pith. sign in

arxiv: 2604.08437 · v1 · submitted 2026-04-09 · 💻 cs.IT · math.IT

Power Amplifier-aware Power Allocation for Noise-limited and Distortion-limited Regimes

Achref Tellili , Nathaniel Paul Epperson , Mohamed Akrout This is my paper

Pith reviewed 2026-05-10 17:01 UTC · model grok-4.3

classification 💻 cs.IT math.IT
keywords power allocationpower amplifier nonlinearityBussgang theoremcapacity optimizationdistortion-limited regimewater-fillingsaturation regimeprojected gradient descent
0
0 comments X

The pith

Power allocation that accounts for amplifier nonlinearity outperforms water-filling in saturation regimes.

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

This paper removes the assumption that power amplifiers operate linearly and instead folds their nonlinear behavior directly into the power allocation problem. It applies the Bussgang theorem to replace the hard-limiting amplifier with an equivalent linear gain plus uncorrelated distortion noise whose variance grows with input power. From this model the authors obtain both a projected gradient descent optimizer that jointly tunes power levels and spatial back-off, and a closed-form threshold on thermal noise variance that cleanly separates noise-limited from distortion-limited operation. A reader should care because conventional water-filling wastes performance once amplifiers enter saturation; the new allocation extracts extra capacity precisely where the linear model breaks down.

Core claim

By statistically linearizing the power amplifier via the Bussgang theorem, the power-allocation problem becomes an optimization over both signal gain and power-dependent distortion; the resulting projected gradient descent procedure yields higher mutual information than water-filling, while a simple expression involving thermal noise variance, distortion variance, and the channel Frobenius norm marks the boundary between the two operating regimes.

What carries the argument

Bussgang decomposition of the hard-limiting amplifier into a power-dependent linear gain plus additive uncorrelated distortion, optimized by projected gradient descent and delimited by a closed-form thermal-noise threshold.

If this is right

  • Capacity gains appear once the amplifier enters the saturation regime.
  • A single closed-form threshold on thermal noise variance identifies the switch between noise-limited and distortion-limited strategies.
  • Optimal spatial back-off can be found jointly with the power vector.
  • The linear-RF-chain assumption is no longer required for power-allocation design.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The regime threshold could be embedded in real-time adaptive algorithms that monitor noise and channel strength.
  • The same Bussgang linearization might be applied to other nonlinear RF components such as mixers or low-noise amplifiers.
  • Hardware validation would test whether the statistical model remains accurate for the specific waveforms used in modern standards.

Load-bearing premise

The Bussgang theorem supplies an accurate statistical description of the amplifier's hard limiter that correctly trades signal gain against power-dependent distortion for the subsequent optimization.

What would settle it

A hardware experiment that measures achievable rate under the proposed allocation versus standard water-filling while the amplifier is driven into saturation; absence of measurable gains or mismatch between predicted and observed distortion statistics would falsify the claim.

Figures

Figures reproduced from arXiv: 2604.08437 by Achref Tellili, Mohamed Akrout, Nathaniel Paul Epperson.

Figure 1
Figure 1. Figure 1: Linear gain α and distortion variance σ 2 η as a function of transmit power PT for varied supply voltages VCC. The vertical dotted lines indicate the theoretical saturation power thresholds Psat = (VCC/G) 2 between the linear and nonlinear operational regimes. III. CAPACITY ANALYSIS AND NOISE-DISTORTION REGIMES In a MIMO transmitter with NT antennas, each RF chain is equipped with an independent power ampl… view at source ↗
Figure 3
Figure 3. Figure 3: Capacity versus thermal noise variance σ 2 n (dBm) showing the transition from distortion-limited to noise-limited regimes. 20 40 60 80 100 120 140 160 180 200 Time slots 0 5 10 15 20 25 30 35 Channel Rank 101 102 103 Capacity Channel Rank Standard Waterfilling Amplifier-aware Waterfilling [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Evolution of channel rank and capacity over 200 time s [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
read the original abstract

The conventional power allocation strategy via water-filling relies on the premise that the power amplifier (PA) operates sufficiently below saturation such that a linear RF chain model holds. This work integrates the PA nonlinearity directly into the power allocation formulation, thereby removing the linearity assumption altogether and enabling operation in regimes where distortion noise is non-negligible. Leveraging the Bussgang theorem, we establish a statistical linearization of the PA's hard-limiting model to characterize the trade-off between signal gain and power-dependent distortion. We propose a projected gradient descent algorithm that optimizes power allocation while identifying an optimal spatial back-off strategy. We also derive a closed-form thermal noise variance threshold that separates the noise-limited and distortion-limited operating regimes as a function of the distortion noise variance and the channel Frobenius norm. Numerical simulations validate that our amplifier-aware strategy provides significant capacity gains in the saturation regime compared to standard water-filling.

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

3 major / 2 minor

Summary. The paper integrates power amplifier (PA) nonlinearity into MIMO power allocation by applying the Bussgang theorem to statistically linearize a hard-limiting PA model, trading off signal gain against power-dependent distortion. It proposes a projected gradient descent algorithm to optimize the allocation (including spatial back-off) and derives a closed-form threshold on thermal noise variance, expressed in terms of distortion variance and channel Frobenius norm, that separates noise-limited from distortion-limited regimes. Numerical simulations are reported to demonstrate significant capacity gains over conventional water-filling when operating in the PA saturation regime.

Significance. If the Bussgang-based gains survive end-to-end validation under the exact nonlinear PA, the work would be significant for practical high-power MIMO systems: it removes the low-power linearity assumption that underpins standard water-filling and supplies both an algorithmic solution and an analytical regime boundary. The closed-form threshold is a concrete, usable contribution that could guide system design without requiring repeated optimization.

major comments (3)
  1. [Abstract / Numerical Simulations] Abstract and Numerical Simulations section: the reported capacity gains are obtained under the Bussgang-linearized model; the manuscript does not present mutual-information or rate results when the optimized power vector is passed through the exact hard-clipping nonlinearity. Because the distortion term is no longer guaranteed to be Gaussian or uncorrelated after re-allocation, it is unclear whether the claimed gains persist in the physical model that the paper ultimately targets.
  2. [Threshold derivation / Optimization section] Derivation of the closed-form threshold (presumably in the section following the Bussgang linearization): the threshold is stated to separate noise-limited and distortion-limited regimes as a function of distortion variance and channel Frobenius norm. However, the optimization itself is performed via projected gradient descent on the linearized objective; it is not shown that the same threshold remains optimal or even meaningful once the true nonlinear PA is substituted.
  3. [System Model / Bussgang Linearization] Model section (Bussgang application): the linearization replaces the PA with an equivalent gain plus additive distortion whose variance depends on input power. The subsequent capacity expression assumes this distortion remains additive and independent of the signal after power re-allocation; no analytic or numerical check is provided that this independence holds for the non-uniform power allocations produced by the algorithm.
minor comments (2)
  1. [Notation / Equations] Notation for the distortion variance and the back-off factor should be introduced once and used consistently; several equations appear to reuse symbols without re-definition.
  2. [Algorithm description] The projected gradient descent algorithm description would benefit from explicit step-size selection rule and a convergence criterion; currently only the update form is given.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We are grateful to the referee for the thorough review and valuable feedback, which will help improve the manuscript. We provide point-by-point responses to the major comments below.

read point-by-point responses

  1. Referee: [Abstract / Numerical Simulations] Abstract and Numerical Simulations section: the reported capacity gains are obtained under the Bussgang-linearized model; the manuscript does not present mutual-information or rate results when the optimized power vector is passed through the exact hard-clipping nonlinearity. Because the distortion term is no longer guaranteed to be Gaussian or uncorrelated after re-allocation, it is unclear whether the claimed gains persist in the physical model that the paper ultimately targets.

    Authors: We thank the referee for highlighting this important point. The Bussgang linearization provides a statistically equivalent model for the hard-limiting PA under Gaussian inputs, where the output is represented as a scaled input plus uncorrelated distortion. Our capacity expressions and optimization are derived within this framework, which is standard in the literature for analyzing nonlinear PAs. However, to directly address the concern, we will include additional simulations in the revised manuscript that apply the optimized power allocations to the exact hard-clipping nonlinearity and compute the achievable rates, for example using Monte Carlo estimation of mutual information. This will validate whether the gains persist under the true model. revision: yes

  2. Referee: [Threshold derivation / Optimization section] Derivation of the closed-form threshold (presumably in the section following the Bussgang linearization): the threshold is stated to separate noise-limited and distortion-limited regimes as a function of distortion variance and channel Frobenius norm. However, the optimization itself is performed via projected gradient descent on the linearized objective; it is not shown that the same threshold remains optimal or even meaningful once the true nonlinear PA is substituted.

    Authors: The closed-form threshold is derived analytically from the linearized model to identify when the distortion variance dominates the thermal noise in the capacity expression. Since both the optimization and the regime separation are based on the same Bussgang-linearized model, the threshold is meaningful and optimal within the scope of our analysis. We agree that its applicability to the exact nonlinear model is not explicitly verified. In the revision, we will add a discussion noting that the threshold serves as a design guideline under the approximation and will include numerical checks comparing the regimes under both models. revision: partial

  3. Referee: [System Model / Bussgang Linearization] Model section (Bussgang application): the linearization replaces the PA with an equivalent gain plus additive distortion whose variance depends on input power. The subsequent capacity expression assumes this distortion remains additive and independent of the signal after power re-allocation; no analytic or numerical check is provided that this independence holds for the non-uniform power allocations produced by the algorithm.

    Authors: The Bussgang theorem ensures that for each individual PA, the distortion component is uncorrelated with the input signal to that PA, regardless of the input power level. Since our model applies a separate PA per transmit antenna with per-antenna power allocation, the distortion at each antenna remains uncorrelated with its own input. The overall received signal can thus be expressed as a linear transformation of the input plus additive distortion noise with a diagonal covariance matrix determined by the per-antenna distortion variances. We will add a clarification in the system model section to explicitly state this property and provide a brief proof or reference to the theorem's applicability to non-uniform allocations. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation uses external Bussgang theorem and derives threshold from linearized expressions without fitting or self-referential reduction.

full rationale

The paper invokes the established Bussgang theorem (an external statistical result for Gaussian inputs through memoryless nonlinearities) to obtain a linear gain plus uncorrelated distortion model. From this, it formulates an optimization problem solved by projected gradient descent and derives a closed-form noise/distortion threshold expressed directly in terms of the distortion variance and channel Frobenius norm. Neither the threshold nor the allocation rule is obtained by fitting to the optimizer's output or by renaming the input; the numerical simulations serve as external validation of the resulting policy rather than as a definitional step. No self-citation chains or ansatz smuggling appear in the load-bearing steps, so the claimed capacity gains are not tautological with the inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the applicability of the Bussgang theorem to the hard-limiting PA model and on the ability of projected gradient descent to locate a useful operating point for the resulting non-convex problem.

axioms (1)
  • standard math Bussgang theorem supplies a statistically equivalent linear model (gain plus uncorrelated distortion) for a memoryless nonlinearity driven by Gaussian inputs
    Invoked to replace the hard limiter with a linear gain plus power-dependent distortion noise

pith-pipeline@v0.9.0 · 5453 in / 1198 out tokens · 54175 ms · 2026-05-10T17:01:11.415985+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

9 extracted references · 9 canonical work pages

  1. [1]

    Holma and A

    H. Holma and A. Toskala, LTE for UMTS: Evolution to LTE-advanced . John Wiley & Sons, 2011

    work page 2011

  2. [2]

    Schenk, RF imperfections in high-rate wireless systems: impact and digital compensation

    T. Schenk, RF imperfections in high-rate wireless systems: impact and digital compensation. Springer, 2008

    work page 2008

  3. [3]

    M assive mimo systems with non-ideal hardware: Energy efficiency, estima tion, and capacity limits,

    E. Bj¨ ornson, J. Hoydis, M. Kountouris, and M. Debbah, “M assive mimo systems with non-ideal hardware: Energy efficiency, estima tion, and capacity limits,” IEEE Transactions on information theory, vol. 60, no. 11, pp. 7112–7139, 2014

    work page 2014

  4. [4]

    R. W. Heath Jr and A. Lozano, F oundations of MIMO communication . Cambridge University Press, 2018

    work page 2018

  5. [5]

    Crosscorrelation functions of amplitu de-distorted gaus- sian signals,

    J. J. Bussgang, “Crosscorrelation functions of amplitu de-distorted gaus- sian signals,” 1952

    work page 1952

  6. [6]

    P erformance limits of mimo systems with nonlinear power amplifiers,

    M. Fozooni, M. Matthaiou, E. Bjornson, and T. Q. Duong, “P erformance limits of mimo systems with nonlinear power amplifiers,” in 2015 IEEE Global Communications Conference (GLOBECOM) . IEEE, 2015, pp. 1–7

    work page 2015

  7. [7]

    Circuit aware de sign of power-efficient short range communication systems,

    A. Mezghani, N. Damak, and J. A. Nossek, “Circuit aware de sign of power-efficient short range communication systems,” in 2010 7th International Symposium on Wireless Communication System s. IEEE, 2010, pp. 869–873

    work page 2010

  8. [8]

    Analysis and compensation of power am plifier nonlinearity in mimo transmit diversity systems,

    J. Qi and S. Aissa, “Analysis and compensation of power am plifier nonlinearity in mimo transmit diversity systems,” IEEE Transactions on V ehicular Technology, vol. 59, no. 6, pp. 2921–2931, 2010

    work page 2010

  9. [9]

    Theoretical analysis and p erformance of ofdm signals in nonlinear awgn channels,

    P . Banelli and S. Cacopardi, “Theoretical analysis and p erformance of ofdm signals in nonlinear awgn channels,” IEEE Transactions on Communications, vol. 48, no. 3, pp. 430–441, 2002

    work page 2002