pith. sign in

arxiv: 2604.10296 · v1 · submitted 2026-04-11 · 📡 eess.SP

DBU-OFDM: A Trainable Deep Block-Unitary OFDM Waveform for Integrated Sensing and Communication

Cheng Luo , Luping Xiang , Hankun Zhang , Yi Luo , Chen Huang , Yi Zhang , Cheng-Xiang Wang , Kun Yang This is my paper

Pith reviewed 2026-05-10 15:31 UTC · model grok-4.3

classification 📡 eess.SP
keywords DBU-OFDMOFDM waveform designintegrated sensing and communicationPAPR reductiondeep learning waveformsHouseholder reflectionsblock-unitary transformfrequency diversity
0
0 comments X

The pith

A block-unitary transform learned only over OFDM data subcarriers reduces PAPR and improves both data reliability and sensing accuracy while keeping the original pilot structure fixed.

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

The paper develops DBU-OFDM as a structure-preserving way to adapt OFDM waveforms for joint communication and sensing. Learning is confined to a unitary map applied only to data subcarriers and realized through recursive Householder reflections, leaving pilots, nulls, and the DFT backbone untouched. This yields lower peak power, added frequency diversity for fading channels, and better range-velocity estimates in sensing, all with hardware-verified compatibility and low overhead.

Core claim

DBU-OFDM restricts learning to a block-unitary transformation over data subcarriers parameterized by recursive Householder reflections; the resulting waveform achieves PAPR tails close to block-pilot DFT-s-OFDM while retaining comb-type pilots, improves communication reliability via frequency-domain diversity, and enhances range and velocity estimation especially in dimension-limited settings, all while preserving DFT-based structure and low-complexity equalization.

What carries the argument

The block-unitary transformation over data subcarriers, parameterized by recursive Householder reflections to enforce strict unitarity while remaining differentiable and complexity-controllable.

Load-bearing premise

That restricting the learned map to a unitary transformation on data subcarriers will produce measurable gains in PAPR, reliability, and sensing without violating pilot protection or adding unacceptable overhead in real channels.

What would settle it

A side-by-side test in a frequency-selective channel where the learned DBU-OFDM shows no improvement in PAPR tail or velocity estimation error relative to conventional DFT-s-OFDM with the same pilot pattern.

Figures

Figures reproduced from arXiv: 2604.10296 by Cheng Luo, Cheng-Xiang Wang, Chen Huang, Hankun Zhang, Kun Yang, Luping Xiang, Yi Luo, Yi Zhang.

Figure 1
Figure 1. Figure 1: Two pilot insertion patterns in OFDM resource allocation. (a) Comb [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: End-to-end transceiver pipeline of the proposed DBU-OFDM system. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Construction of the DBU-OFDM transformation matrix. The permutation matrix contains binary entries and realizes structured index remapping across [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Recursive construction of the trainable unitary matrix via Householder [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: CCDFs for different waveforms, with parameters [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Block-wise structure of Udata with cases B = 1, B = 4, and B = Ndata, respectively. the additive noise vector. Note that Λd,m is assumed to remain constant within the considered coherent processing block. At the receiver, conventional one-tap minimum mean square error (MMSE) equalization is adopted. Let Gd,m denotes the equalizer matrix, the equalized transformed-symbol vector is then expressed as ˆ¯sd,m =… view at source ↗
Figure 7
Figure 7. Figure 7: BER performance of conventional OFDM and DBU-OFDM under the [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: BLER performance of conventional OFDM and DBU-OFDM under [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Sensing performance comparison between conventional OFDM and [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: USRP-based over-the-air experimental setup for validating conven [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 12
Figure 12. Figure 12: Experimental PAPR comparison between conventional OFDM and [PITH_FULL_IMAGE:figures/full_fig_p011_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: FPGA architecture of the merged Householder hardware module. [PITH_FULL_IMAGE:figures/full_fig_p012_13.png] view at source ↗
read the original abstract

Orthogonal frequency-division multiplexing (OFDM) is a dominant waveform in modern wireless systems, yet its high peak-to-average power ratio (PAPR) and limited adaptability hinder efficient support for integrated communication and sensing. This paper proposes deep block-unitary precoded OFDM (DBU-OFDM), a structure-preserving learning framework that enables trainable waveform adaptation while preserving the DFT-based signal structure, pilot/null resource protection, and compatibility with low-complexity frequency-domain equalization. The proposed design restricts learning to a block-unitary transformation over data subcarriers and preserves pilot and null resources for structural compatibility. The transform is parameterized by recursive Householder reflections, ensuring strict unitarity as well as differentiable, numerically stable, and complexity-controllable implementation. Results show that DBU-OFDM achieves PAPR tails close to block-pilot DFT-s-OFDM while retaining comb-type pilots, improves communication reliability in frequency-selective fading via frequency-domain diversity, and enhances range and velocity estimation in direct sensing, especially in dimension-limited settings. Over-the-air USRP experiments and FPGA prototyping further verify its practical feasibility, demonstrating low error vector magnitude (EVM), clear PAPR reduction in real transmission, and hardware throughput up to 200~MS/s with microsecond-level latency. DBU-OFDM therefore offers a practical intermediate solution between conventional model-based OFDM waveforms and unconstrained neural transceivers for next-generation integrated communication and sensing systems.

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

1 major / 0 minor

Summary. The manuscript proposes DBU-OFDM, a structure-preserving trainable waveform for integrated sensing and communication. It restricts adaptation to a block-unitary precoding matrix applied only to data subcarriers, parameterized via recursive Householder reflections to enforce unitarity while preserving the DFT-based OFDM structure, comb-type pilots, null subcarriers, and claimed compatibility with low-complexity frequency-domain equalization. Simulation results are presented for PAPR reduction (close to block-pilot DFT-s-OFDM), improved communication reliability via frequency-domain diversity in fading, and enhanced range/velocity estimation in sensing (especially dimension-limited cases). These are supplemented by over-the-air USRP experiments and FPGA prototyping showing low EVM, real PAPR reduction, and hardware throughput up to 200 MS/s with microsecond latency.

Significance. If the performance claims and low-complexity compatibility hold, the work would provide a useful intermediate design between conventional model-based OFDM and unconstrained neural transceivers for ISAC. The explicit use of recursive Householder reflections for a differentiable, numerically stable, and complexity-controllable unitary parameterization is a technical strength, as is the hardware verification demonstrating practical feasibility beyond simulation.

major comments (1)
  1. Abstract and design description: The central claim of compatibility with low-complexity frequency-domain equalization is load-bearing but appears inconsistent with the block-unitary transform. Applying a non-diagonal unitary U (even if Householder-parameterized) to data subcarriers within each block produces an effective channel matrix H U, where H is diagonal but the product is dense. Symbol recovery then requires block-level matrix inversion or equalization (scaling as O(B^2) or O(B^3) per block of size B) rather than independent scalar operations per subcarrier. The manuscript must clarify the receiver processing (e.g., in the equalization or detection section) and provide explicit complexity analysis showing how O(1) per-subcarrier FDE is retained without unacceptable overhead.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and valuable comments on our work. We provide a point-by-point response to the major comment below, and we will update the manuscript to incorporate the necessary clarifications.

read point-by-point responses

  1. Referee: Abstract and design description: The central claim of compatibility with low-complexity frequency-domain equalization is load-bearing but appears inconsistent with the block-unitary transform. Applying a non-diagonal unitary U (even if Householder-parameterized) to data subcarriers within each block produces an effective channel matrix H U, where H is diagonal but the product is dense. Symbol recovery then requires block-level matrix inversion or equalization (scaling as O(B^2) or O(B^3) per block of size B) rather than independent scalar operations per subcarrier. The manuscript must clarify the receiver processing (e.g., in the equalization or detection section) and provide explicit complexity analysis showing how O(1) per-subcarrier FDE is retained without unacceptable overhead.

    Authors: We appreciate the referee pointing out the need for clarity on this key aspect. The unitary precoding U is applied to the data symbols before the OFDM modulation (IFFT) at the transmitter. At the receiver, following the FFT operation, the standard low-complexity frequency-domain equalization is applied independently to each subcarrier by dividing the received value by the estimated channel gain (interpolated from comb pilots). This produces an estimate of the precoded symbols z ≈ U s. The original data symbols are then obtained via s = U^H z, where the inverse unitary transform is applied block-wise using the adjoint of the Householder reflections. This preserves the per-subcarrier O(1) complexity of the FDE step. The block-wise unitary application adds a controllable complexity of O(B) per block when using the efficient Householder implementation (with the number of reflections limited). We will add a dedicated paragraph in the receiver processing section detailing this procedure and provide an explicit complexity breakdown in the revised manuscript, demonstrating that the overhead remains negligible for practical block sizes while enabling the observed performance gains. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's core contribution is a parameterization of a block-unitary transform via recursive Householder reflections to enforce unitarity while preserving OFDM structure, pilots, and nulls. This rests on standard linear-algebra properties of unitary matrices and Householder reflections rather than any self-referential definition, fitted input renamed as prediction, or load-bearing self-citation. No equations reduce claimed PAPR, diversity, or sensing gains to quantities defined by the learned parameters themselves; performance claims are presented as empirical outcomes from simulation and hardware experiments. The design is therefore self-contained against external benchmarks of unitary-matrix theory and OFDM compatibility.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 1 invented entities

The central claim rests on standard OFDM structural assumptions plus the new parameterized unitary transform; no free parameters are numerically specified in the abstract.

free parameters (1)
  • Trainable parameters of the block-unitary transform
    The deep model parameters are learned from data to adapt the waveform.
axioms (2)
  • domain assumption A block-unitary transform applied only to data subcarriers preserves the DFT-based signal structure, pilot/null resources, and compatibility with frequency-domain equalization.
    Invoked to maintain hardware and receiver compatibility.
  • standard math Recursive Householder reflections yield a strictly unitary, differentiable, and numerically stable parameterization.
    Standard linear-algebra fact used to enable training.
invented entities (1)
  • DBU-OFDM waveform no independent evidence
    purpose: Trainable yet structure-preserving OFDM for integrated sensing and communication.
    Newly introduced framework whose performance claims rest on the proposed parameterization.

pith-pipeline@v0.9.0 · 5574 in / 1789 out tokens · 56677 ms · 2026-05-10T15:31:28.194212+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

33 extracted references · 33 canonical work pages

  1. [1]

    On the road to 6G: Visions, requirements, key technologies, and testbeds,

    C.-X. Wang, X. You, X. Gao, X. Zhu, Z. Li, C. Zhang, H. Wang, Y . Huang, Y . Chen, H. Haaset al., “On the road to 6G: Visions, requirements, key technologies, and testbeds,”IEEE Communications Surveys & Tutorials, vol. 25, no. 2, pp. 905–974, 2023

    work page 2023

  2. [2]

    Ten challenges in advancing machine learning technologies toward 6G,

    N. Kato, B. Mao, F. Tang, Y . Kawamoto, and J. Liu, “Ten challenges in advancing machine learning technologies toward 6G,”IEEE Wireless Communications, vol. 27, no. 3, pp. 96–103, 2020

    work page 2020

  3. [3]

    Deep learning in physical layer communications,

    Z. Qin, H. Ye, G. Y . Li, and B.-H. F. Juang, “Deep learning in physical layer communications,”IEEE Wireless Communications, vol. 26, no. 2, pp. 93–99, 2019

    work page 2019

  4. [4]

    Large AI model-based semantic communications,

    F. Jiang, Y . Peng, L. Dong, K. Wang, K. Yang, C. Pan, and X. You, “Large AI model-based semantic communications,”IEEE Wireless Com- munications, vol. 31, no. 3, pp. 68–75, 2024

    work page 2024

  5. [5]

    An enhanced 6G pervasive channel model towards standardization,

    C.-X. Wang, Z. Lv, C. Huang, Y . Huang, J. Wang, J. Huang, and X. You, “An enhanced 6G pervasive channel model towards standardization,” Science China Information Sciences, vol. 68, no. 6, p. 162301, 2025

    work page 2025

  6. [6]

    Trainable communication systems: Concepts and proto- type,

    S. Cammerer, F. Ait Aoudia, S. D ¨orner, M. Stark, J. Hoydis, and S. Ten Brink, “Trainable communication systems: Concepts and proto- type,”IEEE Transactions on Communications, vol. 68, no. 9, pp. 5489– 5503, 2020

    work page 2020

  7. [7]

    An introduction to deep learning for the physical layer,

    T. O’shea and J. Hoydis, “An introduction to deep learning for the physical layer,”IEEE Transactions on Cognitive Communications and Networking, vol. 3, no. 4, pp. 563–575, 2017

    work page 2017

  8. [8]

    End-to-end design of polar coded integrated data and energy networking,

    J. Hu, J. Cui, L. Xiang, and K. Yang, “End-to-end design of polar coded integrated data and energy networking,”IEEE Transactions on Communications, vol. 72, no. 11, pp. 7017–7031, 2024

    work page 2024

  9. [9]

    Polar coded integrated data and energy networking: A deep neural network assisted end-to-end design,

    L. Xiang, J. Cui, J. Hu, K. Yang, and L. Hanzo, “Polar coded integrated data and energy networking: A deep neural network assisted end-to-end design,”IEEE Transactions on Vehicular Technology, vol. 72, no. 8, pp. 11 047–11 052, 2023

    work page 2023

  10. [10]

    Pruning the pilots: Deep learning- based pilot design and channel estimation for MIMO-OFDM systems,

    M. B. Mashhadi and D. G ¨und¨uz, “Pruning the pilots: Deep learning- based pilot design and channel estimation for MIMO-OFDM systems,” IEEE Transactions on Wireless Communications, vol. 20, no. 10, pp. 6315–6328, 2021

    work page 2021

  11. [11]

    End-to-end learning for OFDM: From neural receivers to pilotless communication,

    F. Ait Aoudia and J. Hoydis, “End-to-end learning for OFDM: From neural receivers to pilotless communication,”IEEE Transactions on Wireless Communications, vol. 21, no. 2, pp. 1049–1063, 2021

    work page 2021

  12. [12]

    Deep-waveform: A learned OFDM receiver based on deep complex-valued convolutional networks,

    Z. Zhao, M. C. Vuran, F. Guo, and S. D. Scott, “Deep-waveform: A learned OFDM receiver based on deep complex-valued convolutional networks,”IEEE Journal on Selected Areas in Communications, vol. 39, no. 8, pp. 2407–2420, 2021

    work page 2021

  13. [13]

    Waveform learning for next-generation wireless communication systems,

    F. Ait Aoudia and J. Hoydis, “Waveform learning for next-generation wireless communication systems,”IEEE Transactions on Communica- tions, vol. 70, no. 6, pp. 3804–3817, 2022

    work page 2022

  14. [14]

    Low PAPR MIMO-OFDM design based on convolutional autoencoder,

    Y . Huleihel and H. H. Permuter, “Low PAPR MIMO-OFDM design based on convolutional autoencoder,”IEEE Transactions on Communi- cations, vol. 72, no. 5, pp. 2779–2792, 2024

    work page 2024

  15. [15]

    End- to-end optimized non-orthogonal multicarrier waveform design via deep learning,

    C. Liu, G. Liu, F. Xu, Q. Li, H. Liu, L. Zhang, and M. A. Imran, “End- to-end optimized non-orthogonal multicarrier waveform design via deep learning,”IEEE Transactions on Communications, 2025

    work page 2025

  16. [16]

    Waveform design for MIMO-OFDM integrated sensing and communication system: An information theoretical approach,

    Z. Wei, J. Piao, X. Yuan, H. Wu, J. A. Zhang, Z. Feng, L. Wang, and P. Zhang, “Waveform design for MIMO-OFDM integrated sensing and communication system: An information theoretical approach,”IEEE Transactions on Communications, vol. 72, no. 1, pp. 496–509, 2023

    work page 2023

  17. [17]

    A dual-functional sensing- communication waveform design based on OFDM,

    Y . He, G. Yu, Z. Tang, J. Wang, and H. Luo, “A dual-functional sensing- communication waveform design based on OFDM,”IEEE Transactions on Wireless Communications, vol. 23, no. 11, pp. 16 962–16 975, 2024

    work page 2024

  18. [18]

    Cross- domain dual-functional OFDM waveform design for accurate sens- ing/positioning,

    F. Zhang, T. Mao, R. Liu, Z. Han, S. Chen, and Z. Wang, “Cross- domain dual-functional OFDM waveform design for accurate sens- ing/positioning,”IEEE Journal on Selected Areas in Communications, vol. 42, no. 9, pp. 2259–2274, 2024

    work page 2024

  19. [19]

    Deep learning- based design of uplink integrated sensing and communication,

    Q. Qi, X. Chen, C. Zhong, C. Yuen, and Z. Zhang, “Deep learning- based design of uplink integrated sensing and communication,”IEEE Transactions on Wireless Communications, vol. 23, no. 9, pp. 10 639– 10 652, 2024

    work page 2024

  20. [20]

    ISAC- NET: Model-driven deep learning for integrated passive sensing and communication,

    W. Jiang, D. Ma, Z. Wei, Z. Feng, P. Zhang, and J. Peng, “ISAC- NET: Model-driven deep learning for integrated passive sensing and communication,”IEEE Transactions on Communications, vol. 72, no. 8, pp. 4692–4707, 2024

    work page 2024

  21. [21]

    Model-based end-to-end learning for multi-target integrated sensing and communication under hardware impairments,

    J. M. Mateos-Ramos, C. H ¨ager, M. F. Keskin, L. Le Magoarou, and H. Wymeersch, “Model-based end-to-end learning for multi-target integrated sensing and communication under hardware impairments,” IEEE Transactions on Wireless Communications, vol. 24, no. 3, pp. 2574–2589, 2025

    work page 2025

  22. [22]

    CP-OFDM achieves the lowest average ranging sidelobe under QAM/PSK constellations,

    F. Liu, Y . Zhang, Y . Xiong, S. Li, W. Yuan, F. Gao, S. Jin, and G. Caire, “CP-OFDM achieves the lowest average ranging sidelobe under QAM/PSK constellations,”IEEE Transactions on Information Theory, 2025

    work page 2025

  23. [23]

    Modeling, capacity studies, antenna and system designs for 6G/B6G 3D continuous-space radio channels enabled by electromagnetic information theory,

    C.-X. Wang, J. Li, J. Huang, C. Huang, Z. Zhang, Y . Liu, S. Zhou, Y . Chen, X. You, X. Gaoet al., “Modeling, capacity studies, antenna and system designs for 6G/B6G 3D continuous-space radio channels enabled by electromagnetic information theory,”IEEE Communications Surveys & Tutorials, 2025

    work page 2025

  24. [24]

    From multilayer perceptron to GPT: A reflection on deep learning research for wireless physical layer,

    M. Akrout, A. Mezghani, E. Hossain, F. Bellili, and R. W. Heath, “From multilayer perceptron to GPT: A reflection on deep learning research for wireless physical layer,”IEEE Communications Magazine, vol. 62, no. 7, pp. 34–41, 2024

    work page 2024

  25. [25]

    Orthogonal time frequency space modula- tion,

    R. Hadani, S. Rakib, M. Tsatsanis, A. Monk, A. J. Goldsmith, A. F. Molisch, and R. Calderbank, “Orthogonal time frequency space modula- tion,” in2017 IEEE wireless communications and networking conference (WCNC). IEEE, 2017, pp. 1–6

    work page 2017

  26. [26]

    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, pp. 8214–8229, 2023

    work page 2023

  27. [27]

    Unitary checkerboard pre- coded OFDM for low-PAPR optical wireless communications,

    T. E. Abrudan, S. Kucera, and H. Claussen, “Unitary checkerboard pre- coded OFDM for low-PAPR optical wireless communications,”Journal of Optical Communications and Networking, vol. 14, no. 4, pp. 153–164, 2022

    work page 2022

  28. [28]

    Engineering of arbitrary u (n) transformations by quantum householder reflections,

    P. A. Ivanov, E. Kyoseva, and N. V . Vitanov, “Engineering of arbitrary u (n) transformations by quantum householder reflections,”Physical Review A—Atomic, Molecular, and Optical Physics, vol. 74, no. 2, p. 022323, 2006

    work page 2006

  29. [29]

    DFT-spread OFDM with frequency domain reference symbols,

    A. Sahin, E. Bala, R. Yang, and R. L. Olesen, “DFT-spread OFDM with frequency domain reference symbols,” inGLOBECOM 2017-2017 IEEE Global Communications Conference. IEEE, 2017, pp. 1–6

    work page 2017

  30. [30]

    Fast fourier transform v9. 1 logicore IP product guide,

    I. Xilinx, “Fast fourier transform v9. 1 logicore IP product guide,” 2022

    work page 2022

  31. [31]

    A flexible-channel MDF architecture for pipelined radix-2 FFT,

    X. Zhou, X. Chen, Y . He, and X. Mou, “A flexible-channel MDF architecture for pipelined radix-2 FFT,”IEEE Access, vol. 11, pp. 38 023–38 033, 2023

    work page 2023

  32. [32]

    An area-energy- efficient 64–2048 point FFT with approximate plane-fitting complex multipliers,

    W. Shi, J. Wu, Y . Yuan, Z. Mo, C. Wu, and J. He, “An area-energy- efficient 64–2048 point FFT with approximate plane-fitting complex multipliers,”IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2025

    work page 2048

  33. [33]

    An efficient implementation approach to FFT processor for spectral analysis,

    J. Hazarika, M. T. Khan, S. R. Ahamed, and H. B. Nemade, “An efficient implementation approach to FFT processor for spectral analysis,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–13, 2023. 13

    work page 2023