Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distributed vs. Centralized Precoding in Cell-Free Systems: Impact of Realistic Per-AP Power Limits

Wei Jiang, Hans D. Schotten

TL;DR

When two simple heuristics are applied to enforce the per-AP instantaneous power constraint, the centralized performance superiority disappears, making distributed precoding a robust option.

Abstract

In cell-free massive MIMO, centralized precoding is {theoretically known} to {remarkably} outperform its distributed counterparts, albeit {with} high implementation complexity. However, this letter highlights a practical limitation {often overlooked:} {widely used closed-form} centralized {precoders} are typically derived under a sum-power constraint, which often demands unrealistic power allocation that exceeds hardware capabilities. {When two simple heuristics (global power scaling and local normalization) are applied to enforce the per-AP instantaneous power constraint}, the centralized performance superiority disappears, making distributed precoding {a robust option}.

Distributed vs. Centralized Precoding in Cell-Free Systems: Impact of Realistic Per-AP Power Limits

TL;DR

When two simple heuristics are applied to enforce the per-AP instantaneous power constraint, the centralized performance superiority disappears, making distributed precoding a robust option.

Abstract

In cell-free massive MIMO, centralized precoding is {theoretically known} to {remarkably} outperform its distributed counterparts, albeit {with} high implementation complexity. However, this letter highlights a practical limitation {often overlooked:} {widely used closed-form} centralized {precoders} are typically derived under a sum-power constraint, which often demands unrealistic power allocation that exceeds hardware capabilities. {When two simple heuristics (global power scaling and local normalization) are applied to enforce the per-AP instantaneous power constraint}, the centralized performance superiority disappears, making distributed precoding {a robust option}.
Paper Structure (15 sections, 2 theorems, 17 equations, 2 figures, 1 table)

This paper contains 15 sections, 2 theorems, 17 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. System Model
  3. Downlink Transmission
  4. Distributed Strategy
  5. Centralized Strategy
  6. Precoder Design under Power Constraints
  7. Unconstrained Precoder Formulation
  8. Power-Constrained Normalization
  9. Distributed Per-AP Normalization
  10. Centralized Sum-Power Normalization
  11. Per-AP Constrained Centralized Precoding
  12. Per-AP Constrained Power Scaling (PS)
  13. Per-AP Constrained Precoder Local Normalization (LN)
  14. Simulation Results and Discussions
  15. Conclusion

Key Result

Proposition 1

An achievable downlink SE for user $k$ under distributed precoding is given by $R_k = \mathbb{E} \left[ \log_2 \left( 1 + \gamma_k \right) \right]$, where the instantaneous effective signal-to-interference-plus-noise ratio (SINR) is

Figures (2)

  • Figure 1: Illustration of the power concentration effect: distribution of normalized transmit power across APs for a CF system with $50$ single-antenna APs and $10$ users, using centralized ZF precoder with suboptimal power allocation of $\epsilon_1=\ldots=\epsilon_K$ (following Eq. (21) in Ref_nayebi2017precoding). The horizontal (red) dashed line marks the normalized per-AP power constraint of $1/L = 0.02$.
  • Figure 2: Performance comparison between distributed and centralized precoding for (a) MR, (b) RZF, and (c) MMSE, under both equal-power and max–min power optimization. All three subfigures use the same markers as in (a) to denote the different options.

Theorems & Definitions (3)

  • Proposition 1
  • Proposition 2
  • Remark 1