Pith. sign in

REVIEW 3 major objections 5 minor 35 references

Deep Learning for Joint Narrowband Interference Cancellation and Soft Demodulation in OFDM Systems

T0 review · 3 major / 5 minor · reviewed 2026-07-10 · glm-5.2

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

Pith's one-line read Neural pipeline cancels narrowband interference without knowing how many interferers exist

desk verdict Joint neural NBI cancellation + soft demodulation for OFDM: solid engineering, but AWGN-only evaluation limits the central whitening claim. read the letter →

arxiv 2607.08717 v1 pith:WIFDQBAB submitted 2026-07-09 cs.LG eess.SP

Emmanouil Kavvousanos , Francky Catthoor , Vassilis Paliouras This is my paper · ORCID
classification cs.LGeess.SP
keywords narrowbandinterferenceOFDMdeeplearningLLRestimationcompressedsensingDirichletkernelLDPCdecodingcancellation
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 proposes two neural networks that work together to fix a broken link in OFDM receivers. The first, NBI-CNet, estimates and subtracts narrowband interference in a single pass using a physics-informed convolutional architecture that models spectral leakage via the Dirichlet kernel. The second, LLR-CNet, remaps the non-Gaussian residual left behind after imperfect cancellation into well-calibrated soft bits for the decoder. Together they eliminate error floors that plague classical compressed-sensing pipelines, which suffer from sequential latency, dependence on knowing the exact interferer count, and a mismatch between their structured residuals and the Gaussian assumption of standard demappers.

What carries the argument

The Dirichlet kernel spectral leakage model, a differentiable analytical reconstruction layer mapping predicted physical parameters to an interference waveform, circular padding in convolutional layers to preserve OFDM spectral wrap-around, binary cross-entropy loss driving network logits to exact LLR convergence, and dynamic conditional routing that evaluates parameter heads only at detected interference locations.

What would settle it

A deployment scenario over a realistic frequency-selective fading channel with imperfect channel estimation, where the post-equalization residual structure differs from the AWGN-trained statistics, causing LLR-CNet to produce miscalibrated soft bits and reintroducing the error floors it was designed to eliminate.

Watch

Extended reading notes

Core claim

The central discovery is that the pipeline mismatch between imperfect interference cancellation and Gaussian-soft-demodulation can be resolved by a two-stage neural architecture: a physics-informed estimator that reconstructs the interference waveform analytically from predicted physical parameters (gain, fractional offset, phase), and a lightweight neural LLR estimator trained with binary cross-entropy that naturally converges to exact log-likelihood ratios and acts as a structural whitener for non-Gaussian residuals. The architecture operates without prior knowledge of the active interferer count, generalizes across FFT sizes via a closed-form gain rescaling, and eliminates the error floor

Load-bearing premise

The evaluation assumes a perfect AWGN channel for the signal of interest, isolating the interference and demodulation modules from channel estimation and equalization errors. Real OFDM systems operate over multipath fading channels, and if residual fading after imperfect equalization interacts with NBI residual structure, the learned residual statistics may no longer match, potentially reintroducing error floors.

Editorial extensions

If this is right

  • Receivers in congested spectrum bands (e.g., 6G FR3, IoT, power-line communications) could deploy a single trained model that adapts to varying interference density without retraining or external interferer-count detectors.
  • The structural-whitener approach to non-Gaussian residuals could generalize to other types of structured interference beyond narrowband tones, such as impulsive noise or multi-user interference, where classical Gaussian demappers also fail.
  • The scale-invariant design principle—training on a small FFT size and transferring via analytical amplitude rescaling—could reduce the data-generation and training cost for neural physical-layer components across heterogeneous system configurations.
  • Eliminating the dependency on prior interferer-count estimation removes a single point of failure in classical CS pipelines, making the receiver robust to dynamic environments where interferers appear and disappear on a per-symbol basis.

Reading between the lines

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

  • If channel estimation errors in real multipath fading channels produce residual structures distinct from those generated by NBI-cancellation errors alone, LLR-CNet's learned residual statistics may not generalize without co-training under realistic channel conditions.
  • The scale-invariance claim rests on the assumption that the 1/√N amplitude scaling captures all FFT-size-dependent effects; if convolutional receptive fields interact differently with spectral leakage patterns at larger N, a single rescaling factor may be insufficient.
  • The dynamic conditional routing that bypasses parameter heads at interference-free subcarriers implicitly assumes the gain head achieves near-perfect detection—false negatives would leave interferers unmitigated, a failure mode not characterized in the paper.
  • Training exclusively under AWGN may limit LLR-CNet's ability to whiten residuals when post-equalization fading artifacts introduce their own non-Gaussian structure, potentially requiring a more diverse training distribution.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes a unified deep-learning framework for joint narrowband interference (NBI) cancellation and soft demodulation in OFDM systems. The framework comprises two components: NBI-CNet, a physics-informed convolutional network that estimates multi-tone NBI parameters (gain, fractional frequency offset, phase) in a single forward pass without requiring prior knowledge of the interferer count; and LLR-CNet, a lightweight neural network that maps non-Gaussian post-mitigation residuals onto calibrated log-likelihood ratios (LLRs) for LDPC decoding. The authors show via simulation that the joint architecture eliminates error floors inherent to classical compressed-sensing baselines (OMP-IDS, EOMP-IDS) paired with Gaussian max-log demappers, operates within 0.2–0.5 dB of the optimal iterative baseline under severe interference (SIR=−10 dB), and delivers >3 dB coding gain under mild interference (SIR=10 dB, Q=12) where greedy algorithms suffer signal-peak confusion. The architecture is shown to generalize across FFT sizes without retraining via a closed-form gain rescaling.

Significance. The paper addresses a practically important problem: the mismatch between imperfect NBI cancellation and downstream soft demodulation in OFDM systems. The joint architecture is well-motivated, and the physics-informed design of NBI-CNet—explicitly modeling Dirichlet kernel spectral leakage with a differentiable reconstruction layer—is a strength. The complexity analysis (Section VIII) is detailed, translating classical CS FLOP counts into comparable metrics and demonstrating up to 60% reduction over EOMP-IDS at high density (N=2048, Q=64). The LLR-CNet design, leveraging BCE loss convergence to exact LLRs, is a clean and falsifiable contribution. The demonstration that the pipeline circumvents error floors from imperfect interferer-count estimation (Fig. 10) is a notable practical result. However, the significance is tempered by the restricted evaluation regime (AWGN channel, perfect equalization), which limits the generalizability of the central 'structural whitener' claim.

major comments (3)
  1. §II.C, Eq. (4): The entire evaluation assumes an AWGN channel with H_k=1, which the paper justifies as mirroring 'the effective background noise floor of a post-equalization signal.' This assumes perfect equalization. The central claim that LLR-CNet acts as a 'structural whitener' (§VI) depends on the residual structure being statistically stable and learnable. In a real multipath OFDM system, post-equalization residuals include channel estimation error and residual ISI, whose statistics differ from AWGN. LLR-CNet takes both the cleaned signal and the estimated interference footprint as inputs (Eq. 34); if the signal component carries residual fading structure that LLR-CNet never saw during training, its learned mapping to calibrated soft metrics would be mismatched. The paper does not test this. This is the soft spot: the 'structural whitener' claim is validated only in the regime where
  2. §VII.D, Eq. (40): The scale-invariance claim assumes the only FFT-size-dependent effect is the 1/√N amplitude scaling. However, the convolutional receptive field (kernel size K=7, padding P=3) operates on a fixed local window. If spectral leakage patterns or the interaction between adjacent interferers change with N (e.g., the density of interferers relative to the grid changes), the learned features may not generalize. The paper demonstrates generalization from N=256 to N=512 (Fig. 10), but this is a modest 2× scaling. A more rigorous test would involve larger scaling factors (e.g., N=1024 or N=2048) to confirm that the local convolutional features remain valid.
  3. §VII.C, Fig. 9: The claim of >3 dB coding gain under mild interference (SIR=10 dB, Q=12) is attributed to classical algorithms suffering 'signal-peak confusion.' However, the classical baselines (OMP-IDS, EOMP-IDS) are provided with perfect prior knowledge of the interferer count Q in Figs. 9a–c, but it is unclear whether this perfect knowledge is also provided in Figs. 9d–f (SIR=10 dB). If the classical algorithms are disadvantaged by imperfect Q estimation in the mild-interference regime, the comparison may not be fair. The paper should clarify whether the baselines in Fig. 9 operate with perfect or imperfect Q estimation.
minor comments (5)
  1. §IV.A, Eq. (20): The notation χ^(0) ∈ R^{N×2} is introduced but the dimension F_l for intermediate feature maps is not clearly defined until later. A brief forward reference would improve readability.
  2. Table I: The 'Total' parameter count is listed as 14,468, but the sum of the individual layer parameters (480+7,200+2,176+65+2,176+65+2,176+130) equals 14,468. This is correct, but the table would benefit from subtotals for the feature extraction, state conditioning, and output stages.
  3. §VII.A, Fig. 5: The y-axis label 'Power (dB)' is ambiguous—should specify whether this is relative to the signal power or noise floor.
  4. §VIII.E, Fig. 11: The intersection point where NBI-CNet becomes more efficient than EOMP-IDS is stated as Q=26, but the figure only shows curves up to Q=64. A marker or annotation at the intersection would aid verification.
  5. §IX: The conclusion states 'up to a 60% reduction in absolute operations for large grid sizes (N=2048, Q=64),' but Table IV shows the reduction is approximately 59.5% (101.9M vs 41.2M FLOPs). The rounding is acceptable but should be stated precisely.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; one minor self-citation for architectural template, not load-bearing for mathematical claims

full rationale

The paper's two main derivation chains are self-contained. (1) NBI-CNet predicts physical parameters (gain, fractional offset, phase) via a CNN, then reconstructs the interference using the Dirichlet kernel formula (Eq. 8, 22-25). The functional equivalence to CS reconstruction (Eq. 25: ˆE = ˆΦˆd mirroring Eq. 10: E = Φd) is explicitly acknowledged as a reformulation, not presented as a novel derivation. The network is trained against ground-truth interference parameters and time-domain waveforms (Eq. 29), not against its own outputs. (2) LLR-CNet's convergence to true LLRs is derived independently in Eq. 35-38: BCE loss minimization → posterior probability convergence → logit = log(P(c=1|Y)/P(c=0|Y)) = LLR. This is a standard ML result reproduced in-paper, not imported via citation. The self-citation [28] (Kavvousanos & Paliouras, SiPS 2025) is acknowledged as the architectural template for the 1D-CNN demapper, but the mathematical justification for LLR convergence does not depend on [28]. The scale-invariance claim (Eq. 40) derives from the 1/√N dependence in the Dirichlet kernel (Eq. 8), a physics-based scaling rather than a fitted parameter renamed as prediction. No step in the derivation chain reduces to its own inputs by construction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new physical entities or postulated phenomena. The neural network architectures (NBI-CNet, LLR-CNet) are engineering constructs, not theoretical entities. The free parameters are standard deep learning hyperparameters (learning rate, kernel sizes, filter counts, regularization weights) and training distribution parameters. The most consequential axioms are the AWGN channel assumption and the FFT-size scale-invariance assumption, both of which are load-bearing for the generalization claims.

free parameters (7)
  • Sparsity weight λ = 0.3
    Empirically determined regularization weight in the hybrid loss (Eq. 29), stated as 'determined empirically' in §IV.E
  • Learning rate η = 10^{-3}
    Adam optimizer learning rate, Table II
  • Conv1D kernel sizes K = 7, 5
    Kernel sizes for NBI-CNet (K=7) and LLR-CNet (K=5), Tables I and III, chosen by architecture design
  • Number of filters F = 32, 16
    Feature extraction filter counts for NBI-CNet (F=32) and LLR-CNet (F=16), Tables I and III
  • Training SNR range = U(7,15) dB
    Uniform distribution range for training data generation, Table II; defines the operational regime the network is optimized for
  • Training SIR range = U(-30,10) dB
    Uniform distribution range for interference power during training, Table II
  • Minimum NBI distance d_m = 2 (training), 4 (eval)
    Minimum subcarrier separation between interferers; set to 2 during training (Table II) but 4 in evaluation (§VII), creating a potential train-test mismatch
assumptions (5)
  • domain assumption AWGN channel assumption (H_k=1, ∀k) for the signal of interest
    §II.C: 'We assume an AWGN channel for the signal of interest (H_k=1, ∀k), which mirrors the effective background noise floor of a post-equalization signal.' This sidesteps multipath and channel estimation errors.
  • domain assumption NBI parameters vary independently from one OFDM symbol to the next
    §II.C: 'the active interferer count Q and its constituent parameters vary independently from one OFDM symbol to the next.' This is a worst-case assumption that may not hold in practice.
  • standard math BCE loss minimization drives network outputs to converge to true posterior LLRs
    §VI.B, Eqs. 35–38: Standard result from neural network theory; the derivation is correct but assumes sufficient training data and capacity.
  • ad hoc to paper Scale-invariance via 1/√N gain rescaling (Eq. 40) is sufficient for FFT-size generalization
    §VII.D: The claim that applying ˆg' = ˆg·√(N'/N) enables generalization to arbitrary N' assumes the only FFT-size-dependent effect is the amplitude scaling, ignoring potential changes in spectral leakage patterns and convolutional receptive field interactions.
  • domain assumption Dirichlet kernel accurately models asynchronous NBI spectral leakage
    §II.B, Eq. 8: Standard OFDM signal model; the Dirichlet kernel is the exact DFT of a finite-length complex exponential, so this is mathematically correct under the assumed signal model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning for Joint Narrowband Interference Cancellation and Soft Demodulation in OFDM Systems." pith.science (2026). https://pith.science/paper/WIFDQBAB

@misc{pith2026260708717,
  author       = {Pith},
  title        = {Pith review of: Deep Learning for Joint Narrowband Interference Cancellation and Soft Demodulation in OFDM Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WIFDQBAB}},
  note         = {Machine review of arXiv:2607.08717}
}
abstract

Narrowband interference (NBI) severely degrades orthogonal frequency-division multiplexing (OFDM) systems by corrupting subcarriers and rendering classical soft demodulation ineffective. Conventional compressed-sensing (CS) mitigation exhibits high sequential latency and leaves structured, non-Gaussian residuals that cause log-likelihood ratio (LLR) unreliability, decoder saturation, and severe error floors when employing classical Gaussian demappers. We resolve this pipeline mismatch using a unified deep learning framework for joint NBI cancellation and robust soft demodulation. First, NBI-CNet employs a physics-informed convolutional architecture to estimate NBI parameters and remove multi-tone interference in a single forward pass. Without requiring prior knowledge of the active interferer count, NBI-CNet reduces computational complexity by up to 60% ($N{=}2048, Q{=}64$) compared to the state-of-the-art EOMP-IDS algorithm. Second, LLR-CNet acts as a structural whitener by mapping non-Gaussian post-mitigation residuals onto well-calibrated soft metrics. Simulations demonstrate that this joint framework eliminates the error floors inherent to traditional baselines across dense grids. Under severe interference ($\text{SIR}{=}{-}10$ dB), the pipeline operates within a $0.2$ to $0.5$ dB SNR margin of the optimal iterative baseline at a target block error rate (BLER) of $10^{-4}$. Under mild interference ($\text{SIR}{=}10$ dB) with heavy spectral overlap ($Q{=}12$), where classical greedy algorithms erroneously subtract valid data components and corrupt the payload, NBI-CNet avoids signal-peak confusion to deliver a coding gain exceeding $3$ dB. Finally, the architecture circumvents the $2{\times}10^{-4}$ error floor triggered by interferer-estimation errors, while its scale-invariant design enables robust generalization across arbitrary FFT sizes without retraining.

Figures

Figures reproduced from arXiv: 2607.08717 by the authors.

Figure 1
Figure 1. Block diagram of the OFDM communication system with NBI mitigation. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 3
Figure 3. Circular Padding for improved spectral leakage identification. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 2
Figure 2. Model Architecture of NBI-CNet. TABLE I NBI-CNET ARCHITECTURE AND PARAMETER COUNT Layer In Shape Out Shape Details Params Feature Extraction Input (Yin) [N, 2] [N, 2] Re/Im Stack 0 Circular Pad [N, 2] [N + 6, 2] P =3 0 Conv1D [N + 6, 2] [N, 32] F =32, K=7, ReLU 480 Circular Pad [N, 32] [N + 6, 32] P =3 0 Conv1D [N + 6, 32] [N, 32] F =32, K=7, ReLU 7,200 State Conditioning Feat. Concat. Conv [N, 32] σ 2 w [N, 1] [N, … view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Model Architecture of LLR-CNet. The mathematical stages natively [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Frequency-domain snapshot of the residual interference power [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: ICR performance of the different algorithms (Q [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: BLER vs. SNR performance comparison between LLR-CNet and the LLR erasure method (Q=24, SIR=−10 dB, LDPC(2048,1024), 16-QAM). −50 0 50 0 5 10 ·10−2 LLR values Probability (a) −10 0 10 0 5 10 ·10−2 LLR values Probability (b) [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: LLR values post NBI Mitigation. (a) Gaussian-based Max-log, (b) [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: BLER vs. SNR for the proposed and baseline CS NBI mitigation algorithms under different interference scenarios (LDPC(1024, 512), 16-QAM). 6 7 8 9 10 11 12 13 14 15 16 10−6 10−5 10−4 10−3 10−2 10−1 100 SNR (dB) BLER No NBI + Max-Log OMP-IDS + Max-Log OMP-IDS + Max-Log (…
Figure 10
Figure 10. Figure 10: BLER vs. SNR performance comparisons (Q=8, SIR=−10 dB, LDPC(2048,1024), 16-QAM). The abbreviation IE indicates Imperfect Esti￾mation of the number of active NBI sources. interference suppression while bypassing the execution latency of sequential searches. Under mild …
Figure 11
Figure 11. Figure 11: Computational complexity (FLOPs) scaling versus the number of [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [1]

    Interference management issues for the future 5G network: A review,

    F. Qamar, M. H. D. N. Hindia, K. Dimyati, K. A. Noordin, and I. S. Amiri, “Interference management issues for the future 5G network: A review,”Telecommunication Systems, vol. 71, no. 4, pp. 627–643, 2019

    work page 2019

  2. [2]

    URLLC in Beyond 5G and 6G Networks: An Interference Management Perspective,

    M. U. A. Siddiqui, H. Abumarshoud, L. Bariah, S. Muhaidat, M. A. Imran, and L. Mohjazi, “URLLC in Beyond 5G and 6G Networks: An Interference Management Perspective,”IEEE Access, vol. 11, pp. 54 639–54 663, 2023

    work page 2023

  3. [3]

    From 5G to 6G Networks: A Survey on AI-Based Jamming and Interference Detection and Mitigation,

    P. Lohan, B. Kantarci, M. Amine Ferrag, N. Tihanyi, and Y . Shi, “From 5G to 6G Networks: A Survey on AI-Based Jamming and Interference Detection and Mitigation,”IEEE Open Journal of the Communications Society, vol. 5, pp. 3920–3974, 2024

    work page 2024

  4. [4]

    A Systematic Review of Interference Mitigation Techniques in Current and Future UA V-Assisted Wireless Networks,

    H. Shakhatreh, A. Sawalmeh, K. F. Hayajneh, S. Abdel-Razeq, W. Malkawi, and A. Al-Fuqaha, “A Systematic Review of Interference Mitigation Techniques in Current and Future UA V-Assisted Wireless Networks,”IEEE Open Journal of the Communications Society, vol. 5, pp. 2815–2846, 2024

    work page 2024

  5. [5]

    Integrating Terrestrial and Satellite Multibeam Systems Toward 6G: Techniques and Challenges for Interference Mitigation,

    D. Peng, D. He, Y . Li, and Z. Wang, “Integrating Terrestrial and Satellite Multibeam Systems Toward 6G: Techniques and Challenges for Interference Mitigation,”IEEE Wireless Communications, vol. 29, no. 1, pp. 24–31, 2022

    work page 2022

  6. [6]

    Interference Management for Integrated Sensing and Communication Systems: A Survey,

    Y . Niu, Z. Wei, L. Wang, H. Wu, and Z. Feng, “Interference Management for Integrated Sensing and Communication Systems: A Survey,”IEEE Internet of Things Journal, vol. 12, no. 7, pp. 8110–8134, 2025

    work page 2025

  7. [7]

    Interference Burden in Wireless Communi- cations: A Comprehensive Survey From PHY Layer Perspective,

    A. Tusha and H. Arslan, “Interference Burden in Wireless Communi- cations: A Comprehensive Survey From PHY Layer Perspective,”IEEE Communications Surveys & Tutorials, vol. 27, no. 4, pp. 2204–2246, 2025

    work page 2025

  8. [8]

    Upper Mid-Band Spectrum for 6G: Vision, Opportunity and Challenges,

    A. Bazzi, R. Bomfin, M. Mezzavilla, S. Rangan, T. Rappaport, and M. Chafii, “Upper Mid-Band Spectrum for 6G: Vision, Opportunity and Challenges,”IEEE Communications Magazine, vol. 64, no. 1, pp. 206– 212, 2026

    work page 2026

Show all 35 references
  1. [9]

    Spectrum Sharing Across Terrestrial and Non-Terrestrial Services in the FR3 Upper Mid- band,

    P. Testolina, E. Beshaj, M. Polese, and T. Melodia, “Spectrum Sharing Across Terrestrial and Non-Terrestrial Services in the FR3 Upper Mid- band,” in2025 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), 2025, pp. 1–9

    2025

  2. [10]

    6G Wireless Communications in 7– 24 GHz Band: Opportunities, Techniques, and Challenges,

    Z. Cui, P. Zhang, and S. Pollin, “6G Wireless Communications in 7– 24 GHz Band: Opportunities, Techniques, and Challenges,” in2025 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), 2025, pp. 1–8

    2025

  3. [11]

    The effect of narrowband interfer- ence on wideband wireless communication systems,

    A. Giorgetti, M. Chiani, and M. Win, “The effect of narrowband interfer- ence on wideband wireless communication systems,”IEEE Transactions on Communications, vol. 53, no. 12, pp. 2139–2149, 2005

    2005

  4. [12]

    Narrowband interference mitigation in OFDM systems,

    A. Batra and J. R. Zeidler, “Narrowband interference mitigation in OFDM systems,” inMILCOM 2008 - 2008 IEEE Military Communica- tions Conference, 2008, pp. 1–7

    2008

  5. [13]

    Multiple Interacting Narrowband Interferences Suppression Algorithm for OFDM Systems,

    C. Zhang, X. Hu, Y . Liu, and S. Wang, “Multiple Interacting Narrowband Interferences Suppression Algorithm for OFDM Systems,”IEEE Access, vol. 8, pp. 62 310–62 321, 2020

    2020

  6. [14]

    Narrowband Interference Mitigation Techniques: A Survey,

    M. Aygur, S. Kandeepan, A. Giorgetti, A. Al-Hourani, E. Arbon, and M. Bowyer, “Narrowband Interference Mitigation Techniques: A Survey,”IEEE Communications Surveys & Tutorials, pp. 1–1, 2025

    2025

  7. [15]

    Deep Learning- Based Multi-Tone Interference Suppression for Short Polar Codes,

    B. Husain, H. Mohamed, L. Haering, and A. Czylwik, “Deep Learning- Based Multi-Tone Interference Suppression for Short Polar Codes,” in 2021 IEEE 93rd V ehicular Technology Conference (VTC2021-Spring), 2021, pp. 1–5

    2021

  8. [16]

    Eliminating NB-IoT Interference to LTE System: A Sparse Machine Learning-Based Approach,

    S. Liu, L. Xiao, Z. Han, and Y . Tang, “Eliminating NB-IoT Interference to LTE System: A Sparse Machine Learning-Based Approach,”IEEE Internet of Things Journal, vol. 6, no. 4, pp. 6919–6932, 2019

    2019

  9. [17]

    Joint Impulsive Noise and Narrowband Inter- ference Mitigation in Time-Varying OFDM Communications,

    W. Gao and X. Kuai, “Joint Impulsive Noise and Narrowband Inter- ference Mitigation in Time-Varying OFDM Communications,”IEEE Communications Letters, vol. 29, no. 8, pp. 1799–1803, 2025

    2025

  10. [18]

    Improving Soft Decoding by Spectral Leakage Reduction in Presence of Narrow Band Interference in PLC,

    L. Mart ´ınez Marrero, L. Sadamori, S. Dominiak, U. Dersch, and J. T. G´omez, “Improving Soft Decoding by Spectral Leakage Reduction in Presence of Narrow Band Interference in PLC,”IEEE Access, vol. 7, pp. 79 491–79 502, 2019

    2019

  11. [19]

    NBI Spectral Leakage Mitigation Based on Compressed Sensing in OFDM Systems,

    K.-H. Kim, “NBI Spectral Leakage Mitigation Based on Compressed Sensing in OFDM Systems,”IEEE Transactions on Broadcasting, vol. 62, no. 2, pp. 409–416, 2016

    2016

  12. [20]

    Narrowband Interference Cancellation for OFDM Based on Deep Learning and Compressed Sensing,

    Y . Hu, S. Huang, L. Zhao, and M. Jiang, “Narrowband Interference Cancellation for OFDM Based on Deep Learning and Compressed Sensing,”IEEE Transactions on Signal Processing, vol. 73, pp. 1612– 1625, 2025

    2025

  13. [21]

    Interference Suppression Using Deep Learning: Current Approaches and Open Chal- lenges,

    T. Oyedare, V . K. Shah, D. J. Jakubisin, and J. H. Reed, “Interference Suppression Using Deep Learning: Current Approaches and Open Chal- lenges,”IEEE Access, vol. 10, pp. 66 238–66 266, 2022

    2022

  14. [22]

    Signal Recovery From Random Measurements Via Orthogonal Matching Pursuit,

    J. A. Tropp and A. C. Gilbert, “Signal Recovery From Random Measurements Via Orthogonal Matching Pursuit,”IEEE Transactions on Information Theory, vol. 53, no. 12, pp. 4655–4666, 2007

    2007

  15. [23]

    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

    2017

  16. [24]

    Model-Driven Deep Learning for Physical Layer Communications,

    H. He, S. Jin, C.-K. Wen, F. Gao, G. Y . Li, and Z. Xu, “Model-Driven Deep Learning for Physical Layer Communications,”IEEE Wireless Communications, vol. 26, no. 5, pp. 77–83, 2019

    2019

  17. [25]

    Narrowband Interference Detection via Deep Learning,

    C. P. Robinson, D. Uvaydov, S. D’Oro, and T. Melodia, “Narrowband Interference Detection via Deep Learning,” inICC 2023 - IEEE Inter- national Conference on Communications, 2023, pp. 6379–6384

    2023

  18. [26]

    Deep Learning Noncoherent UWB Receiver Design,

    S. Sharma, K. Deka, and M. Mandloi, “Deep Learning Noncoherent UWB Receiver Design,”IEEE Sensors Letters, vol. 5, no. 6, pp. 1–4, 2021

    2021

  19. [27]

    ”Machine LLRning

    O. Shental and J. Hoydis, “”Machine LLRning”: Learning to Softly Demodulate,” in2019 IEEE Globecom Workshops (GC Wkshps), 2019, pp. 1–7

    2019

  20. [28]

    Design and Implementation of a Low-Complexity Neural Detector for Correlated Noise,

    E. Kavvousanos and V . Paliouras, “Design and Implementation of a Low-Complexity Neural Detector for Correlated Noise,” in2025 IEEE Workshop on Signal Processing Systems (SiPS), 2025, pp. 1–5

    2025

  21. [29]

    Interference Mitigation in Turbo-Coded OFDM Systems Using Robust LLRs,

    S. Kalyani and K. Giridhar, “Interference Mitigation in Turbo-Coded OFDM Systems Using Robust LLRs,” in2008 IEEE International Conference on Communications, 2008, pp. 646–651

    2008

  22. [30]

    Liu,Research on the Key Technologies in Narrowband Interference and Impulsive Noise Mitigation and Cancellation, ser

    S. Liu,Research on the Key Technologies in Narrowband Interference and Impulsive Noise Mitigation and Cancellation, ser. Springer Theses. Singapore: Springer, 2021

    2021

  23. [31]

    Complex Signal Denoising and Interference Mitigation for Automotive Radar Using Convolutional Neural Networks,

    J. Rock, M. Toth, E. Messner, P. Meissner, and F. Pernkopf, “Complex Signal Denoising and Interference Mitigation for Automotive Radar Using Convolutional Neural Networks,” in2019 22th International Conference on Information Fusion (FUSION), 2019, pp. 1–8

    2019

  24. [32]

    Tensorflow: A system for large-scale machine learning,

    M. Abadi, P. Barham, J. Chen, Z. Chen, A. Davis, J. Dean, M. Devin, S. Ghemawat, G. Irving, M. Isard, M. Kudlur, J. Levenberg, R. Monga, S. Moore, D. G. Murray, B. Steiner, P. Tucker, V . Vasudevan, P. Warden, M. Wicke, Y . Yu, and X. Zheng, “Tensorflow: A system for large-sca...

    2016 arXiv

  25. [33]

    Hoydis, S

    J. Hoydis, S. Cammerer, F. Ait Aoudia, M. Nimier-David, L. Maggi, G. Marcus, A. Vem, and A. Keller, “Sionna,” 2022, https://nvlabs.github.io/sionna/

    2022

  26. [34]

    17.1 NVIDIA GB10: SoC Built for AI Acceleration,

    A. Skende, G. Rosseel, and N. Pinckney, “17.1 NVIDIA GB10: SoC Built for AI Acceleration,” in2026 IEEE International Solid-State Circuits Conference (ISSCC), vol. 69, 2026, pp. 298–300

    2026

  27. [35]

    TensorRT Implementations of Model Quantization on Edge SoC,

    Y . Zhou, Z. Guo, Z. Dong, and K. Yang, “TensorRT Implementations of Model Quantization on Edge SoC,” in2023 IEEE 16th International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MC- SoC), 2023, pp. 486–493

    2023

Pith tools

Reviewed July 10, 2026 · model on record in the stance chip above.