Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Nonlinear Power Amplifier-Resilient Cell-Free Massive MIMO: A Joint Optimization Approach

Wei Jiang, Hans D. Schotten

TL;DR

This work tackles the adverse impact of nonlinear power amplifiers in cell-free massive MIMO downlinks by introducing a Bussgang-based PA distortion model and a unified SE expression applicable to arbitrary linear precoding. It then develops a joint optimization of user association and max-min power allocation, reformulated as a rotated second-order cone program and solved via a bisection-based algorithm. To scale to larger networks, a low-complexity two-stage method fixes associations and optimizes power within active links, achieving substantial fairness and throughput gains over conventional baselines. The results demonstrate PA-resilient performance improvements across MR, ZF, RZF, and MMSE precoding, with significant gains in 95th-percentile SE and manageable computational costs. Overall, the paper provides a practical framework for mitigating PA nonlinearities in CF-mMIMO deployments.

Abstract

This letter analyzes the effects of power amplifiers (PAs) on the downlink of cell-free massive MIMO systems. We model signal transmission incorporating nonlinear PA distortion and derive a unified spectral efficiency (SE) expression applicable to arbitrary precoding schemes. To combat PA-induced performance degradation, a joint optimization approach for user association and max-min power control is proposed. Furthermore, a low-complexity alternative is developed to approximate the joint optimization with reduced computational overhead. Simulations validate the analysis and demonstrate significant performance gains of the proposed approaches over conventional techniques.

Nonlinear Power Amplifier-Resilient Cell-Free Massive MIMO: A Joint Optimization Approach

TL;DR

This work tackles the adverse impact of nonlinear power amplifiers in cell-free massive MIMO downlinks by introducing a Bussgang-based PA distortion model and a unified SE expression applicable to arbitrary linear precoding. It then develops a joint optimization of user association and max-min power allocation, reformulated as a rotated second-order cone program and solved via a bisection-based algorithm. To scale to larger networks, a low-complexity two-stage method fixes associations and optimizes power within active links, achieving substantial fairness and throughput gains over conventional baselines. The results demonstrate PA-resilient performance improvements across MR, ZF, RZF, and MMSE precoding, with significant gains in 95th-percentile SE and manageable computational costs. Overall, the paper provides a practical framework for mitigating PA nonlinearities in CF-mMIMO deployments.

Abstract

This letter analyzes the effects of power amplifiers (PAs) on the downlink of cell-free massive MIMO systems. We model signal transmission incorporating nonlinear PA distortion and derive a unified spectral efficiency (SE) expression applicable to arbitrary precoding schemes. To combat PA-induced performance degradation, a joint optimization approach for user association and max-min power control is proposed. Furthermore, a low-complexity alternative is developed to approximate the joint optimization with reduced computational overhead. Simulations validate the analysis and demonstrate significant performance gains of the proposed approaches over conventional techniques.

Paper Structure

This paper contains 10 sections, 2 theorems, 23 equations, 2 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Model
  3. Downlink Transmission
  4. Nonlinear Power Amplification
  5. Precoding Schemes
  6. Performance Analysis
  7. Joint Optimization of User Association and Max-Min Power Allocation
  8. Computational Cost and Low-Complexity Alternative
  9. Simulation Results and Discussions
  10. Conclusion

Key Result

Proposition 1

The achievable SE of user $k$ is $R_k=\mathbb{E} \Bigl[\log_2(1+\gamma_k) \Bigr]$, where the expectation is over channel realizations and the instantaneous effective SINR is

Figures (2)

  • Figure 1: The CDF of achievable spectral efficiency under MR, RZF, and MMSE precoding. User fairness (worst-case performance) is represented by the 5th percentile of each curve. All curves use the same marker styles as in Fig. 1a.
  • Figure 2: Comparison of computational complexity.

Theorems & Definitions (3)

  • Proposition 1
  • Remark 1
  • Lemma 1