pith. sign in

arxiv: 2512.23381 · v2 · submitted 2025-12-29 · 📡 eess.SP

On Signal Peak Power Constraint of Over-the-Air Federated Learning

Lorenz Bielefeld , Paul Zheng , Oner Hanay , Yao Zhu , Yulin Hu , Anke Schmeink This is my paper

Pith reviewed 2026-05-16 19:40 UTC · model grok-4.3

classification 📡 eess.SP
keywords over-the-air computationfederated learningpeak power constraintpower amplifierclipping and filteringOFDMsignal distortionAirComp-FL
0
0 comments X

The pith

Over-the-air federated learning often violates power amplifier limits, causing performance loss when corrected by clipping.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper examines the practical issue of instantaneous peak power constraints in over-the-air computation for federated learning. It shows that the transmitted signals frequently exceed the linear range of power amplifiers. The authors apply iterative clipping and filtering to enforce these limits and demonstrate resulting degradations in learning accuracy. The problem is more severe in multi-carrier OFDM systems because of added in-band distortions. This highlights a gap between theoretical AirComp-FL designs and real hardware constraints.

Core claim

The central claim is that existing AirComp-FL overlooks instantaneous peak-power constraints from non-linear power amplifiers. Operating beyond linearity causes distortions, and the default mitigation of iterative amplitude clipping with filtering leads to FL performance degradation. This degradation is more pronounced in multi-carrier OFDM systems due to in-band effects from clipping and filtering. Simulations confirm that transmit power regularly exceeds limits in practical settings.

What carries the argument

Iterative amplitude clipping combined with filtering applied to the superimposed wireless signals in AirComp-FL to enforce peak power limits.

Load-bearing premise

The iterative clipping and filtering method used is representative of practical mitigation techniques, and the channel and amplifier models capture the main real-world effects.

What would settle it

An experiment measuring the actual model accuracy in a wireless testbed with real power amplifiers and AirComp transmission, comparing clipped versus ideal linear cases.

Figures

Figures reproduced from arXiv: 2512.23381 by Anke Schmeink, Lorenz Bielefeld, Oner Hanay, Paul Zheng, Yao Zhu, Yulin Hu.

Figure 1
Figure 1. Figure 1: Typical input and output power characteristics curve for a power [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Flow Diagram for ICF in multi-carrier OFDM systems. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 5
Figure 5. Figure 5: Test Accuracy w.r.t. communication rounds. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Loss of test accuracy by imposing the peak power constraint (after [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
read the original abstract

Federated learning (FL) has been considered a promising privacy preserving distributed edge learning framework. Over-the-air computation (AirComp) leveraging analog transmission enables the aggregation of local updates directly over-the-air by exploiting the superposition properties of wireless multiple-access channels, thereby alleviating the communication bottleneck issues of FL compared with digital transmission schemes. This work points out that existing AirComp-FL overlooks a key practical constraint, the instantaneous peak-power constraints due to the non-linearity of radio-frequency power amplifiers. Operating directly in non-linear region causes in-band and out-of-band distortions. We present and analyze the effect of the default method that limits the signal's peak power and out-of-band distortions, iterative amplitude clipping combined with filtering. We investigate the effect of imposing instantaneous peak-power constraints in AirComp-FL for both single-carrier and multi-carrier orthogonal frequency-division multiplexing (OFDM) systems. Simulation results demonstrate that, in practical settings, the instantaneous transmit power in AirComp-FL regularly exceeds the power-amplifier linearity limit. As the first work of this line of research, it is essential to evaluate if this is an actual problem that has an impact on FL performance. We therefore apply the classic method of iterative clipping and filtering, and show that the FL performance degrades more or less depending on the scenarios. The degradation becomes pronounced especially in multi-carrier OFDM systems due to the in-band distortions caused by clipping and filtering.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript claims that existing AirComp-FL literature overlooks instantaneous peak-power constraints arising from RF power-amplifier non-linearity. Through simulations of both single-carrier and multi-carrier OFDM systems, it shows that the transmit signal regularly exceeds the amplifier linearity region in practical settings; applying the standard iterative clipping-and-filtering procedure to enforce the constraint produces FL performance degradation that is especially severe in OFDM due to in-band distortion of the superimposed gradient.

Significance. If the simulation results are representative, the work is the first to quantify a hardware-level limitation that can materially affect convergence and accuracy in AirComp-FL. It supplies concrete evidence that peak-power violations are not rare and that conventional clipping mitigation is costly, thereby motivating the design of AirComp-aware PAPR-reduction techniques.

major comments (2)
  1. [Simulation Results] The central performance-degradation claim rests exclusively on the iterative clipping-and-filtering mitigation (described in the simulation section). No comparison is provided against AirComp-specific alternatives such as per-subcarrier power scaling, tone reservation, or optimized precoding that enforce the same peak-power limit without explicit amplitude clipping of the transmitted signals. Because the received aggregate in AirComp is the superposition of the clipped waveforms, it is unclear whether the reported FL degradation is fundamental to the hardware constraint or an artifact of the chosen mitigation.
  2. [Simulation Results] The manuscript states that degradation 'becomes pronounced especially in multi-carrier OFDM systems due to the in-band distortions caused by clipping and filtering,' yet provides no quantitative breakdown (e.g., EVM or gradient distortion metrics before/after clipping) that isolates the contribution of in-band versus out-of-band effects to the observed FL accuracy loss.
minor comments (2)
  1. [Simulation Setup] Clarify the exact PAPR threshold, clipping ratio, and filter parameters used in the iterative clipping-and-filtering procedure so that the experiments can be reproduced.
  2. [Introduction] Add a brief discussion of how the peak-power constraint interacts with the analog superposition property that is the core advantage of AirComp.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. The feedback correctly identifies areas where our simulation analysis can be strengthened. We address each major comment below and indicate planned revisions.

read point-by-point responses

  1. Referee: The central performance-degradation claim rests exclusively on the iterative clipping-and-filtering mitigation. No comparison is provided against AirComp-specific alternatives such as per-subcarrier power scaling, tone reservation, or optimized precoding that enforce the same peak-power limit without explicit amplitude clipping.

    Authors: We chose iterative clipping-and-filtering as it is the standard, widely referenced method for enforcing peak-power limits while suppressing out-of-band emissions. The manuscript's purpose is to demonstrate that this conventional approach already produces noticeable FL degradation in practical AirComp settings, thereby motivating the need for better techniques. We agree that direct comparisons to AirComp-aware methods (per-subcarrier scaling, tone reservation, optimized precoding) would clarify whether the observed loss is fundamental to the hardware constraint. In the revision we will add a dedicated discussion subsection on these alternatives and their potential to reduce in-band distortion under superposition, although a full comparative simulation campaign lies outside the scope of this initial problem-identification study. revision: partial

  2. Referee: The manuscript states that degradation becomes pronounced especially in multi-carrier OFDM systems due to the in-band distortions caused by clipping and filtering, yet provides no quantitative breakdown such as EVM or gradient distortion metrics before/after clipping to isolate in-band versus out-of-band effects.

    Authors: We acknowledge the absence of explicit quantitative isolation. While the text attributes the OFDM degradation to in-band distortion, we did not report supporting metrics such as EVM or the norm of the distortion in the received aggregate gradient. In the revised manuscript we will include new simulation results (additional figures/tables) that report EVM and gradient-distortion values before and after clipping/filtering for both single-carrier and OFDM cases, thereby separating the in-band and out-of-band contributions to the observed accuracy loss. revision: yes

Circularity Check

0 steps flagged

No circularity: simulation study applies external clipping method

full rationale

The paper is a simulation study that applies the established iterative amplitude clipping-and-filtering procedure (described as the 'classic method' and 'default method') to evaluate peak-power effects in AirComp-FL for single-carrier and OFDM systems. No load-bearing derivations, fitted parameters renamed as predictions, or self-citation chains appear in the provided text. The central claims rest on simulation outcomes using standard external techniques rather than any self-referential construction, so the analysis is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper relies on standard wireless channel models, power-amplifier non-linearity curves, and federated-learning aggregation rules taken from prior literature; no new free parameters, axioms, or invented entities are introduced.

pith-pipeline@v0.9.0 · 5565 in / 1190 out tokens · 26819 ms · 2026-05-16T19:40:23.202128+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages · 1 internal anchor

  1. [1]

    Communication-efficient learning of deep networks from decentralized data,

    H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. Y . Arcas, “Communication-efficient learning of deep networks from decentralized data,” inAISTATS, Fort Lauderdale, Florida, USA, 2017

    work page 2017

  2. [2]

    Federated Learning: Strategies for Improving Communication Efficiency

    J. Kone ˇcn´y, H. B. McMahan, F. X. Yu, P. Richt ´arik, A. T. Suresh, and D. Bacon, “Federated learning: Strategies for improving communication efficiency,” 2017, arXiv:1610.05492

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  3. [3]

    Federated learning in mobile edge networks: A comprehensive survey,

    W. Y . B. Lim, N. C. Luong, D. T. Hoang, Y . Jiao, Y .-C. Liang, Q. Yang, D. Niyato, and C. Miao, “Federated learning in mobile edge networks: A comprehensive survey,”IEEE Commun. Surv. & Tut., vol. 22, no. 3, pp. 2031–2063, 2020

    work page 2031

  4. [4]

    Computation over multiple-access channels,

    B. Nazer and M. Gastpar, “Computation over multiple-access channels,” IEEE Trans. Inf. Theory, vol. 53, no. 10, pp. 3498–3516, 2007

    work page 2007

  5. [5]

    Robust analog function computation via wireless multiple-access channels,

    M. Goldenbaum and S. Sta ´nczak, “Robust analog function computation via wireless multiple-access channels,”IEEE Trans. Commun., vol. 61, no. 9, pp. 3863–3877, 2013

    work page 2013

  6. [6]

    Federated learning via over- the-air computation,

    K. Yang, T. Jiang, Y . Shi, and Z. Ding, “Federated learning via over- the-air computation,”IEEE Trans. Wireless Commun., vol. 19, no. 3, pp. 2022–2035, 2020

    work page 2022

  7. [7]

    Machine learning at the wireless edge: Distributed stochastic gradient descent over-the-air,

    M. M. Amiri and D. G ¨und¨uz, “Machine learning at the wireless edge: Distributed stochastic gradient descent over-the-air,”IEEE Trans. Signal Process., vol. 68, pp. 2155–2169, 2020. (a) Single-carrier (b) Multi-carrier. Fig. 5. Test Accuracy w.r.t. communication rounds. (a) Single-carrier Communication Round t 0 100 200 300 400 500 Loss of Test Accuracy ...

    work page 2020

  8. [8]

    Broadband analog aggregation for low- latency federated edge learning,

    G. Zhu, Y . Wang, and K. Huang, “Broadband analog aggregation for low- latency federated edge learning,”IEEE Trans. Wireless Commun., vol. 19, no. 1, pp. 491–506, 2020

    work page 2020

  9. [9]

    Gradient statistics aware power control for over- the-air federated learning,

    N. Zhang and M. Tao, “Gradient statistics aware power control for over- the-air federated learning,”IEEE Trans. Wireless Commun., vol. 20, no. 8, pp. 5115–5128, 2021

    work page 2021

  10. [10]

    Transmission power control for over-the-air federated averaging at network edge,

    X. Cao, G. Zhu, J. Xu, and S. Cui, “Transmission power control for over-the-air federated averaging at network edge,”IEEE J. Sel. Areas Commun., vol. 40, no. 5, pp. 1571–1586, 2022

    work page 2022

  11. [11]

    Optimized power control design for over-the-air federated edge learning,

    X. Cao, G. Zhu, J. Xu, Z. Wang, and S. Cui, “Optimized power control design for over-the-air federated edge learning,”IEEE J. Sel. Areas Commun., vol. 40, no. 1, pp. 342–358, 2022

    work page 2022

  12. [12]

    Over-the-Air Computing in OFDM Systems,

    N. G. Evgenidis, S. A. Tegos, P. D. Diamantoulakis, and G. K. Kara- giannidis, “Over-the-Air Computing in OFDM Systems,”IEEE Commun. Lett., vol. 28, no. 11, pp. 2523–2527, Nov. 2024

    work page 2024

  13. [13]

    Over-the-air computation in OFDM systems with imperfect channel state information,

    Y . Chen, H. Xing, J. Xu, L. Xu, and S. Cui, “Over-the-air computation in OFDM systems with imperfect channel state information,”IEEE Trans. Commun., vol. 72, no. 5, pp. 2929–2944, 2024

    work page 2024

  14. [14]

    Optimal power control and CSI acquisition for over-the-air computation in OFDM system,

    X. Xie, C. Hua, J. Hong, and W. Xu, “Optimal power control and CSI acquisition for over-the-air computation in OFDM system,”IEEE Trans. Wireless Commun., vol. 23, no. 6, pp. 6533–6545, Jun. 2024

    work page 2024

  15. [15]

    Federated learning with integrated over-the-air computation and sensing in IRS-assisted networks,

    P. Zheng, Y . Zhu, M. Bouchaala, Y . Hu, S. Stanczak, and A. Schmeink, “Federated learning with integrated over-the-air computation and sensing in IRS-assisted networks,” inWSA, Braunschweig, Germany, 2023

    work page 2023

  16. [16]

    Optimized power control for over- the-air computation in fading channels,

    X. Cao, G. Zhu, J. Xu, and K. Huang, “Optimized power control for over- the-air computation in fading channels,”IEEE Transactions on Wireless Communications, vol. 19, no. 11, pp. 7498–7513, 2020

    work page 2020

  17. [17]

    Joint subcarrier and power allocation in uplink OFDMA systems,

    K. Kim, Y . Han, and S.-L. Kim, “Joint subcarrier and power allocation in uplink OFDMA systems,”IEEE Commun. Lett., vol. 9, no. 6, pp. 526– 528, 2005

    work page 2005

  18. [18]

    Low complexity scheduling algorithms for the LTE uplink,

    E. Yaacoub, H. Al-Asadi, and Z. Dawy, “Low complexity scheduling algorithms for the LTE uplink,” inIEEE Symp. Comp. and Commun., 2009, pp. 266–270

    work page 2009

  19. [19]

    An overview of peak-to-average power ratio reduction techniques for multicarrier transmission,

    S. H. Han and J. H. Lee, “An overview of peak-to-average power ratio reduction techniques for multicarrier transmission,”IEEE Wireless Commun., vol. 12, no. 2, pp. 56–65, 2005

    work page 2005

  20. [20]

    Peak-to-average power ratio reduction in OFDM systems: A survey and taxonomy,

    Y . Rahmatallah and S. Mohan, “Peak-to-average power ratio reduction in OFDM systems: A survey and taxonomy,”IEEE Commun. Surv. & Tut., vol. 15, no. 4, pp. 1567–1592, 2013

    work page 2013

  21. [21]

    Computation of the continuous-time PAR of an OFDM signal with BPSK subcarriers,

    C. Tellambura, “Computation of the continuous-time PAR of an OFDM signal with BPSK subcarriers,”IEEE Commun. Lett., vol. 5, no. 5, pp. 185–187, 2001

    work page 2001

  22. [22]

    Peak-to-average power reduction for OFDM by repeated clipping and frequency domain filtering,

    J. Armstrong, “Peak-to-average power reduction for OFDM by repeated clipping and frequency domain filtering,”Electronics Letters, vol. 38, pp. 246–247, 2002

    work page 2002