detector is utilized to compute the uplink SE. We introduce the optimal large-scale fading decoding at the central processing unit and derive closed-form SE expressions utilizing maximum ratio combining at APs for Level 1. Our numerical results reveal that hardware impairments negatively affect SE performance, particularly at the user end. However, this degradation can be mitigated by increasing the number of user antennas. Enhancing the number of APs and STAR-RIS elements also improves performance and mitigates performance degradation. Notably, unlike conventional results based on direct links, our findings show that Level 2 consistently outperforms Level 1 with arbitrary combining schemes for the proposed STAR-RIS-assisted system.

Figures (6)

• Figure 1: A STAR-RIS-Assisted Cell-Free Massive MIMO System deployment.
• Figure 2: Sum SE v.s. Number of Antennas per User with $M=20$, $K=10$, $N_{ap}=4$, $L=16$. (MC Simulations: Solid lines; Analytical Results: Dashed lines with markers)
• Figure 3: Average SE v.s. Number of Antennas per User with $M=20$, $N_{ap}=4$, $L=16$. (MC Simulations: Solid lines; Analytical Results: Dashed lines with markers).
• Figure 4: Sum SE v.s. Number of Antennas per User with $M=20$, $K=10$, $N_{ap}=4$, $N_{u}=4$, $L=16$. (MC Simulations: Solid lines; Analytical Results: Dashed lines with markers).
• Figure 5: Sum SE v.s. Number of APs with $K=10$, $N_{ap}=4$ (until mentioned), $N_{u}=4$, $L=16$. (MC Simulations: Solid lines; Analytical Results: Dashed lines with markers).