Pith. sign in

REVIEW 2 major objections 2 minor 53 references

Neural Augmentation of MIMO-OFDM Receivers for Universal LLR Reconstruction

T0 review · 2 major / 2 minor · reviewed 2026-06-30 · grok-4.3

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

Pith's one-line read A compact neural network can refine soft outputs from any MIMO-OFDM receiver into accurate log-likelihood ratios without knowing the cause of unreliability.

desk verdict The modular DNN add-on for refining LLRs from any MIMO-OFDM receiver is a practical idea, but the universality claim needs checks on impairments left out of training. read the letter →

arxiv 2606.29345 v1 pith:CTPWTHNF submitted 2026-06-28 eess.SP cs.ITmath.IT

Ory Eger , Nir Shlezinger This is my paper
classification eess.SPcs.ITmath.IT
keywords MIMO-OFDMneuralaugmentationLLRreconstructioninterferencecancellationwirelessreceivershardwareimpairmentssoftoutputschanneldecoding
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 sets out to demonstrate that existing MIMO-OFDM receivers, whether model-based or data-driven, can be augmented by a small neural network that corrects their soft outputs in two common failure modes: incomplete information from low-complexity detectors and degraded information from hardware impairments or synchronization issues. The augmentation operates without any details about the specific impairment and is trained to produce well-calibrated log-likelihood ratios that improve channel decoding. A sympathetic reader would care because this modular addition combines the reliability of classical receivers with the adaptability of learned methods while adding little computation, addressing the practical need for robust high-throughput wireless links under real-world distortions.

What carries the argument

Element-wise scaled convolutional neural network for learned interference cancellation across users and subcarriers, paired with a training procedure that targets accurate LLRs for soft decoding.

What would settle it

An experiment in which the augmented receiver shows no improvement in decoding error rate compared with the unaugmented receiver when tested on a previously unseen hardware impairment or synchronization error.

Watch

Extended reading notes

Core claim

An element-wise scaled convolutional neural network can be trained to perform learned interference cancellation across users and neighboring subcarriers on the log-likelihood ratios produced by diverse existing receivers, yielding calibrated outputs suitable for channel decoding in a task-agnostic way that requires no knowledge of the source of unreliability.

Load-bearing premise

An element-wise scaled convolutional neural network trained only on LLR examples can learn to cancel interference and produce accurate soft outputs for any unknown type of impairment without being given domain information about the impairment source.

Editorial extensions

If this is right

  • The augmentation improves decoding performance of multiple different receiver algorithms under challenging channel conditions.
  • It adds only minimal computational overhead to the original receiver.
  • It produces LLRs that are well-calibrated for use by standard channel decoders.
  • It works for both structurally incomplete soft information from reduced-complexity detectors and for degraded outputs caused by impairments.

Reading between the lines

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

  • The same modular augmentation structure could be applied to receiver chains in other multi-carrier or multi-antenna systems beyond the MIMO-OFDM setting examined here.
  • Because the network acts only on existing LLRs, it could be inserted after future receiver designs without requiring changes to their internal structure.
  • Training data collected from a range of known impairments might still allow generalization to entirely new impairment types not seen during training.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The paper proposes a modular DNN-based augmentation for MIMO-OFDM receivers that refines soft outputs into well-calibrated LLRs using an element-wise scaled convolutional neural network. The framework is presented as task-agnostic, requiring no knowledge of the impairment source (hardware nonlinearities, synchronization errors, or reduced-complexity detection), and is shown via numerical results to improve diverse base receivers with low overhead while supporting channel decoding.

Significance. If the claimed generalization and task-agnostic properties hold, the work offers a practical bridge between model-based and data-driven receivers, potentially enabling robust deployment in impaired wireless environments with minimal added complexity.

major comments (2)
  1. [Numerical Results] The central universality claim (abstract and introduction) requires explicit evidence that the CNN augmentation generalizes to impairment classes absent from the training distribution; without held-out impairment tests, the 'without any knowledge of the specific source of unreliability' property remains unverified and load-bearing for the task-agnostic assertion.
  2. [Proposed Method] The description of the element-wise scaled CNN and training algorithm (method section) does not include sufficient detail on how the scaling and loss encourage interference cancellation independent of impairment type; this leaves open whether the learned mapping reduces to in-distribution fitting rather than true cancellation.
minor comments (2)
  1. Clarify the precise definition and placement of the element-wise scaling factors with an equation or diagram, as the current description is high-level.
  2. [Numerical Results] Add error bars or multiple random seeds to the reported performance gains to allow assessment of statistical significance.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address the two major comments below and will incorporate revisions to strengthen the universality claims and methodological clarity.

read point-by-point responses

  1. Referee: [Numerical Results] The central universality claim (abstract and introduction) requires explicit evidence that the CNN augmentation generalizes to impairment classes absent from the training distribution; without held-out impairment tests, the 'without any knowledge of the specific source of unreliability' property remains unverified and load-bearing for the task-agnostic assertion.

    Authors: We agree that explicit held-out impairment tests would provide stronger support for the task-agnostic claim. The current experiments demonstrate consistent gains across multiple receiver types and impairment scenarios, but to directly verify generalization beyond the training distribution we will add new numerical results using impairment classes (e.g., previously unseen synchronization offsets and hardware nonlinearities) excluded from training. These results will be included in the revised manuscript. revision: yes

  2. Referee: [Proposed Method] The description of the element-wise scaled CNN and training algorithm (method section) does not include sufficient detail on how the scaling and loss encourage interference cancellation independent of impairment type; this leaves open whether the learned mapping reduces to in-distribution fitting rather than true cancellation.

    Authors: We will expand the method section with additional explanation of the element-wise scaling mechanism and the training loss. Specifically, we will clarify how the per-element scaling combined with the LLR-oriented loss promotes cancellation of structured interference across subcarriers and users in a manner that does not require explicit impairment modeling. We will also include a short discussion of why this design favors impairment-agnostic behavior over pure in-distribution fitting. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; derivation is self-contained learned augmentation

full rationale

The paper describes a modular DNN augmentation that refines LLRs from existing receivers via an element-wise scaled CNN trained to encourage accurate soft outputs. No equations, self-citations, or training procedures are shown that reduce a claimed prediction or universality property to a fitted parameter or prior result by construction. The task-agnostic claim is presented as an empirical property of the trained network rather than a definitional identity, and the abstract gives no indication that reported gains are forced by the same quantities used to fit the model. This is the normal case of an independent learned correction.

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

Abstract-only review; no explicit free parameters, axioms, or invented entities are stated. The DNN weights themselves constitute implicit fitted parameters, but none are enumerated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neural Augmentation of MIMO-OFDM Receivers for Universal LLR Reconstruction." pith.science (2026). https://pith.science/paper/CTPWTHNF

@misc{pith2026260629345,
  author       = {Pith},
  title        = {Pith review of: Neural Augmentation of MIMO-OFDM Receivers for Universal LLR Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CTPWTHNF}},
  note         = {Machine review of arXiv:2606.29345}
}
read the original abstract

The growing demands for higher throughput and cost-efficient wireless communications drive the need for receivers that are both simple to deploy and robust to hardware impairments and nonlinear environments. While classical model-based receivers and recently proposed deep neural network ( DNN) architectures provide complementary benefits, they either rely on simplified linear Gaussian assumptions, require considerable computational resources, or are tailored for a given setting and modulation. In this work, we propose a compact and modular DNN augmentation that universally refines the soft outputs of existing receivers (model-based or data-driven), addressing two distinct operating regimes: structurally incomplete soft information arising from reduced-complexity detectors, and degraded soft outputs caused by hardware impairments and synchronization errors. A key property of the proposed framework is its task-agnostic nature: operating without any knowledge of the specific source of unreliability, it produces well-calibrated log-likelihood ratios (LLRs) suitable for channel decoding. Our design leverages an element-wise scaled convolutional neural network tailored to perform learned interference cancellation across users and neighboring subcarriers, combined with a training algorithm that encourages accurate LLR s for soft channel decoding. Numerical results demonstrate that the proposed augmentation consistently improves diverse receiver algorithms in challenging channel conditions while incurring minimal overhead.

Figures

Figures reproduced from arXiv: 2606.29345 by the authors.

Figure 1
Figure 1. DNN for soft detection: (a) A DNN serves as an end-to-end soft detector, directly producing LLRs for the channel decoder; (b) The proposed approach: A DNN augments some primary soft detector, refining its output LLRs before channel decoding • Robust and scalable enhancement under mismatches: The proposed augmentation effectively compensates for a wide range of impairments, including ICI, CFO, IQMM and clipping disto… view at source ↗
Figure 2
Figure 2. System model for K users. Pink blocks denote hardware impairments arising in the RF front-end and digital front-end at both transmitter and receiver B. Soft Symbol Detection We focus on the soft detection receiver task, whose goal is to map the received signal Y i into probabilistic estimates of the transmitted bits. In particular, we aim to recover the bit-wise LLRs, denoted {l i b,k,n}, which quantify the reliabil… view at source ↗
Figure 3
Figure 3. ESCNN based receiver augmentation architecture at time i with the corresponding channel outputs Y i ∈ C B×nr , into enhanced LLRs, which we denote by L˜ i k ∈ R B×N for the kth user. The operation is comprised of two key stages: (i) feature pre-processing and (ii) learned processing that applies the ESCNN modules to the extracted features.. 1) Feature Pre-Processing: To support modulation scaling and LLR completion … view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: MI and BLER (LDPC code-rate 0.82) versus SNR, TDL-C channel, 16QAM modulation, 4 users, nr = 8 zero Doppler and low spatial correlation are assumed. Three different 16QAM scenarios are evaluated. In the first scenario, whose results are reported in [PITH_FULL_IMAGE:fi…
Figure 6
Figure 6. Figure 6: MI and BLER (LDPC code-rate 0.7) versus SNR, RMa channel, 16QAM modulation, 4 users, nr = 8, CFO: 4.5 kHz 0 10 20 30 40 SNR (dB) 0 0.2 0.4 0.6 0.8 1 MI Bit-wise Mutual Information 0 10 20 30 40 SNR (dB) 10-5 10-4 10-3 10-2 10-1 100 BLER Block Error Rate LMMSE SPHERE De…
Figure 7
Figure 7. Figure 7: MI and BLER (LDPC code-rate 0.7) versus SNR, UMi, 16QAM modulation, 4 users, nr = 8, CFO: 4.5 kHz [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: MI and BLER (LDPC code-rate 0.92) versus SNR, TDL-C channel, 16QAM modulation, 4 users, nr = 8, IQMM: 1 dB, 5 ◦ TABLE III Transferability between primary detectors: measured SNR at 10% BLER (dB) Primary detector LMMSE Augmentation None LMMSE Sphere DeepSIC DeepRx SNR (…
Figure 9
Figure 9. Figure 9: MI and BLER (LDPC code-rate 0.7) versus SNR, TDL-C channel with medium spatial correlation, 64QAM modulation, 4 users, nr = 4 simultaneously exercises both of its core capabilities: LLR reconstructionofthemissingbitsproducedbyRBSD,andro￾bustness to hardware-induced dis…
Figure 11
Figure 11. Figure 11: MI and BLER (LDPC code-rate 0.75) versus SNR, TDL-C channel, 64QAM modulation, 4 users, nr = 8, CFO: 2.25 kHz 4060 GPU. The results in Table V show that the overhead introduced by the augmentation is negligible compared to that of either of the primary detectors. Spec…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

53 extracted references · 4 canonical work pages

  1. [1]

    Learning to refine LLRs: Modular neural augmentation for MIMO-OFDM receivers,

    O. Eger and N. Shlezinger, “Learning to refine LLRs: Modular neural augmentation for MIMO-OFDM receivers,” in IEEE International Conference on Communications. Conference (ICC), 2026

    2026

  2. [2]

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

    C.-X. Wang et al., “On the road to 6G: Visions, requirements, key technologies, and testbeds,” IEEE Commun. Surveys Tuts., vol. 25, no. 2, pp. 905–974, 2023

    2023

  3. [3]

    A vision of 6G wireless systems: Applications, trends, technologies, and open research problems,

    W. Saad, M. Bennis, and M. Chen, “A vision of 6G wireless systems: Applications, trends, technologies, and open research problems,” IEEE Netw., vol. 34, no. 3, pp. 134–142, 2019

    2019

  4. [4]

    Massive MIMO detection techniques: A survey,

    M. A. Albreem, M. Juntti, and S. Shahabuddin, “Massive MIMO detection techniques: A survey,” IEEE Commun. Surveys Tuts., vol. 21, no. 4, pp. 3109–3132, 2019

    2019

  5. [5]

    Fifty years of MIMO detection: The road to large-scale MIMOs,

    S. Yang and L. Hanzo, “Fifty years of MIMO detection: The road to large-scale MIMOs,” IEEE Commun. Surveys Tuts., vol. 17, no. 4, pp. 1941–1988, 2015

    1941

  6. [6]

    A sphere decoding algorithm for MIMO channels,

    C.-Y. Hung and T.-H. Sang, “A sphere decoding algorithm for MIMO channels,” in Proc. IEEE ISSPIT, 2006, pp. 502–506

    2006

  7. [7]

    Iterative soft interference cancellation for multiple antenna systems,

    W.-J. Choi, K.-W. Cheong, and J. Cioffi, “Iterative soft interference cancellation for multiple antenna systems,” in Proc. IEEE WCNC, 2000

    2000

  8. [8]

    RF impairments in wireless transceivers: Phase noise, CFO, and IQ imbalance – A survey,

    A. Mohammadian and C. Tellambura, “RF impairments in wireless transceivers: Phase noise, CFO, and IQ imbalance – A survey,” IEEE Access, vol. 9, pp. 111 718–111 791, 2021

    2021

Show all 53 references
  1. [9]

    Channel estimation and performance analysis of one-bit massive MIMO systems,

    Y. Li, C. Tao, G. Seco-Granados, A. Mezghani, A. L. Swindlehurst, and L. Liu, “Channel estimation and performance analysis of one-bit massive MIMO systems,” IEEE Trans. Signal Process., vol. 65, no. 15, pp. 4075–4089, 2017

    2017

  2. [10]

    Asymptotic task-based quantization with application to massive MIMO,

    N. Shlezinger, Y. C. Eldar, and M. R. Rodrigues, “Asymptotic task-based quantization with application to massive MIMO,” IEEE Trans. Signal Process., vol. 67, no. 15, pp. 3995–4012, 2019

    2019

  3. [11]

    Iterative methods for cancellation of intercarrier interference in OFDM systems,

    A. F. Molisch, M. Toeltsch, and S. Vermani, “Iterative methods for cancellation of intercarrier interference in OFDM systems,” IEEE Trans. Veh. Technol., vol. 56, no. 4, pp. 2158–2167, 2007

    2007

  4. [12]

    Digital postcompensation using volterra series transfer function,

    F. P. Guiomar, J. D. Reis, A. L. Teixeira, and Pinto, “Digital postcompensation using volterra series transfer function,” IEEE Photon. Technol. Lett., vol. 23, no. 19, pp. 1412–1414, 2011

    2011

  5. [13]

    Deep learning for wireless communications: An emerging interdisciplinary paradigm,

    L. Dai, R. Jiao, F. Adachi, H. V. Poor, and L. Hanzo, “Deep learning for wireless communications: An emerging interdisciplinary paradigm,” IEEE Wireless Commun., vol. 27, no. 4, pp. 133–139, 2020

    2020

  6. [14]

    Neural network detection of data sequences in communication systems,

    N. Farsad and A. Goldsmith, “Neural network detection of data sequences in communication systems,” IEEE Trans. Signal Process., vol. 66, no. 21, pp. 5663–5678, 2018

    2018

  7. [15]

    Machine learning for MU-MIMO receive processing in OFDM systems,

    M. Goutay, F. Ait Aoudia, J. Hoydis, and J.-M. Gorce, “Machine learning for MU-MIMO receive processing in OFDM systems,” IEEE J. Sel. Areas Commun., vol. 39, no. 8, pp. 2318–2332, 2021

    2021

  8. [16]

    A deep learning receiver for non-linear transmitter,

    H. Farhadi, J. Haraldson, and M. Sundberg, “A deep learning receiver for non-linear transmitter,” IEEE Access, vol. 11, pp. 2796–2803, 2023

    2023

  9. [17]

    DeepRx: Fully convolutional deep learning receiver,

    M. Honkala, D. Korpi, and J. M. Huttunen, “DeepRx: Fully convolutional deep learning receiver,” IEEE Trans. Wireless Commun., vol. 20, no. 6, pp. 3925–3940, 2021

    2021

  10. [18]

    Deep learning OFDM receivers for improved power efficiency and coverage,

    J. Pihlajasalo, D. Korpi, M. Honkala, J. M. Huttunen, T. Riihonen, J. Talvitie, A. Brihuega, M. A. Uusitalo, and M. Valkama, “Deep learning OFDM receivers for improved power efficiency and coverage,” IEEE Trans. Wireless Commun., vol. 22, no. 8, pp. 5518–5535, 2023

    2023

  11. [19]

    Deep learning based OFDM physical-layer receiver with multidilated convolutions,

    J. Pihlajasalo, D. Korpi, E. Raninen, and M. Valkama, “Deep learning based OFDM physical-layer receiver with multidilated convolutions,” in IEEE Conference on Standards for Communications and Networking (CSCN), 2025

    2025

  12. [20]

    Joint detection and decoding: A graph neural network approach,

    J. Clausius, M. Rübenacke, D. Tandler, and S. Ten Brink, “Joint detection and decoding: A graph neural network approach,” IEEE Trans. Commun., vol. 73, no. 10, pp. 9043–9058, 2025

    2025

  13. [21]

    Graph Neural Networks-Enhanced expectation propagation algorithm for MIMO turbo receiver,

    X. Zhou, J. Zhang, C.-K. Wen, S. Jin, and S. Han, “Graph Neural Networks-Enhanced expectation propagation algorithm for MIMO turbo receiver,” IEEE Trans. Signal Process., vol. 71, pp. 3458–3473, 2023

    2023

  14. [22]

    Hypernetwork- enhanced GEPNet for MIMO-OFDM receiver with imperfect CSI,

    X. Li, J. Zhang, X. Zhou, Y. Li, and S. Jin, “Hypernetwork- enhanced GEPNet for MIMO-OFDM receiver with imperfect CSI,” IEEE Wireless Commun. Lett., vol. 14, 2025

    2025

  15. [23]

    A neural receiver for 5G NR multi-user MIMO,

    S. Cammerer, F. Aït Aoudia, J. Hoydis, A. Oeldemann, A. Roessler, T. Mayer, and A. Keller, “A neural receiver for 5G NR multi-user MIMO,” in IEEE Globecom Workshops, 2023, pp. 329–334

    2023

  16. [24]

    Unsupervised linear and nonlinear channel equalization and decoding using variational autoencoders,

    A. Caciularu and D. Burshtein, “Unsupervised linear and nonlinear channel equalization and decoding using variational autoencoders,” IEEE Trans. on Cogn. Commun. Netw., vol. 6, no. 3, pp. 1003–1018, 2020

    2020

  17. [25]

    Comm-transformer: A robust deep learning-based receiver for OFDM system under TDL chan- nel,

    Y. Xie, K. C. Teh, and A. C. Kot, “Comm-transformer: A robust deep learning-based receiver for OFDM system under TDL chan- nel,” IEEE Trans. Commun., vol. 72, no. 4, pp. 2014–2026, 2023

    2014

  18. [26]

    Convolutional self-attention-based multi-user MIMO demapper,

    A. Michon, F. Aït Aoudia, and K. P. Srinath, “Convolutional self-attention-based multi-user MIMO demapper,” in IEEE International Conference on Communications (ICC), 2022, pp. 2621–2626

    2022

  19. [27]

    Deep learning in wireless communication receivers: A survey,

    S. R. Doha and A. Abdelhadi, “Deep learning in wireless communication receivers: A survey,” IEEE Access, vol. 13, pp. 113 586–113 605, 2025

    2025

  20. [28]

    Design of a standard-compliant real-time neural receiver for 5G NR,

    R. Wiesmayr, S. Cammerer, F. Aït Aoudia, J. Hoydis, J. Zakrzewski, and A. Keller, “Design of a standard-compliant real-time neural receiver for 5G NR,” in IEEE International Conf. Machine Learning and Communications and Networking (MLCN), 2025

    2025

  21. [29]

    Adaptive and flexible model-based AI for deep receivers in dynamic channels,

    T. Raviv, S. Park, O. Simeone, Y. C. Eldar, and N. Shlezinger, “Adaptive and flexible model-based AI for deep receivers in dynamic channels,” IEEE Wireless Commun., vol. 31, no. 4, pp. 163–169, 2024

    2024

  22. [30]

    Model-based deep learning,

    N. Shlezinger and Y. C. Eldar, “Model-based deep learning,” Foundations and Trends® in Signal Processing, vol. 17, no. 4, pp. 291–416, 2023

    2023

  23. [31]

    Comprehensive review of deep unfolding techniques for next-generation wireless communication systems,

    S. Deka, K. Deka, N. T. Nguyen, S. Sharma, V. Bhatia, and N. Rajatheva, “Comprehensive review of deep unfolding techniques for next-generation wireless communication systems,” IEEE Internet Things J., vol. 13, no. 6, pp. 10 379–10 406, 2026

    2026

  24. [32]

    Deep unfolding for communications systems: A survey and some new directions,

    A. Balatsoukas-Stimming and C. Studer, “Deep unfolding for communications systems: A survey and some new directions,” arXiv preprint arXiv:1906.05774, 2019

    1906

  25. [33]

    Deep unfolding: Recent developments, theory, and design guidelines,

    N. Shlezinger, S. Segarra, Y. Zhang, D. A vrahami, Z. Davidov, T. Routtenberg, and Y. C. Eldar, “Deep unfolding: Recent developments, theory, and design guidelines,” arXiv preprint arXiv:2512.03768, 2025

    2025

  26. [34]

    A comprehensive survey of knowledge-driven deep learning for intelligent wireless network optimization in 6G,

    R. Sun, N. Cheng, C. Li, W. Quan, H. Zhou, Y. Wang, W. Zhang, and X. Shen, “A comprehensive survey of knowledge-driven deep learning for intelligent wireless network optimization in 6G,” IEEE Commun. Surveys Tuts., vol. 28, pp. 1099–1135, 2025

    2025

  27. [35]

    EqDeepRx: Learning a scalable MIMO receiver,

    M. Honkala, D. Korpi, E. Raninen, and J. M. Huttunen, “EqDeepRx: Learning a scalable MIMO receiver,” arXiv preprint arXiv:2602.11834, 2026

    2026

  28. [36]

    Deep learning- based low complexity MIMO detection via partial MAP,

    L. Bai, Q. Zeng, R. Han, J. Choi, and W. Zhang, “Deep learning- based low complexity MIMO detection via partial MAP,” IEEE Trans. Wireless Commun., vol. 24, no. 3, pp. 2126–2139, 2025

    2025

  29. [37]

    Learning successive interference cancellation for low-complexity soft-output MIMO detection,

    B. Fesl and F. Capar, “Learning successive interference cancellation for low-complexity soft-output MIMO detection,” arXiv preprint arXiv:2601.16586, 2026

    2026

  30. [38]

    “Machine LLRning

    O. Shental and J. Hoydis, ““Machine LLRning”: Learning to softly demodulate,” in IEEE Globecom Workshops, 2019

    2019

  31. [39]

    Adaptive neural signal detection for massive MIMO,

    M. Khani, M. Alizadeh, J. Hoydis, and P. Fleming, “Adaptive neural signal detection for massive MIMO,” IEEE Trans. Wireless Commun., vol. 19, no. 8, pp. 5635–5648, 2020

    2020

  32. [40]

    Model-driven deep learning for MIMO detection,

    H. He, C.-K. Wen, S. Jin, and G. Y. Li, “Model-driven deep learning for MIMO detection,” IEEE Trans. Signal Process., vol. 68, pp. 1702–1715, 2020

    2020

  33. [41]

    GNN-assisted BiG-AMP: Joint channel estimation and data detection for massive MIMO receiver,

    Z. Liu, N. Wu, D. He, W. Yuan, Y. Li, and T. Q. Quek, “GNN-assisted BiG-AMP: Joint channel estimation and data detection for massive MIMO receiver,” IEEE Trans. Wireless Commun., vol. 24, no. 6, pp. 4631–4646, 2025

    2025

  34. [42]

    DeepSIC: Deep soft interference cancellation for multiuser MIMO detection,

    N. Shlezinger, R. Fu, and Y. C. Eldar, “DeepSIC: Deep soft interference cancellation for multiuser MIMO detection,” IEEE Trans. Wireless Commun., vol. 20, no. 2, pp. 1349–1362, 2021

    2021

  35. [43]

    Deep learning based successive interference cancellation for the non-orthogonal downlink,

    T. Van Luong, N. Shlezinger, C. Xu, T. M. Hoang, Y. C. Eldar, and L. Hanzo, “Deep learning based successive interference cancellation for the non-orthogonal downlink,” IEEE Trans. Veh. Technol., vol. 71, no. 11, pp. 11 876–11 888, 2022

    2022

  36. [44]

    Model-based deep learning receiver design for rate-splitting multiple access,

    R. C. Loli, O. Dizdar, B. Clerckx, and C. Ling, “Model-based deep learning receiver design for rate-splitting multiple access,” IEEE Trans. Wireless Commun., vol. 22, no. 11, pp. 8352–8365, 2023

    2023

  37. [45]

    Study on channel model for frequencies from 0.5 to 100 GHz,

    3rd Generation Partnership Project (3GPP), “Study on channel model for frequencies from 0.5 to 100 GHz,” Tech. Rep. 38.901, 2025, https://www.3gpp.org/dynareport/38901.htm

    2025

  38. [46]

    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

  39. [47]

    Quadriga: A 3-D multi-cell channel model with time evolution for enabling virtual field trials,

    S. Jaeckel, L. Raschkowski, K. Börner, and L. Thiele, “Quadriga: A 3-D multi-cell channel model with time evolution for enabling virtual field trials,” IEEE Trans. Antennas Propag., vol. 62, no. 6, pp. 3242–3256, 2014

    2014

  40. [48]

    Joint estimation of carrier frequency offset and channel impulse response for linear periodic channels,

    R. Shaked, N. Shlezinger, and R. Dabora, “Joint estimation of carrier frequency offset and channel impulse response for linear periodic channels,” IEEE Trans. Commun., vol. 66, no. 1, pp. 302–319, 2017

    2017

  41. [49]

    Asynchronous online adaptation via modular drift detection for deep receivers,

    N. Uzlaner, T. Raviv, N. Shlezinger, and K. Todros, “Asynchronous online adaptation via modular drift detection for deep receivers,” IEEE Trans. Wireless Commun., vol. 24, no. 5, pp. 4454–4468, 2025

    2025

  42. [50]

    Algorithm unrolling: Inter- pretable, efficient deep learning for signal and image processing,

    V. Monga, Y. Li, and Y. C. Eldar, “Algorithm unrolling: Inter- pretable, efficient deep learning for signal and image processing,” IEEE Signal Process. Mag., vol. 38, no. 2, pp. 18–44, 2021

    2021

  43. [51]

    Online learning of modular bayesian deep receivers: Single-step adaptation with streaming data,

    Y. Gusakov, O. Simeone, T. Routtenberg, and N. Shlezinger, “Online learning of modular bayesian deep receivers: Single-step adaptation with streaming data,” IEEE Trans. Signal Process., vol. 74, pp. 2192–2207, 2026

    2026

  44. [52]

    Modular hypernetworks for scalable and adaptive deep MIMO receivers,

    T. Raviv and N. Shlezinger, “Modular hypernetworks for scalable and adaptive deep MIMO receivers,” IEEE Open Journal of Signal Processing, vol. 6, pp. 256–265, 2025

    2025

  45. [53]

    Performance assessment of MIMO-BICM demodulators based on mutual information,

    P. Fertl, J. Jalden, and G. Matz, “Performance assessment of MIMO-BICM demodulators based on mutual information,” IEEE Trans. Signal Process., vol. 60, no. 3, pp. 1366–1382, 2011

    2011

Pith tools

Reviewed June 30, 2026 · model on record in the stance chip above.