Pith. sign in

REVIEW 4 major objections 5 minor 3 cited by

Affine Filter Bank Modulation: A New Waveform for High Mobility Communications

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims AFBM achieves AFDM-like quasi-orthogonality in doubly-dispersive channels while reporting PAPR 3 dB lower (2 dB in simulation) and OOBE down to -100 dB.

desk verdict AFBM is a sensible combination of DAFT precoding and DFT-spread FBMC, but the abstract's simultaneous claims are not supported because the best-OOBE configuration (PHYDYAS O=4) violates the paper's own O≤1.5 orthogonality-restoration condition and no BER curves are shown. read the letter →

arxiv 2505.03589 v1 pith:VT3ZBJO4 submitted 2025-05-06 eess.SP

Henrique L. Senger , Gustavo P. Gonçalves , Bruno S. Chang , Hyeon Seok Rou , Kuranage Roche Rayan Ranasinghe , Giuseppe Thadeu Freitas de Abreu , Didier Le Ruyet This is my paper · ORCID
classification eess.SP
keywords affinefilterbankmodulationAFBMdoubly-dispersivechannelspeak-to-averagepowerratioout-of-bandemissionsDAFTprecodingmulticarrierintegratedsensingandcommunications
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces Affine Filter Bank Modulation (AFBM), a waveform built by inserting a DAFT precoding stage into an FBMC filter bank, and argues that it combines properties no prior waveform in this line has delivered together: robustness to doubly-dispersive channels comparable to AFDM, peak-to-average power ratio several decibels lower (the abstract says 3 dB; the simulation figure shows about 2 dB), and out-of-band emissions as low as -100 dB when a PHYDYAS prototype filter with overlap factor 4 is used. A sympathetic reader would care because high-mobility and integrated-sensing links need exactly this combination: DD robustness for reliable transmission, low PAPR for efficient amplification, and low OOBE for spectral coexistence. The paper supports the claim with an effective-channel analysis showing that the delay-Doppler shifts of the channel map to deterministic diagonal shifts, as in AFDM, and with PAPR and OOBE simulations; it does not report BER curves.

What carries the argument

The load-bearing object is the compensation precoder $C_f = W_L \operatorname{diag}\{\tilde{b}\}$, an $L$-point discrete affine Fourier transform multiplied by a diagonal filter-compensation vector whose nonzero entries ($\tilde{b}_{\tilde{l}} = 1/\sqrt{\tilde{c}_{\tilde{l}}}$) are set only in the first and last $L/4$ positions, where the data actually live. This $C_f$ is what turns a conventional FBMC filter bank into a complex-orthogonality-preserving structure, because the DAFT spreads each data symbol over chirp subcarriers while the compensation cancels the interference introduced by the prototype filter. The second essential element is the truncated IDAFT plus frequency-domain zero padding (the $Q_P$ matrix) that keeps the chirp sampling below Nyquist and contains the spectrum. Together they make the effective channel $\mathbf{H}_{\mathrm{eff}}$ behave like AFDM's: each propagation path produces a deterministic diagonal shift determined by its delay and Doppler indices.

What would settle it

Simulate AFBM with the PHYDYAS $O=4$ filter over the paper's three-path doubly-dispersive channel, compare BER against AFDM with identical chirp parameters, and inspect the off-diagonal energy of $C_f^H Q_P^H \tilde{G}^T \tilde{G} Q_P C_f$. A multi-dB SNR loss or an error floor in that configuration would disprove the simultaneous claim, because the -100 dB OOBE result is only achieved at $O=4$.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that AFBM is the first waveform to simultaneously offer AFDM-like quasi-orthogonality in doubly-dispersive channels, a low PAPR close to that of DAFT-spread AFDM, and FBMC-grade spectral containment. The mechanism is a precoding matrix $C_f = W_L \operatorname{diag}\{\tilde{b}\}$ that combines an $L$-point DAFT with a filter compensation vector, followed by a truncated IDAFT, frequency-domain zero padding, and a block-Toeplitz prototype filter; the compensation is designed so that $C_f^H Q_P^H \tilde{G}^T \tilde{G} Q_P C_f \approx U$ holds, restoring complex orthogonality for overlap factors $O \le 1.5$. The effective channel $\mathbf{H}_{\mathrm{eff}} = Q_P^H G^H \left(\sum_{r} h_r \Phi_r Z^{f_r} \Pi^{\ell_r}\right) G Q_P$ retains the deterministic band-diagonal shift structure that gives AFDM its DD robustness. Simulation comparisons against AFDM report a PAPR advantage (about 2 dB in the figure, 3 dB in the abstract) and OOBE down to about -100 dB with the PHYDYAS filter at $O=4$.

Load-bearing premise

The central claim stands on the unquantified assumption that the signal-to-interference loss caused by using a high-overlap PHYDYAS filter ($O=4$) is small enough that AFBM keeps near-AFDM bit-error-rate performance; the paper guarantees orthogonality only for $O \le 1.5$ and gives no BER curves or SIR numbers.

Editorial extensions

If this is right

  • If AFBM delivers on its claims, ISAC systems can use one waveform for both communication and sensing in high-mobility channels without choosing between spectral containment and PAPR efficiency.
  • The $O \le 1.5$ compensation guarantee gives a concrete low-latency operating point with controlled self-interference, while larger overlap factors trade SIR for deeper OOBE suppression.
  • Because the effective channel preserves AFDM's deterministic path-dependent diagonal structure, existing AFDM channel estimation and equalization techniques can be adapted to AFBM rather than designed from scratch.
  • The low PAPR means high-power amplifiers need less back-off, so the good OOBE is less likely to be destroyed by nonlinear distortion in realistic transmitters.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper leaves open whether the -100 dB OOBE configuration and the quasi-orthogonality claim can coexist: the unquantified SIR loss at $O=4$ could be small enough to hide in BER figures, or large enough to split AFBM into two regimes, one for spectral containment and one for high-rate transmission.
  • One testable implication beyond the paper: comparing BER at $O=1.5$ vs $O=4$ on the same channel would quantify the SIR penalty and show whether an iterative receiver, which the authors list as future work, is necessary.
  • Because the effective channel's diagonal shifts encode delay and Doppler directly, AFBM's structure suggests radar parameter estimation similar to AFDM should be possible; the paper defers sensing evaluation, so this is a projection.
  • A natural extension the authors do not pursue is optimizing the chirp parameters of the DAFT stage for both PAPR and channel orthogonality simultaneously, which they mention as future work on PAPR specifically.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes Affine Filter Bank Modulation (AFBM), a waveform that combines a DAFT (chirp) precoding stage, a frequency-domain zero-padded IDAFT, and a filter-bank transmit structure, with a compensation vector intended to restore complex orthogonality. The authors claim that AFBM simultaneously provides quasi-orthogonality in doubly-dispersive channels comparable to AFDM, PAPR about 3 dB lower than AFDM, and OOBE as low as -100 dB when a PHYDYAS prototype filter is used. The manuscript develops a matrix model of the transceiver, derives a compensation condition, presents an effective-channel analysis for a single symbol, and gives simulation results for PAPR and OOBE. It does not provide BER results, despite announcing them in the introduction, and it does not quantify the signal-to-interference ratio degradation for high-overlap filters.

Significance. If the simultaneous claims were fully demonstrated, AFBM would be a meaningful contribution to high-mobility and ISAC waveform design, since the cited prior art (AFDM, DAFT-s-AFDM, FBMC) does not jointly achieve DD robustness, low PAPR, and low OOBE. The paper contributes a clear matrix formulation of the proposed transceiver, an explicit compensation construction, and initial PAPR/OOBE simulation evidence that the waveform inherits the expected filter-bank spectral containment and low-PAPR properties. However, the central quasi-orthogonality claim is currently supported only by structural plots and a K=1 analysis, with no detection-performance or interference-quantification evidence, so the significance is conditional on substantial additional validation.

major comments (4)
  1. [Section III / Abstract] The paper does not deliver the promised BER comparison. The introduction states that simulation results will compare AFBM against AFDM 'in terms of bit error rate (BER), PAPR, and OOBE,' but Section III contains only PAPR and OOBE figures. Since the abstract's central claim is that AFBM 'maintains quasi-orthogonality similar to that of AFDM' in doubly-dispersive channels, this claim is currently unsupported by any detection-performance result. The authors should add BER versus SNR curves for AFBM and AFDM over the doubly-dispersive channel, or explicitly remove the quasi-orthogonality claim.
  2. [Section II-B, Eqs. (11)-(13)] The compensation construction is only approximate and is guaranteed only for O <= 1.5, as stated immediately after Eq. (13). The O=4 PHYDYAS configuration used to obtain the -100 dB OOBE result in Fig. 4 falls exactly in the regime where the text says interference components outside the main diagonal reduce the SIR. Since no SIR or BER quantification is provided for O=4, the paper never demonstrates the simultaneous realization of -100 dB OOBE and quasi-orthogonality in a single configuration. The authors should quantify the SIR as a function of O and provide BER results for the O=4 case, or restrict the OOBE claim to configurations for which the orthogonality restoration is established.
  3. [Section II-C, Eq. (19)] The effective-channel analysis is restricted to K=1 and is said to be 'without loss of generality,' but the full transmit model in Eq. (8) uses a block-Toeplitz filter matrix G that introduces inter-symbol overlap for K>1. Setting K=1 removes all overlapping blocks, so the analysis cannot capture inter-symbol interference that the filter bank introduces in the actual K=8 simulation configuration. The quasi-orthogonality conclusion therefore rests on a structural plot rather than on a metric for the full block system. The authors should either extend the effective-channel analysis to K>1, provide a rigorous argument that K=1 is representative, or give a quantitative interference-energy metric for the block system.
  4. [Abstract and Fig. 3] The abstract claims PAPR levels '3 dB lower' than AFDM, but the simulation result in Fig. 3 reports 'an advantage of 2 dB with respect to regular AFDM.' These numbers should be reconciled. If the 3 dB figure refers to a different operating point or parameter set, that should be stated explicitly; otherwise the abstract overstates the simulation evidence.
minor comments (5)
  1. [Eq. (7)] The block-Toeplitz matrix display in Eq. (7) is garbled and appears to have missing entries and misplaced zeros; it should be redrawn with a clear definition of the blocks G_p and their positions.
  2. [Section I] The introduction announces a BER comparison, but no BER results appear in Section III; either add the comparison or correct the description of the simulation section.
  3. [Eq. (11)] The 'approximately equals' in Eq. (11) should be replaced by a precise design criterion or an explicit equality under the stated conditions, since the subsequent compensation derivation relies on diagonal dominance of the bracketed matrix.
  4. [Fig. 2 caption] The caption contains the typo 'PHYDYAS4' and should read 'PHYDYAS' with a space before the overlap factor.
  5. [Notation throughout] The notation for the filter matrix, e.g., G_p, \tilde G, and the relationship between N, P, and L, is introduced quickly and is hard to follow; a short table of dimensions and symbols would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: AFBM is a constructive combination of DAFT precoding and FBMC filtering; PAPR/OOBE are simulated, and DD robustness is inherited from AFDM by design.

full rationale

The derivation chain is self-contained in the relevant sense. The compensation vector b~ is solved in closed form from the known prototype filter (Eqs. 10-13), so it is not a fitted parameter renamed as a prediction. The DD-channel robustness is inherited because the transmitter retains an IDAFT/DAFT structure like AFDM, and Eq. (19) with Fig. 2 illustrates the resulting banded effective channel rather than fitting a target metric. Two self-citations are used: [6] for the chirp orthogonality condition and [14] for the O<=1.5 diagonal-dominance guarantee. Both are prior published results with stated assumptions; neither is fitted to the present data, and neither is invoked as a uniqueness argument to rule out alternatives, so they do not make the claim circular. The abstract's simultaneous -100 dB OOBE and quasi-orthogonality claims are not demonstrated in one configuration (the O=4 PHYDYAS case violates the O<=1.5 guarantee stated in Section II-B, and no BER results are shown despite the introduction's promise), but that is an evidentiary gap and an internal consistency risk, not a reduction of the derivation to its inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central design depends on three hand-chosen or prior-derived elements: the DAFT chirp parameters and guard width from the AFDM condition, the prototype filter and overlap choice, and the assumption that a K=1 effective-channel analysis represents the full overlapping filter-bank system. The diagonal compensation vector is computed from known filter coefficients rather than fitted to data. No new physical entities are introduced.

free parameters (3)
  • DAFT chirp parameters (c1, c2) = not specified in paper
    The DAFT/IDAFT precoding is chirp-based; the parameters are required to satisfy the AFDM orthogonality inequality 2(f_max+xi)(ell_max+1)+ell_max <= P. No explicit values are given in the paper, and they depend on channel delay and Doppler knowledge.
  • Guard width xi = not specified in paper
    Free parameter in the chirp orthogonality condition controlling additional guard elements around channel diagonals; its value is not reported in the simulations.
  • Prototype filter and overlap factor O = Hermite O=1.5 or PHYDYAS O=4
    The OOBE and orthogonality results are sensitive to this choice. PHYDYAS O=4 gives the -100 dB OOBE but violates the paper's O<=1.5 condition for guaranteed complex orthogonality; Hermite O=1.5 preserves the guarantee with worse spectral containment. This design choice is hand-selected per simulation, not derived.
assumptions (4)
  • domain assumption AFDM chirp orthogonality condition: 2(f_max+xi)(ell_max+1)+ell_max <= P, cited from prior AFDM work [6].
    The chirp parameters of AFBM are selected using this prior condition. It is an unproved-in-this-paper background result from the authors' own earlier survey/work on AFDM.
  • domain assumption Diagonal compensation achieves complex orthogonality when overlap factor O <= 1.5, based on [14].
    The paper states the compensation stage satisfies Eq. (11) approximately and that this is guaranteed for overlap factors up to 1.5, invoking prior work. The approximation itself is not derived in this paper.
  • domain assumption Circular convolutional doubly-dispersive channel model with phase-prefix and roots-of-unity matrices.
    The channel model in Section II-C is a standard circular-convolution model for AFDM/OTFS analysis, used to define the effective channel. It assumes ideal cyclic behavior that may not hold for finite-length filter-bank transmission.
  • ad hoc to paper Single-symbol K=1 analysis is representative of the full overlapping filter-bank system.
    Section II-C says 'without loss of generality' for K=1, but the filter bank has overlap across symbols (block Toeplitz G and overlap factor O), so inter-symbol interference is not captured. This is a load-bearing simplification for the quasi-orthogonality claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Affine Filter Bank Modulation: A New Waveform for High Mobility Communications." pith.science (2026). https://pith.science/paper/VT3ZBJO4

@misc{pith2026250503589,
  author       = {Pith},
  title        = {Pith review of: Affine Filter Bank Modulation: A New Waveform for High Mobility Communications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VT3ZBJO4}},
  note         = {Machine review of arXiv:2505.03589}
}
read the original abstract

We propose a new waveform suitable for integrated sensing and communications (ISAC) systems facing doubly-dispersive (DD) channel conditions, as typically encountered in high mobility scenarios. Dubbed Affine Filter Bank Modulation (AFBM), this novel waveform is designed based on a filter-bank structure, known for its ability to suppress out-of-band emissions (OOBE), while integrating a discrete affine Fourier transform (DAFT) precoding stage which yields low peak-to-average power ratio (PAPR) and robustness to DD distortion, as well as other features desirable for ISAC. Analytical and simulation results demonstrate that AFBM maintains quasi-orthogonality similar to that of affine frequency division multiplexing (AFDM) in DD channels, while achieving PAPR levels 3 dB lower, in addition to OOBE as low as -100 dB when implemented with PHYDYAS prototype filters.

Figures

Figures reproduced from arXiv: 2505.03589 by the authors.

Figure 1
Figure 1. Illustration of the full AFBM modulation framework. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. 3D illustration of a three-path AFBM effective channel of the DAFT [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. PAPR performance of AFDM and AFBM. -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Normalized Frequency -150 -100 -50 0 P S D (d B) AFDM AFBM, Hermite, O = 1:5 AFBM, PHYDYAS, O = 4 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: OOBE Performance of AFDM and AFBM with the Hermite and [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Affine Filter Bank Modulation (AFBM): A Novel 6G ISAC Waveform with Low PAPR and OOBE

    eess.SP 2025-09 conditional novelty 5.0 of 10

    AFBM, a filter-banked chirp-precoded waveform, achieves lower PAPR and OOBE than AFDM in simulation while supporting a belief-propagation data receiver and an EM-assisted sensing receiver.

  2. Low-Complexity Receiver Design for Affine Filter Bank Modulation

    eess.SP 2025-06 conditional novelty 5.0 of 10

    A Gaussian Belief Propagation detector for AFBM is shown, via simulation, to outperform AFDM by about 2 dB at BER 10^-3 in doubly-dispersive channels at low per-iteration cost.

  3. AFDM: Evolving OFDM Towards 6G+

    eess.SP 2026-02 conditional novelty 4.0 of 10

    AFDM is presented as an OFDM-backward-compatible 6G+ waveform whose added transceiver cost is two O(N) chirp rotations, supported by a generalized pulse-shaped FDFD channel formulation.

Reference graph

Works this paper leans on

15 extracted references · 14 canonical work pages · cited by 3 Pith papers

  1. [1]

    A survey on high mobility wireless communications: Challenges, opportunities and solutions,

    J. Wu and P. Fan, “A survey on high mobility wireless communications: Challenges, opportunities and solutions,”IEEE Access, vol. 4, pp. 450– 476, 2016

    2016

  2. [2]

    Integrated sensing and communication signals toward 5G-A and 6G: A survey,

    Z. Weiet. al, “Integrated sensing and communication signals toward 5G-A and 6G: A survey,”IEEE Internet of Things Journal, vol. 10, no. 13, 2023

    work page 2023

  3. [3]

    The integrated sensing and communication revolution for 6G: Vision, techniques, and applications,

    N. Gonz ´alez-Prelcicet. al, “The integrated sensing and communication revolution for 6G: Vision, techniques, and applications,”Proceedings of the IEEE, vol. 112, no. 7, 2024

    work page 2024

  4. [4]

    Joint channel, data, and radar parameter estimation for afdm systems in doubly-dispersive channels,

    K. R. R. Ranasingheet. al, “Joint channel, data, and radar parameter estimation for afdm systems in doubly-dispersive channels,”IEEE Transactions on Wireless Communications, vol. 24, no. 2, 2025

    work page 2025

  5. [5]

    Comparison of CP-OFDM and OFDM/OQAM in doubly dispersive channels,

    J. Du and S. Signell, “Comparison of CP-OFDM and OFDM/OQAM in doubly dispersive channels,” inFuture Generation Communication and Networking (FGCN 2007), vol. 2, 2007

    work page 2007

  6. [6]

    H. S. Rouet. al, “From orthogonal time–frequency space to affine frequency-division multiplexing: A comparative study of next-generation waveforms for integrated sensing and communications in doubly disper- sive channels,”IEEE Signal Process. Mag., vol. 41, no. 5, 2024

    work page 2024

  7. [7]

    Affine frequency division multiplexing for next generation wireless communications,

    A. Bemani, N. Ksairi, and M. Kountouris, “Affine frequency division multiplexing for next generation wireless communications,”IEEE Trans- actions on Wireless Communications, vol. 22, no. 11, 2023

    work page 2023

  8. [8]

    DAFT-spread affine frequency division multiple access for downlink transmission,

    Y . Taoet. al, “DAFT-spread affine frequency division multiple access for downlink transmission,” inGLOBECOM 2024 - 2024 IEEE Global Communications Conference, 2024

    work page 2024

Show all 15 references
  1. [9]

    Spectrum design for orthogonal chirp division multiplexing transmissions,

    M. S. Omar and X. Ma, “Spectrum design for orthogonal chirp division multiplexing transmissions,”IEEE Wireless Communications Letters, vol. 9, no. 11, 2020

    2020

  2. [10]

    Special cases of DFT-based modulation and demodulation for affine frequency division multiplexing,

    V . Savaux, “Special cases of DFT-based modulation and demodulation for affine frequency division multiplexing,”IEEE Transactions on Com- munications, vol. 72, no. 12, 2024

    2024

  3. [11]

    A novel filter-bank multicarrier scheme to mitigate the intrinsic interference: Application to mimo systems,

    R. Zakaria and D. Le Ruyet, “A novel filter-bank multicarrier scheme to mitigate the intrinsic interference: Application to mimo systems,”IEEE Transactions on Wireless Communications, vol. 11, no. 3, 2012

    2012

  4. [12]

    Pruned DFT-spread FBMC: Low PAPR, low la- tency, high spectral efficiency,

    R. Nissel and M. Rupp, “Pruned DFT-spread FBMC: Low PAPR, low la- tency, high spectral efficiency,”IEEE Transactions on Communications, vol. 66, no. 10, 2018

    2018

  5. [13]

    A novel DFT precoded filter bank system with iterative equalization,

    R. P. Junior, C. A. F. d. Rocha, B. S. Chang, and D. Le Ruyet, “A novel DFT precoded filter bank system with iterative equalization,”IEEE Wireless Communications Letters, vol. 10, no. 3, 2021

    2021

  6. [14]

    A generalized DFT precoded filter bank system,

    ——, “A generalized DFT precoded filter bank system,”IEEE Wireless Communications Letters, vol. 11, no. 6, 2022

    2022

  7. [15]

    A two-dimensional FFT precoded filter bank scheme,

    ——, “A two-dimensional FFT precoded filter bank scheme,”IEEE Transactions on Wireless Communications, vol. 22, no. 11, 2023

    2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.