Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Effect of Realistic Oscillator Phase Noise on the Performance of Cell-Free Massive MIMO Systems

Igor Zhilin, Evgenii Vinogradov, Ian Akyildiz

TL;DR

This work investigates the impact of phase noise on cell-free massive MIMO systems operating under sub-6 GHz 5G NR conditions. It advances a hardware-inspired PN model that captures LO drift in practical SDRs and PLL-based architectures, moving beyond simplistic Wiener-process models. Through centralized CF processing and FR1 numerology with a 1 ms TTI, the study demonstrates that realistic PN causes negligible uplink SE degradation even with low-cost LOs, highlighting the practicality of CF mMIMO in current 5G deployments and potential 6G integration. The proposed PN model is broadly applicable to RF systems in the sub-6 GHz band and can inform numerology and pilot allocation strategies for robust coherent CF operation.

Abstract

As the demand for 6G technologies continues to grow, the radio access infrastructure is expected to become increasingly dense. Cell-free (CF) Massive MIMO systems provide remarkable flexibility by enabling coherent service to users through multiple Access Points (APs). This innovative paradigm necessitates precise and stable phase synchronization. This paper examines the standardized 5G New Radio (NR) framework, focusing on subcarrier spacing, OFDM symbol duration, and allocation, while investigating the impact of Phase Noise (PN) on the performance of scalable massive MIMO cell-free systems. Unlike existing studies that typically employ a simplified model of a free-running oscillator characterized by a Wiener process, we present a realistic phase noise model inspired by actual hardware, designed to accurately capture the Local Oscillator (LO) phase drift. Furthermore, our PN model extends its applicability beyond cell-free systems, making it relevant for any RF system operating within the sub-6 GHz band. This model provides a robust foundation for the practical design of cell-free systems, encompassing numerology and pilot allocation strategies. Our findings reveal that even cost-effective low-cost Local Oscillators can achieve sufficient stability, resulting in negligible degradation of uplink Spectral Efficiency (SE) within the standardized 5G Transmission Time Interval of 1 ms. These results affirm the viability of cell-free massive MIMO systems based on 5G standards and their potential integration into future 6G networks.

Effect of Realistic Oscillator Phase Noise on the Performance of Cell-Free Massive MIMO Systems

TL;DR

This work investigates the impact of phase noise on cell-free massive MIMO systems operating under sub-6 GHz 5G NR conditions. It advances a hardware-inspired PN model that captures LO drift in practical SDRs and PLL-based architectures, moving beyond simplistic Wiener-process models. Through centralized CF processing and FR1 numerology with a 1 ms TTI, the study demonstrates that realistic PN causes negligible uplink SE degradation even with low-cost LOs, highlighting the practicality of CF mMIMO in current 5G deployments and potential 6G integration. The proposed PN model is broadly applicable to RF systems in the sub-6 GHz band and can inform numerology and pilot allocation strategies for robust coherent CF operation.

Abstract

As the demand for 6G technologies continues to grow, the radio access infrastructure is expected to become increasingly dense. Cell-free (CF) Massive MIMO systems provide remarkable flexibility by enabling coherent service to users through multiple Access Points (APs). This innovative paradigm necessitates precise and stable phase synchronization. This paper examines the standardized 5G New Radio (NR) framework, focusing on subcarrier spacing, OFDM symbol duration, and allocation, while investigating the impact of Phase Noise (PN) on the performance of scalable massive MIMO cell-free systems. Unlike existing studies that typically employ a simplified model of a free-running oscillator characterized by a Wiener process, we present a realistic phase noise model inspired by actual hardware, designed to accurately capture the Local Oscillator (LO) phase drift. Furthermore, our PN model extends its applicability beyond cell-free systems, making it relevant for any RF system operating within the sub-6 GHz band. This model provides a robust foundation for the practical design of cell-free systems, encompassing numerology and pilot allocation strategies. Our findings reveal that even cost-effective low-cost Local Oscillators can achieve sufficient stability, resulting in negligible degradation of uplink Spectral Efficiency (SE) within the standardized 5G Transmission Time Interval of 1 ms. These results affirm the viability of cell-free massive MIMO systems based on 5G standards and their potential integration into future 6G networks.
Paper Structure (22 sections, 16 equations, 6 figures, 1 table)

This paper contains 22 sections, 16 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. System Model
  3. The OFDM System Model
  4. The Phase Noise Model
  5. The OFDM Signal Model with Phase Noise
  6. The Channel Model
  7. Resource Element Mapping and Channel Estimation
  8. The Scalable Cell-Free system
  9. The CF Channel Model
  10. Scalable CF
  11. The Phase Noise in CF
  12. The Received Signal Model
  13. The Phase Noise Model
  14. Simplified PN Model: Free Running Oscillator
  15. Hardware-Inspired Phase Noise Model
  16. ...and 7 more sections

Figures (6)

  • Figure 1: Mapping of the Resource Elements within a coherence block.
  • Figure 2: CF massive MIMO system with user-centric clustering.
  • Figure 3: PSD of an ovenized crystal oscillator (OCXO) phase noise versus frequency offset with asymptots corresponding to different noise parts: phase white noise $~\sim f^{0}$, phase flicker noise $~\sim f^{-1}$, frequency white noise $~\sim f^{-2}$, frequency flicker noise $~\sim f^{-3}$, frequency random walk $~\sim f^{-4}$. Courtesy of Rohde&Schwarz RSAN2015AllanVar.
  • Figure 4: A single loop of PLL.
  • Figure 5: Comparison of the phase drift generated by different models: Realistic LOs are more stable.
  • ...and 1 more figures