Pith. sign in

REVIEW 3 major objections 5 minor 33 references

A JEPA-Based Field-Layer World Model for Bridging Channel Prediction and Estimation

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read JEPA-based latent-field prediction preserves the spatial structure beamforming needs.

desk verdict A genuinely useful JEPA-for-wireless paper with a clean latent-field idea and an honest experimental write-up; the cross-band results are real but rest on a shared-geometry simulation that should be disclosed more prominently. read the letter →

arxiv 2608.10222 v1 pith:5W6IQQTO submitted 2026-08-10 eess.SP

Yuzhi Yang , Brahim Mefgouda , Hang Zou , Lina Bariah , Anis Bara , Yuhuan Lu , Hao Zhang , MérouaneDebbah This is my paper · ORCID
classification eess.SP
keywords channelstateinformationpredictionreconstructionJEPAworldmodelMIMO-OFDMbeamformingcross-bandtransfer
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

This paper tries to establish that wireless channel prediction is more reliable and more useful when done in a learned latent propagation-field space rather than at the level of raw complex coefficients. The authors build a JEPA-based world model that maps multi-resolution CSI observations across several carrier bands into a shared latent field, predicts masked or future field states there, and then fuses the predicted latent with sparse current pilots to reconstruct the channel. The claimed payoff is concrete: improved symbol detection and beamforming performance close to an oracle that uses the true current latent, even when coefficient-level NMSE improvements over an end-to-end baseline are small. A reader should care because the approach separates predictable propagation structure from phase-sensitive detail, which points toward lower pilot overhead and transfer of channel knowledge across frequency bands.

What carries the argument

The central object is the latent propagation-field representation: a shared latent space in which each time–carrier-band cell is encoded as a small set of tokens by scale-specific tokenizer heads, with a shared JEPA backbone predicting masked or future cells. The 'field layer' means propagation-level structure — delay-angle support, dominant paths, spatial covariance, and dominant eigenspaces — rather than individual complex coefficients. The training objective combines latent prediction with a diversity regularizer and uses an exponential-moving-average target tokenizer, so the model encodes what is predictable across time and band while avoiding collapsed representations. The load-bearing identity is that beamforming quality follows the Rayleigh quotient of the true channel covariance evaluated at the reconstructed beamformer, so preserving the dominant eigenvector and covariance of the latent field converts directly into beamforming gain. A frozen-reference incremental alignment procedure lets new observation scales be added without retraining the shared backbone.

What would settle it

Regenerate or measure the scenario with band-dependent path visibility — ray-tracing each carrier band independently so paths can appear or disappear with frequency, or using measured sub-6-GHz and mmWave channel pairs — and compare FWM against the end-to-end baseline on bands without pilots; if the beamforming gain and dominant-eigenvector similarity fall to baseline levels, the claim that the latent field transfers shared spatial structure across bands is falsified.

Watch

Extended reading notes

Core claim

The central claim is that a shared latent propagation state, learned by JEPA-style masked prediction, carries the spatial structure that matters for communication tasks, and that this structure survives prediction even when individual CSI coefficients do not. On a ray-traced city scenario with five sub-6-GHz and three mmWave carrier bands, the paper's FWM gives each observation scale its own tokenizer head, maps all scales into a common latent field, predicts the current-time field from history, and uses the predicted latent as a structured prior for a reconstruction head that also takes sparse noisy pilots. In the cross-band setting, where some carrier bands have no pilots, the predicted latent is calibrated using pilot-conditioned latents from the observed bands. The paper reports that FWM and a fully end-to-end trained baseline show comparable reconstruction NMSE, yet FWM retains higher dominant-eigenvector similarity and lower covariance NMSE, yielding band-wise beamforming ratios close to a current-latent oracle. The authors conclude that JEPA pretraining better preserves the dominant spatial eigendirection and covariance structure of the missing-band channel, which govern the beamforming Rayleigh quotient.

Load-bearing premise

The load-bearing premise is that all eight simulated carrier bands see exactly the same propagation paths, gains, delays, and angles, so the latent field learned from one band is genuinely shared by the others; if real frequency-dependent effects make paths appear or disappear across bands, the cross-band transfer gains may not survive.

Editorial extensions

If this is right

  • Fewer pilots are needed for the same reconstruction quality, because the predicted latent field already supplies the spatial subspace that sparse pilots would otherwise have to recover.
  • Carrier bands with no current pilots can still be reconstructed from observed bands plus the latent prior, which bears directly on FDD-style reciprocity problems.
  • Coefficient-level NMSE is a poor proxy for task value: models with similar NMSE can differ sharply in beamforming gain, so communication-level metrics should accompany CSI metrics.
  • New observation formats can be introduced incrementally with a frozen reference branch, avoiding retraining the whole model when resolution or bandwidth changes.
  • Adding raw-CSI grounding objectives to the latent target is counterproductive for prediction; diversity-based JEPA pretraining outperforms direct CSI, amplitude, and diffusion grounding.

Reading between the lines

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

  • Our inference: if propagation-geometry preservation is the operative mechanism, then the FWM latent should also support other geometry-driven tasks such as localization, sensing, and beam management; the paper lists these as future work rather than demonstrated results.
  • Our inference: the cross-band gains are likely to shrink under frequency-dependent path visibility, because the dataset deliberately reuses one path set across all bands; a direct test is to regenerate channels with per-band ray tracing or measured dual-band data and compare beamforming ratios.
  • Our inference: the evaluation principle here — judge predicted CSI by what it does to beamforming and detection, not by NMSE — could reasonably be adopted by the wider channel-prediction literature, but the paper only demonstrates it in its own scenario.
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

3 major / 5 minor

Summary. The paper proposes FWM, a JEPA-based latent world model for MIMO-OFDM CSI. It maps multi-resolution CSI observations from several carrier bands into a shared latent propagation-field space via scale-specific tokenizer heads, trains a latent prediction backbone with a masked/future prediction objective, and uses the predicted latent field as a structured prior for downstream channel reconstruction, either per band or across bands with missing pilots. The authors evaluate latent-prediction NMSE, multi-scale alignment strategies, and downstream reconstruction NMSE, symbol error rate, and single-stream beamforming metrics on a Sionna RT dataset generated in a Chicago city-center scenario. The central empirical claim is that JEPA pretraining preserves the dominant spatial eigendirection and covariance structure of missing-band channels, yielding substantial beamforming gains even when coefficient-level NMSE is comparable to an end-to-end baseline.

Significance. If the results hold beyond the specific simulation setup, the paper makes a useful contribution by shifting channel prediction from coefficient-level regression to latent predictive modeling and by demonstrating that task-relevant spatial structure can transfer across observation scales and carrier bands. The multi-scale tokenizer architecture and incremental alignment strategy are clearly formulated, and the downstream fusion of a predicted latent prior with sparse pilots is a sensible design. The beamforming evaluation metrics (eigenvector similarity, covariance NMSE, and Rayleigh quotient ratio) are appropriate and provide a credible explanation for the observed gap between reconstruction NMSE and beamforming gain. The provision of dataset and code in the footnote is a strength. However, the cross-band experiments rely on a data-generation protocol that reuses a single path layout across all bands, which severely limits the generality of the stated conclusions about physics-aware field representation.

major comments (3)
  1. [Section VI.A and Section VIII.C] The central cross-band beamforming claim is established only under a dataset where all eight carrier bands share exactly the same propagation paths, gains, delays, and angles, computed once at 3.5 GHz. As the authors acknowledge in Section VI.A, the dataset does not model frequency-dependent material responses or band-dependent path visibility (paths appearing or disappearing due to penetration, reflection, diffraction, or blockage). In this setting, the missing-band channel is essentially a phase-shifted, carrier-dependent version of the observed-band channel, which is precisely the regime most favorable to a latent propagation-field prior that preserves the spatial support. The conclusion in Section IX that FWM 'provides a step from channel approximation toward physics-aware field representation' therefore goes beyond what the experiment can support. To make the cross-band claim load-bearing, the authors should test the method on data where the path set itself varies across bands, for example by running Sionna RT independently at each carrier frequency or by explicitly adding/removing paths per band. Without such a test, the spatial-subspace preservation in Fig. 8 and the associated beamforming gains may be an artifact of the shared-geometry assumption.
  2. [Section VIII.A] The statement that 'We do not report pure-prediction baselines because none converged in our experiments, including [15], [16] and the prediction-only counterpart of the proposed method' is an unsupported empirical claim that is important for positioning the paper. No convergence curves, final NMSE values, or implementation details are given for these baselines, so the reader cannot tell whether the failure is due to the dataset's phase perturbations, an implementation issue, or a fundamental property of coefficient-level prediction. Since the paper motivates JEPA precisely by arguing that raw-CSI prediction is fragile, this baseline evidence is relevant. The authors should either provide the quantitative results and learning curves for these baselines or soften the claim to an observation about their specific setup rather than a general statement about prior methods.
  3. [Section VIII.C, Fig. 9] The paper asserts in the conclusion that the latent prior provides 'substantial band-wise beamforming gain even when its NMSE advantage over sufficiently trained end-to-end baselines is small or inconsistent.' However, the comparison between FWM and DT uses different training schedules (FWM downstream heads trained for 200 epochs, DT trained for 500 epochs), and the text states that the DT training schedule is longer because 'learning the complete predictor–reconstructor from scratch is more difficult.' This is a reasonable choice, but it makes the comparison less clean; the claim that JEPA is responsible for the beamforming gain would be stronger if the DTs were trained until their loss plateaued and the same number of epochs were used for the downstream stage. Please clarify whether the 500-epoch DT has converged and report its final NMSE and beamforming metrics at the same evaluation points as FWM.
minor comments (5)
  1. [Section VIII.A] The baseline 'SP' is mentioned ('SP and SE are retained as single-purpose baselines') but never defined; only SE is introduced in the baseline list. Please define SP or remove it.
  2. [Section VI.C] The sentence 'In all cases, the allotted training schedule is sufficient for convergence' is an assertion without evidence. Presenting learning curves or final loss/epoch plots for the main JEPA and downstream training runs would make this claim verifiable.
  3. [Figure 9] The downstream correspondence results in Fig. 9 are described only qualitatively in the text. Reporting numeric values for the grounding-objective ablations and for the training-strategy comparison would allow readers to judge the magnitude of the differences and would also avoid the impression that the conclusions rest on visual inspection of curves.
  4. [Appendix B] The statement 'The target branch is stopped from gradient back-propagation' should read 'gradient backpropagation' or 'gradient back-propagation' for consistency; also consider specifying whether the target tokenizer's batch normalization or layer normalization statistics are updated by EMA or kept frozen.
  5. [Author line] The author name 'M ´erouane' has a formatting issue with the accent; it should appear as 'Mérouane'.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the downstream evaluation is held out and externally supervised; the shared-path dataset is an acknowledged realism caveat, not a circular construction.

full rationale

The paper's load-bearing chain is not circular. The JEPA pretraining objective (Eq. 8) is a masked latent-prediction loss against EMA target tokenizer outputs, but the downstream reconstruction head is trained separately under Eq. (17) using true channel matrices, and all task-level metrics (beamforming ratio, dominant-eigenvector similarity, covariance NMSE, SER) are computed from held-out true channels via Eqs. (22)-(25). Whole UE trajectories are split into disjoint train/test sets, so the pretraining and downstream objectives cannot peek at the test channel. The beamforming claim in Sec. VIII-C ('JEPA pretraining better preserves the dominant spatial eigendirection and covariance structure of the missing-band channel') is an empirical reading of measured S and covariance NMSE, not an equation that reduces to the model's inputs. The shared-path dataset construction in Sec. VI.A ('path gains, delays, and angles ... are then reused to synthesize the channel responses at all considered carrier frequencies') makes cross-band transfer favorable, and the paper explicitly says it 'does not model frequency-dependent material responses or band-dependent path visibility'; this is an acknowledged generalization limitation and should be weighed as such, not as a circularity. The only author-overlapping citations (e.g., [1]) appear in related-work background and are not invoked to justify the field-layer mechanism, the JEPA objective, or the beamforming advantage. No fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The framework introduces a learned latent propagation field as the central abstraction, with no external physical grounding. The cross-band results rely on the synthetic assumption that all bands share one path layout. Several hyperparameters and the perturbation scale are hand-selected and materially influence the reported gains.

free parameters (5)
  • Diversity regularization weight lambda_div = 0.1
    Selected by single-scale ablation on 32x16 observations; without it, latent-prediction NMSE worsens by more than an order of magnitude, so the reported gains depend on this choice.
  • Grounding loss weight lambda_g = 0.2 (ablations only)
    Set in preliminary experiments for CSI-domain grounding ablations; not used in the default FWM objective.
  • EMA momentum mu = 0.95
    Chosen by hand for target tokenizer updates; standard JEPA momentum value.
  • Historical masking schedule parameters = initial p=0.3, final Beta(4.0,1.2) mix
    The progressive masking schedule controls prediction difficulty during pretraining and is fixed by the authors rather than derived.
  • UE position perturbation standard deviation sigma_p = 5 cm
    Chosen as the mobility uncertainty scenario; at 28 GHz this is 4.67 lambda and makes adjacent-sample phase correlation near zero, which is the premise that raw coefficient prediction fails.
assumptions (4)
  • domain assumption There exists a shared underlying propagation state S_t such that every CSI observation is H_t^(omega)=O_omega(S_t)+N_t^(omega).
    Equation (1) defines the field-layer abstraction; the entire multi-scale alignment and cross-band transfer rests on this common-state existence.
  • domain assumption All carrier bands share the same set of propagation paths, gains, delays, and angles computed at 3.5 GHz; no frequency-dependent path visibility or material response.
    Section VI.A dataset generation; this makes cross-band structure artificially shared and is explicitly acknowledged by the authors.
  • domain assumption Position perturbations are independent Gaussian with sigma=5 cm, so adjacent-sample phase correlation is near zero.
    Section VI.B; it ensures raw-CSI phase prediction is infeasible, motivating the latent approach. If perturbations were smaller or correlated, the premise weakens.
  • ad hoc to paper JEPA target embeddings from an EMA teacher provide a stable supervision target for latent prediction.
    Section IV.B; a design choice standard in JEPA literature, not independently justified for wireless channels.
invented entities (1)
  • Latent propagation-field state S_t / field layer
    purpose: Serves as the central modeled object; heterogeneous CSI observations are mapped to this shared latent space and predictions are made there.
    The paper explicitly defines field layer as an abstraction (Section III) and states no latent coordinate is uniquely identifiable with a path, angle, or delay. Predictive and downstream performance are internal to the same synthetic dataset; no external falsifiable handle is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A JEPA-Based Field-Layer World Model for Bridging Channel Prediction and Estimation." pith.science (2026). https://pith.science/paper/5W6IQQTO

@misc{pith2026260810222,
  author       = {Pith},
  title        = {Pith review of: A JEPA-Based Field-Layer World Model for Bridging Channel Prediction and Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5W6IQQTO}},
  note         = {Machine review of arXiv:2608.10222}
}
read the original abstract

Channel state information (CSI) acquisition, reconstruction, and prediction are fundamental yet costly tasks in modern MIMO-OFDM wireless systems. Direct coefficient-level prediction of raw CSI is fragile in realistic propagation environments, since small spatial perturbations, local scattering changes, and phase variations can cause large errors in the complex channel domain. However, the underlying wireless propagation field still contains stable and predictable structures that can be exploited across time, frequency, antenna, and carrier dimensions. Motivated by this observation, we propose a JEPA-based field-layer world model (FWM) that learns a shared latent propagation state from multi-resolution CSI observations across the considered carrier bands and predicts its task-relevant evolution in the latent domain. The proposed FWM maps multiple CSI observation resolutions to a shared latent propagation-field space through scale-specific tokenizer heads. A latent prediction backbone is then trained to infer masked or future field states, while an incremental multi-scale alignment strategy allows new observation scales to be incorporated without retraining the entire model from scratch. For downstream reconstruction, the predicted latent field is used as a structured prior and combined with sparse current pilots. Experiments on single-band and cross-band reconstruction demonstrate improved symbol detection and, more notably, substantial beamforming gains despite modest NMSE improvements, indicating that FWM captures task-relevant spatial propagation structure beyond coefficient-wise CSI fitting.

Figures

Figures reproduced from arXiv: 2608.10222 by the authors.

Figure 1
Figure 1. Overview of the JEPA channel representation and prediction architecture. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the downstream channel estimation task. The cross-band [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. The scenario considered for generating dataset based on the actual maps [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Single-scale objective ablation on 32 × 16 observations, measured by test JEPA prediction NMSE [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Test JEPA prediction NMSE under different multi-scale training [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 7
Figure 7. Figure 7: Single-band downstream performance without frequency alignment [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Frequency-alignment performance on the missing carrier bands for [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Downstream reconstruction performance corresponding to the pretraining studies, evaluated versus fixed-observation NMSE. [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 11
Figure 11. Figure 11: The structure of the encoder and predictor for JEPA. [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: The structure of the reconstruction head for the downstream task. [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 17 canonical work pages

  1. [15]

    ODE-Former for mobile channel prediction: A novel learning structure leveraging the physics continuity,

    Z. Xiao, Y . Huang, Y . Xu, et al., “ODE-Former for mobile channel prediction: A novel learning structure leveraging the physics continuity,” IEEE Wireless Communications Letters, 2025

    work page 2025

  2. [16]

    CSI-4CAST: A hybrid deep learning model for CSI prediction with comprehensive robustness and generalization testing,

    S. Cheng, R. Zandehshahvar, H. Zhao, et al., “CSI-4CAST: A hybrid deep learning model for CSI prediction with comprehensive robustness and generalization testing,”IEEE Transactions on Machine Learning in Communications and Networking, 2026

    work page 2026

  3. [1]

    Telecom world models: Unifying digital twins, foundation models, and predictive planning for 6G,

    H. Zou, Y . Yang, L. Bariah, et al., “Telecom world models: Unifying digital twins, foundation models, and predictive planning for 6G,”arXiv preprint arXiv:2604.06882, 2026

    arXiv 2026

  4. [2]

    Structured latent dynam- ics in wireless CSI via homomorphic world models,

    S. Naoumi, M. Bennis, and M. Chafii, “Structured latent dynam- ics in wireless CSI via homomorphic world models,”arXiv preprint arXiv:2603.20048, 2026

    arXiv 2026

  5. [3]

    Self-supervised learning from images with a joint-embedding predictive architecture,

    M. Assran, Q. Duval, I. Misra, et al., “Self-supervised learning from images with a joint-embedding predictive architecture,”arXiv preprint arXiv:2301.08243, 2023

    arXiv 2023

  6. [4]

    Revisiting feature predic- tion for learning visual representations from video,

    A. Bardes, Q. Garrido, J. Ponce, et al., “Revisiting feature predic- tion for learning visual representations from video,”arXiv preprint arXiv:2404.08471, 2024

    arXiv 2024

  7. [5]

    V-JEPA 2.1: Unlock- ing dense features in video self-supervised learning,

    L. Mur-Labadia, M. Muckley, A. Bar, et al., “V-JEPA 2.1: Unlock- ing dense features in video self-supervised learning,”arXiv preprint arXiv:2603.14482, 2026

    arXiv 2026

  8. [6]

    AI reasoning for wireless communica- tions and networking: A survey and perspectives,

    H. Luo, Y . Yan, Y . Bian, et al., “AI reasoning for wireless communica- tions and networking: A survey and perspectives,”ACM Comput. Surv., 2025

    work page 2025

Show all 33 references
  1. [7]

    Wireless large AI model: Shaping the AI-native future of 6G and beyond,

    F. Zhu, X. Wang, S. Jiang, et al., “Wireless large AI model: Shaping the AI-native future of 6G and beyond,”arXiv preprint arXiv:2504.14653, 2025

    2025 arXiv

  2. [8]

    Deep learning for massive MIMO CSI feedback,

    C.-K. Wen, W.-T. Shih, and S. Jin, “Deep learning for massive MIMO CSI feedback,”IEEE Wireless Communications Letters, vol. 7, no. 5, pp. 748–751, 2018

    2018

  3. [9]

    Deep learning-based CSI feedback approach for time-varying massive MIMO channels,

    T. Wang, C.-K. Wen, S. Jin, et al., “Deep learning-based CSI feedback approach for time-varying massive MIMO channels,”IEEE Wireless Communications Letters, vol. 8, no. 2, pp. 416–419, 2018

    2018

  4. [10]

    Accurate channel prediction based on transformer: Making mobility negligible,

    H. Jiang, M. Cui, D. W. K. Ng, et al., “Accurate channel prediction based on transformer: Making mobility negligible,”IEEE Journal on Selected Areas in Communications, vol. 40, no. 9, pp. 2717–2732, 2022

    2022

  5. [11]

    Spatio-temporal neural network for chan- nel prediction in massive MIMO-OFDM systems,

    G. Liu, Z. Hu, L. Wang, et al., “Spatio-temporal neural network for chan- nel prediction in massive MIMO-OFDM systems,”IEEE Transactions on Communications, vol. 70, no. 12, pp. 8003–8016, 2022

    2022

  6. [12]

    Machine learning-based channel prediction in wideband massive mimo systems with small overhead for online training,

    B. Ko, H. Kim, M. Kim, et al., “Machine learning-based channel prediction in wideband massive mimo systems with small overhead for online training,”IEEE Open Journal of the Communications Society, vol. 5, pp. 5289–5305, 2024

    2024

  7. [13]

    Linformer: A linear-based lightweight transformer architecture for time-aware MIMO channel prediction,

    Y . Jin, Y . Wu, Y . Gao, et al., “Linformer: A linear-based lightweight transformer architecture for time-aware MIMO channel prediction,” IEEE Transactions on Wireless Communications, 2025

    2025

  8. [14]

    From data-driven learning to physics- inspired inferring: A novel mobile MIMO channel prediction scheme based on neural ODE,

    Z. Xiao, Z. Zhang, Z. Chen, et al., “From data-driven learning to physics- inspired inferring: A novel mobile MIMO channel prediction scheme based on neural ODE,”IEEE Transactions on Wireless Communications, vol. 23, no. 7, pp. 7186–7199, 2023

    2023

  9. [17]

    Deep transfer learning-based downlink channel prediction for FDD massive MIMO systems,

    Y . Yang, F. Gao, Z. Zhong, et al., “Deep transfer learning-based downlink channel prediction for FDD massive MIMO systems,”IEEE Transactions on Communications, vol. 68, no. 12, pp. 7485–7497, 2020

    2020

  10. [18]

    FIRE: Enabling reciprocity for FDD MIMO systems,

    Z. Liu, G. Singh, C. Xu, et al., “FIRE: Enabling reciprocity for FDD MIMO systems,” inProceedings of the 27th Annual International Conference on Mobile Computing and Networking, 2021, pp. 628–641

    2021

  11. [19]

    LLM4CP: Adapting large language models for channel prediction,

    B. Liu, X. Liu, S. Gao, et al., “LLM4CP: Adapting large language models for channel prediction,”Journal of Communications and Information Networks, vol. 9, no. 2, pp. 113–125, 2024

    2024

  12. [20]

    BERT4MIMO: A foundation model using BERT architecture for massive MIMO channel state information prediction,

    F. O. Catak, M. Kuzlu, and U. Cali, “BERT4MIMO: A foundation model using BERT architecture for massive MIMO channel state information prediction,”arXiv preprint arXiv:2501.01802, 2025

    2025 arXiv

  13. [21]

    LWM-Temporal: Sparse spatio-temporal attention for wireless channel representation learning,

    S. Alikhani, A. Malhotra, S. Hamidi-Rad, et al., “LWM-Temporal: Sparse spatio-temporal attention for wireless channel representation learning,”arXiv preprint arXiv:2603.10024, 2026

    2026

  14. [22]

    AirFM-DDA: Air-interface foundation model in the delay-doppler-angle domain for AI-native 6G,

    K. Bian, M. Tao, J. Mo, et al., “AirFM-DDA: Air-interface foundation model in the delay-doppler-angle domain for AI-native 6G,”arXiv preprint arXiv:2605.00020, 2026

    2026 arXiv

  15. [23]

    WiFo: Wireless foundation model for channel prediction,

    B. Liu, S. Gao, X. Liu, et al., “WiFo: Wireless foundation model for channel prediction,”Science China Information Sciences, vol. 68, no. 6, p. 162 302, 2025

    2025

  16. [24]

    CSI-JEPA: Towards foundation representa- tions for ubiquitous sensing with minimal supervision,

    X. Luo, Z. Li, and Y . Liu, “CSI-JEPA: Towards foundation representa- tions for ubiquitous sensing with minimal supervision,”arXiv preprint arXiv:2605.14171, 2026

    2026 arXiv

  17. [25]

    WirelessJEPA: A multi-antenna foundation model using spatio-temporal wireless latent predictions,

    V . Chu, O. Mashaal, and H. Abou-Zeid, “WirelessJEPA: A multi-antenna foundation model using spatio-temporal wireless latent predictions,” arXiv preprint arXiv:2601.20190, 2026

    2026

  18. [26]

    V-JEPA 2: Self-supervised video models enable understanding, prediction and planning,

    M. Assran, A. Bardes, D. Fan, et al., “V-JEPA 2: Self-supervised video models enable understanding, prediction and planning,”arXiv preprint arXiv:2506.09985, 2025

    2025 arXiv

  19. [27]

    Tutorial on joint em- bedding predictive architectures (JEPA): Foundations, applications, and future directions,

    M. Monemi, M. Chinipardaz, M. Rasti, et al., “Tutorial on joint em- bedding predictive architectures (JEPA): Foundations, applications, and future directions,”Authorea Preprints,

  20. [28]

    Learning latent wireless dynamics from channel state information,

    C. B. Chaaya, A. M. Girgis, and M. Bennis, “Learning latent wireless dynamics from channel state information,”IEEE Wireless Communica- tions Letters, vol. 14, no. 2, pp. 489–493, 2024

    2024

  21. [29]

    LatentWave: JEPA pretraining for wireless foundation models,

    A. Mohamed, A. Aboulfotouh, and H. Abou-Zeid, “LatentWave: JEPA pretraining for wireless foundation models,”arXiv preprint arXiv:2606.06373, 2026

    2026 arXiv

  22. [30]

    Mobiworld: World models for mobile wireless network,

    H. Chai, Y . Yuan, and Y . Li, “Mobiworld: World models for mobile wireless network,”arXiv preprint arXiv:2507.09462, 2025

    2025 arXiv

  23. [31]

    JEPA-MSAC: A joint-embedding predictive architecture for multimodal sensing-assisted communications,

    C. Zheng, J. He, G. Cai, et al., “JEPA-MSAC: A joint-embedding predictive architecture for multimodal sensing-assisted communications,” arXiv preprint arXiv:2603.29796, 2026

    2026

  24. [32]

    A wireless world model for AI-native 6G networks,

    Z. Chen, Y . Ren, Y . Huang, et al., “A wireless world model for AI-native 6G networks,”arXiv preprint arXiv:2603.25216, 2026

    2026

  25. [33]

    Time-series JEPA for pre- dictive remote control under capacity-limited networks,

    A. M. Girgis, A. Valcarce, and M. Bennis, “Time-series JEPA for pre- dictive remote control under capacity-limited networks,”IEEE Internet of Things Journal, 2026

    2026

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.