Link-Level Evaluation of Uplink Cell-Free MIMO in 5G NR Over Frequency-Selective Channels

Cell-free (CF) MIMO has emerged as a promising next-generation technology, primarily due to its ability to provide uniformly high-quality service to all user equipment (UEs), regardless of their location. While existing research has extensively explored various aspects of CF systems–including scalability, clustering strategies, power control, and precoding designs–there remains a notable gap in the literature concerning the physical-layer performance of 5G New Radio (NR) within CF architectures. This paper addresses this gap by focusing on the Physical Uplink Shared Channel (PUSCH) transmission over frequency-selective channels. We develop a comprehensive, 3GPP-compliant link-level simulator to evaluate the performance of CF MIMO under realistic propagation conditions. First, we generate results for selected modulation and coding schemes (MCSs) to confirm the simulator’s alignment with expected performance. Then, the effects of key physical-layer parameters–such as subcarrier spacing (SCS), the number of distributed radio units (RUs), and the number of RU antennas–are evaluated using Block Error Rate (BLER) as the primary performance metric. We also compare the results of the CF-MIMO system with a co-located antenna scenario, serving as the baseline for a traditional MIMO system, and confirm that the CF-MIMO system achieves superior performance due to its spatial diversity advantages. The results also show that employing higher SCS values effectively exploits frequency diversity, particularly when the signal bandwidth exceeds the channel’s coherence bandwidth. As expected, increasing the number of RUs significantly improves BLER due to enhanced spatial diversity and reduced UE-RU path loss. We further examine the impact of practical channel estimation by evaluating four different DMRS configurations, confirming that Type 1 with length 2 provides superior performance under the tested conditions. Finally, we investigate the effect of carrier frequency, showing that higher frequencies lead to increased path loss and degraded performance. The findings offer valuable insights into spatial, frequency, and estimation-related interactions in CF 5G NR, while guiding MCS selection for target BLER-SNR levels and enabling PHY abstraction for higher-layer simulations.