Pith. sign in

REVIEW 3 major objections 5 minor 62 references

Generalizable and Computational Efficient Channel Extrapolation for 6G: A Configurable AI-Driven Framework Built from a Modular Perspective

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Clustering neurons of a pre-trained channel encoder into experts beats joint mixture-of-experts training at 38% lower cost.

desk verdict A promising construction recipe for MoE channel extrapolation, but the paper overstates the complexity saving and never checks whether the clustering step itself is load-bearing. read the letter →

arxiv 2608.04630 v1 pith:P5MTZSOJ submitted 2026-08-05 eess.SP

Yuan Gao , Xinyi Wu , Jiang Jun , Yi Yu , Yanliang Jin , Shunqing Zhang , Zhu Han , Shugong Xu This is my paper · ORCID
classification eess.SP
keywords 6Gchannelextrapolationstateinformationmaskedauto-encodermixtureofexpertsmodularAIgeneralizationcomputationalefficiency
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 is trying to establish that channel extrapolation for 6G can be both generalizable and computationally cheap if the model is built from specialized modules instead of trained as one whole network. Its three-stage framework pre-trains a masked auto-encoder on CSI from mixed indoor and outdoor environments, partitions the encoder's feedforward neurons into equal-size experts by weight-space clustering, and then fine-tunes only a lightweight gating network that selects the relevant experts per scenario. The authors report that this reduces channel extrapolation error by 1.1-19.1 dB and inference cost by 38 percent relative to the dense MAE baseline, and that it beats an end-to-end-trained mixture-of-experts model on in-distribution, out-of-distribution, and unseen indoor-factory channels. The reason to care is that 6G systems need CSI with low pilot overhead across heterogeneous scenarios, and a modular model promises scenario-specific behavior without retraining the whole network. If the central modularity claim holds, the same expert set can be adapted to new environments by training only a small router.

What carries the argument

The central object is the expert: a cluster of hidden neurons in an encoder FFN layer that is supposed to act as a specialized functional module. Expert construction treats the $n$-th hidden neuron as the feature vector $W_1[:, n]$, the $n$-th column of the first FFN weight matrix, and runs balanced K-means with the exact-size constraint $|I_i| = d_{\mathrm{ff}}/k$, then reorders $W_2$ rows and $b_1$ entries with the same index sets so the partition is consistent across all parameters. The gating network built in the third stage takes the masked CSI input, produces softmax scores over experts, and activates only the top-$r$ experts, with all expert parameters frozen during per-scenario fine-tuning. This separation, experts fixed by weight-space clustering and router trained afterward, is what the paper claims converts the pre-trained MAE into an MoE without joint training.

What would settle it

On a held-out scenario, run the full pipeline twice, once with Stage-2 experts from balanced K-means and once with an equal-size random partition of the same pre-trained neurons but identical gating and fine-tuning; if the random partition matches or beats the clustering partition in extrapolation error, expert construction is not doing the work. A complementary check is to silence one expert cluster at a time and look for scenario-specific error spikes, which would show that the clusters track distinct propagation phenomena.

Watch

Extended reading notes

Core claim

At the center of the paper is a construction claim: after MAE pre-training on mixed indoor-LoS and outdoor-NLoS CSI, the feedforward layers of the Transformer encoder carry functionally specialized neurons, so grouping hidden neurons by the columns of the first FFN weight matrix $W_1$ yields reusable experts. The paper forms experts with balanced K-means, each containing exactly $d_{\mathrm{ff}}/k$ neurons, keeps them frozen, and fine-tunes a two-layer MLP gating network that scores experts and applies top-$r$ selection. The authors state that this three-stage procedure yields the same mixture-of-experts structure as their comparison baseline [33] but obtained from modular construction rather than joint training. In simulation, the resulting model outperforms the jointly trained MoE by 0.8-16.2 dB in indoor scenarios and 1.1-18.3 dB in outdoor scenarios, generalizes to indoor-NLoS and outdoor-LoS scenarios it never saw during expert construction, transfers to a QuaDRiGa-generated indoor-factory environment, and lowers FLOPs by 38 percent versus the dense MAE.

Load-bearing premise

The load-bearing premise is that pre-training on mixed indoor and outdoor channel data makes individual hidden neurons of the network specialize into reusable functional modules, so that grouping them by how similar their learned weights are yields genuine experts rather than arbitrary parameter splits; if that specialization is absent, the claimed accuracy and efficiency gains reduce to whatever the gating network and fine-tuning contribute on their own.

Editorial extensions

If this is right

  • A single pre-trained expert set can be reused across in-distribution, out-of-distribution, and unseen indoor-factory channels, so adapting a model to a new scenario reduces to training a small router instead of the whole network.
  • Because only the lightweight gating function is fine-tuned per scenario, the deployment cost for a new propagation environment drops far below end-to-end retraining.
  • Activating only two of eight experts per FFN layer preserves or improves accuracy while lowering FLOPs by 38 percent versus the dense MAE, so modular construction does not trade accuracy for efficiency.
  • The same three-stage pipeline applies to any Transformer-based wireless model, meaning expert emergence and selection could be reused for resource allocation, precoding, beam management, or other tasks once a pre-trained backbone exists.
  • Expert activation patterns differ coherently across scenarios, with LoS versus NLoS driving the largest differences, so the routing statistics themselves carry information about the propagation environment.

Reading between the lines

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

  • Editorial extension: the same balanced-K-means expert extraction could be tested on other wireless foundation models; whether clusters remain functionally coherent for tasks such as beam prediction or positioning is an empirical question the paper does not run.
  • Editorial extension: the scenario-dependent expert-activation vectors suggest a low-dimensional radio-environment fingerprint that could support LoS/NLoS detection or coverage classification, a use the authors do not claim.
  • Editorial extension: the reported 38 percent FLOP reduction is tied to the top-$r$ budget used in the experiments; pushing sparsity further could cut cost more, but the authors' own ablation shows diminishing returns beyond $K=16$, so the optimal sparsity frontier remains open.
  • Editorial extension: all experiments use simulated channels from DeepMIMO and QuaDRiGa; a decisive next step the paper does not take is measuring whether the emergent expert structure survives on real 60 GHz channel measurements.
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 three-stage framework for channel extrapolation in 6G: (1) pre-train a masked auto-encoder (MAE) on mixed indoor-LoS and outdoor-NLoS CSI to let 'experts' emerge in the FFN neurons; (2) construct experts by balanced K-means clustering of W1 columns (Eq. (13)); (3) add a lightweight MLP gating, fine-tune the non-expert parameters, and activate only the top-r experts. The authors compare against a jointly-trained MoE (their prior SSNet [33]) and the MAE backbone, reporting lower extrapolation error in indoor/outdoor, out-of-distribution, and indoor-factory scenarios, plus lower FLOPs than MAE. They also show scenario-dependent expert activation probabilities and an ablation that replaces the learned gating with random expert selection.

Significance. If the central modularity claim holds, the paper offers a useful and practical way to convert a monolithic pre-trained model into a sparse, scenario-adaptive mixture-of-experts, potentially reducing inference cost while improving generalization. The framework is clearly described, the use of standard simulators (DeepMIMO, QuaDRiGa) and an unseen indoor-factory test are strengths, and the complexity table is a helpful addition. However, the core premise that expert emergence and the specific K-means clustering are load-bearing is not directly tested; the only ablation targets the gating mechanism. Without a random-partition control, the central novelty remains unverified, and the headline accuracy/complexity numbers are internally inconsistent.

major comments (3)
  1. [Expert Construction and Results (Figs. 5, 6, 10)] The paper's central claim is that pre-training makes FFN neurons modular and that balanced K-means clustering (Eq. (13)) constructs meaningful experts. The only ablation, 'Proposed framework*', replaces the learned gating with random activation while keeping the K-means experts fixed; this tests the gating, not the clustering. A random equal-size partition of the same pre-trained neurons, combined with the same trained gating, could plausibly perform equally well, which would undercut the 'experts emergent' narrative. Please add a control experiment that uses a random neuron partition (e.g., shuffled neuron indices) with the otherwise identical pipeline, and report its performance in the same figures and tables. This is load-bearing because the expert-construction step is the main claimed novelty over a standard MoE.
  2. [Abstract and Results (Table I)] The abstract claims a 38% FLOPs reduction, but Table I lists FLOPs of 27.61 M for the proposed model and 38.22 M for MAE, which corresponds to a 27.8% reduction; the same 38% figure is repeated in the text near Table I. The abstract also reports gains of 1.1–19.1 dB that do not match the body: the contribution section reports 0.8–16.2 dB and 1.1–18.3 dB, while the discussion reports 1.3–16.2 dB, 1.7–18.3 dB, 0.8–6.4 dB, and 1.1–16.1 dB depending on the scenario. These headline numbers should be reconciled and corrected, as they are the first quantitative claims a reader evaluates.
  3. [Results (Figs. 5–10)] The experimental comparisons appear to be based on a single run per configuration; the figures show no error bars, confidence intervals, or repeated-seed results. Since several claimed gains are small at low SNR (e.g., Figs. 5 and 6 near 0–10 dB), it is not possible to assess whether the differences are statistically meaningful. Please run multiple seeds (at least 3–5) and report mean and variance, or otherwise provide evidence that the results are stable.
minor comments (5)
  1. [Abstract] The abstract contains two typos: 'section modules' should be 'selection modules' and 'db' should be 'dB' (the latter also appears elsewhere in the paper).
  2. [Results / Fig. 6] In the paragraph introducing Fig. 6, the text says 'Fig. 6 illustrate the performance ... in-distribution manner', but the figure and caption describe out-of-distribution performance; please correct this.
  3. [Table I and accompanying text] The parameter-increase statement says the proposed framework increases parameters by '7%' relative to MAE, but Table I shows 7.21 M vs 7.16 M, which is approximately a 0.7% increase; please clarify the intended comparison.
  4. [Table I caption] The caption contains a double 'the' ('between the the MoE') and the meaning of the arrows and parenthetical values is not fully defined; please expand the notation so the table is self-contained.
  5. [Expert Selection (Eq. (15))] It is not clear whether the gating network is shared across all FFN layers or is layer-specific; Eq. (15) suggests a single gating input X_CSI, but the framework is applied to each FFN layer. Please specify the gating architecture precisely.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central claims are empirical comparisons, no equation reduces to a fitted value, and no load-bearing argument rests on an unverified self-citation.

full rationale

The claimed derivation chain is not circular. The three stages (MAE pre-training, balanced K-means clustering of W1 columns, and gating fine-tuning) produce a model that is empirically compared, not derived, against baselines. No equation maps a fitted parameter to the reported gains: the 1.1-19.1 dB improvements come from simulations on held-out in-distribution, out-of-distribution, and indoor-factory data, while the 38% FLOPs reduction is an architectural consequence of activating r=2 of K=8 experts, not a fitted quantity renamed as a result. The statements that the authors 'obtain the same MoE model proposed in [33]' and adopt 'the same expert structure' are explicit admissions of architectural identity used to set up a controlled comparison; they do not define the performance gains into existence. Self-citations appear ([14] for the MAE backbone, [33] as the MoE baseline), but neither is load-bearing: the backbone is a standard architecture and the MoE baseline is an implemented comparison, not an unverified premise imported to force the conclusion. The 'experts emergent' premise is imported from external NLP modularity studies [47,50] and is tested empirically via activation patterns and transfer results; whether it holds for CSI is an evidence question, not circularity. The absence of a random-partition ablation for Stage 2 is a limitation of experimental support, but it does not make any prediction equivalent to its input. No uniqueness theorem is invoked, and no known result is renamed as new without attribution. Verdict: no significant circularity.

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

All neural network weights are learned from data and are not enumerated here. The ledger lists hand-chosen hyperparameters and the structural assumptions the framework depends on. No new physical entities are introduced: the 'experts' are clusters of existing FFN neurons, not independent postulated objects. The main non-standard element is the transfer of NLP emergent-modularity results to CSI MAE pre-training.

free parameters (3)
  • K (number of experts per FFN layer) = 8 (default; ablations K=4,8,16)
    Hand-chosen hyperparameter that controls modularity granularity; the paper selects K=8 based on an ablation, not by derivation.
  • r (number of top experts selected) = 2 (default; ablations (K,r)=(4,1),(8,2),(16,4))
    Hand-chosen routing budget with activation ratio held at 1/4; the performance claims depend on this choice.
  • Pre-training scenario mix = 50% DeepMIMO indoor LoS + 50% outdoor NLoS, 20,000 samples each
    The Stage 1 pre-training data composition is chosen by hand and directly determines which experts emerge.
assumptions (4)
  • domain assumption CSI correlations can be captured by a small number of specialized functional modules (experts) that are activated differently across scenarios.
    Stated in the abstract and throughout; if false, the modular framework has no basis.
  • ad hoc to paper After MAE pre-training on mixed CSI, FFN layers exhibit functional modularity in weight space, so balanced K-means on W1 columns yields meaningful experts.
    Imported from NLP results [47,50] in the 'Experts Emergent via Pre-training' section; no independent CSI-domain verification is provided beyond the paper's own end-to-end experiments.
  • ad hoc to paper Equal-size balanced clustering with the same hidden dimension per expert preserves the functional content of the original FFN.
    The 'Expert Construction' section imposes equal cluster sizes; there is no analysis of whether forced equal sizes degrade modularization.
  • domain assumption Freezing the clustered experts and fine-tuning only the gating network is sufficient to adapt to a new scenario.
    Stage 3 design; the random-expert ablation supports the role of gating, but not the sufficiency of this specific fine-tuning strategy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalizable and Computational Efficient Channel Extrapolation for 6G: A Configurable AI-Driven Framework Built from a Modular Perspective." pith.science (2026). https://pith.science/paper/P5MTZSOJ

@misc{pith2026260804630,
  author       = {Pith},
  title        = {Pith review of: Generalizable and Computational Efficient Channel Extrapolation for 6G: A Configurable AI-Driven Framework Built from a Modular Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P5MTZSOJ}},
  note         = {Machine review of arXiv:2608.04630}
}
abstract

Acquiring channel state information (CSI) with manageable overhead has been essential to provide high-performance communication services, which is extremely challenging in the emerging sixth generation (6G) mobile network. Channel extrapolation has been proposed to infer complete CSI using a small portion of known CSI, its performance can be dramatically enhanced by artificial intelligence (AI). However, AI-driven channel extrapolation suffers from poor generalization across scenarios and high computational complexity, which is common in the broad research of AI and large language models. Inspired by the modular function of human brain, we propose a configurable AI-driven framework to achieve generalizable and computational efficient channel extrapolation from a modular perspective. We propose a three-stage framework, consisting of experts emergent, experts construction and experts selection. This framework assumes that CSI correlations can be captured by a small number of specialized functional modules (experts) that are activated differently across scenarios. Such modularity emerges in the experts emergent stage via pre-training using CSI data covering comprehensive scenarios. The neurons with similar weight-space patterns are grouped as experts in the experts construction stage. A lightweight gating function is added to control the routing of experts and is fine-tuned for each scenario in the experts selection stage. Simulation results demonstrate that the proposed three-stage framework reduce the channel extrapolation error and computational complexities dramatically by $1.1-19.1$ db and $38$ \%, respectively. In addition, attributed to the proposed experts emergent and section modules, the proposed framework outperforms its counterpart mix-of-expert model dramatically in terms of channel extrapolation performance.

Figures

Figures reproduced from arXiv: 2608.04630 by the authors.

Figure 1
Figure 1. Proposed three-stage framework. Architecture of the proposed configurable channel extrapolation framework, which [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Stage 1: experts emergent. Experts emergent by pre-training a masked auto-encoder (MAE)-based model via masked [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Stage 2: experts construction. Experts constructed [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Stage 3: experts selection. After completing the expert construction of the FFN layers in the pre-trained MAE encoder, [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: In-distribution performance. The in-distribution performance comparison between the mixed of expert MoE [ [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Out-of-distribution performance. The out-of-distribution performance comparison between the zeroshot and fine [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Experts activation probabilities in each scenario. The illustration of the activation probability of experts in the proposed [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Experts activation patterns between scenarios. The [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Performance with respect to the number of experts. Channel extrapolation performance of the proposed framework [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Performance in indoor factory (InF) scenarios. The performance comparison between the mixed of expert (MoE) [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 52 canonical work pages

  1. [33]

    SSNet: Flexible and robust channel extrapolation for fluid antenna systems enabled by an self-supervised learning framework,

    Y . Gao, Y . Liu, R. Yu, S. Liu, Y . Jin, S. Zhang, S. Xu, and X. Chu, “SSNet: Flexible and robust channel extrapolation for fluid antenna systems enabled by an self-supervised learning framework,”IEEE Journal on Selected Areas in Communica- tions, vol. 44, pp. 1276–1289, Jan. 2026

    work page 2026

  2. [1]

    Sidelink positioning: Standardization advancements, challenges and opportunities,

    Y . Gao, G. Pan, Z. Zhong, Z. Jinm, Y . Hu, Y . Jin, , and S. Xu, “Sidelink positioning: Standardization advancements, challenges and opportunities,”IEEE Communications Maga- zine, vol. 64, no. 4, pp. 128–134, Jan. 2026

    work page 2026

  3. [2]

    Enhanced fingerprint-based positioning with practical imperfections: Deep learning-based approaches,

    S. Xu, J. Jiang, W. Yu, Y . Gao, G. Pan, S. Mu, Z. Ai, Y . Gao, P. Jiang, and C.-X. Wang, “Enhanced fingerprint-based positioning with practical imperfections: Deep learning-based approaches,”IEEE Wirel. Commun., vol. 33, no. 1, pp. 252– 258, Feb. 2026

    work page 2026

  4. [3]

    A stochastic geometry-based analytical framework for integrated localization and communication sys- tems,

    Y . Gao, H. Du, Z. Jiang, H. Hu, J. Zhang, S. Zhang, J. Du, F. R. Yu, and S. Xu, “A stochastic geometry-based analytical framework for integrated localization and communication sys- tems,”IEEE Internet of Things Journal, vol. 12, no. 18, pp. 38 710–38 722, Sep. 2025

    work page 2025

  5. [4]

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

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

    work page 2023

  6. [5]

    Hyper-parameter optimization for wireless network traffic prediction models with a novel meta- learning framework,

    L. Wang, J. Zhang, Y . Gao, J. Zhang, G. Wei, H. Zhou, B. Zhuge, and Z. Zhang, “Hyper-parameter optimization for wireless network traffic prediction models with a novel meta- learning framework,”IEEE Internet of Things Journal, vol. 12, no. 15, pp. 29 514–29 528, May 2025

    work page 2025

  7. [6]

    On the performance of coexisting NR-U and WiGig networks with directional sensing,

    H. Hu, C. Wang, Y . Gao, Y . Dong, Q. Chen, and J. Zhang, “On the performance of coexisting NR-U and WiGig networks with directional sensing,”IEEE Transactions on Communications, vol. 73, no. 1, pp. 469–482, Jul. 2025

    work page 2025

  8. [7]

    On the performance of an integrated communication and localization system: An analytical framework,

    Y . Gao, H. Hu, J. Zhang, Y . Jin, S. Xu, and X. Chu, “On the performance of an integrated communication and localization system: An analytical framework,”IEEE Transactions on V ehic- ular Technology, vol. 73, no. 7, pp. 10 845–10 849, Feb. 2024

    work page 2024

Show all 62 references
  1. [8]

    Fair and efficiency coexistence between nr-u and wigig networks enabled by a matching-based framework with forbidden pairs,

    Y . Gao, Z. Zhang, H. Hu, X. Wang, Y . Jin, and X. Chu, “Fair and efficiency coexistence between nr-u and wigig networks enabled by a matching-based framework with forbidden pairs,” IEEE Transactions on V ehicular Technology, vol. 72, no. 9, pp. 11 814–11 827, Apr. 2023

    2023

  2. [9]

    On the fairness of the coexisting LTE-U and WiFi networks sharing multiple unlicensed channels,

    H. Hu, Y . Gao, J. Zhang, X. Chu, Q. Chen, and J. Zhang, “On the fairness of the coexisting LTE-U and WiFi networks sharing multiple unlicensed channels,”IEEE Transactions on V ehicular Technology, vol. 69, no. 11, pp. 13 890–13 904, Oct. 2020

    2020

  3. [10]

    Licensed and unlicensed bands allocation for cellular users: A matching-based approach,

    Y . Gao, Y . Wu, H. Hu, X. Chu, and J. Zhang, “Licensed and unlicensed bands allocation for cellular users: A matching-based approach,”IEEE Wireless Communications Letters, vol. 8, no. 3, pp. 969–972, Mar. 2019

    2019

  4. [11]

    Coexistence of LTE- LAA and Wi-Fi on 5 GHz with corresponding deployment sce- narios: A survey,

    B. Chen, J. Chen, Y . Gao, and J. Zhang, “Coexistence of LTE- LAA and Wi-Fi on 5 GHz with corresponding deployment sce- narios: A survey,”IEEE Communications Surveys & Tutorials, vol. 19, no. 1, pp. 7–32, Jul. 2016

    2016

  5. [12]

    Generalizable and Robust Beam Prediction for 6G Networks: An Deep-Learning Framework with Positioning Feature Fu- sion,

    Y . Jin, Y . Li, J. Jun, Y . Gao, S. Liu, J. Du, Z. Yang, and S. Xu, “Generalizable and Robust Beam Prediction for 6G Networks: An Deep-Learning Framework with Positioning Feature Fu- sion,”IEEE Transactions on Network Science and Engineering, vol. 13, pp. 9680–9698, 2026

    2026

  6. [13]

    A matching-based pilot assignment algorithm for cell-free massive MIMO networks,

    Y . Gao, H. Hu, J. Chen, X. Wang, X. Chu, and J. Zhang, “A matching-based pilot assignment algorithm for cell-free massive MIMO networks,”IEEE Transactions on V ehicular Technology, vol. 73, no. 1, pp. 1453–1457, Aug. 2023

    2023

  7. [14]

    Channel extrapolation for mimo systems with the assistance of multi-path information induced from channel state information,

    Y . Gao, X. Wu, J. Jun, Z. Zhang, Z. Yang, S. Xu, C.-X. Wang, and Z. Han, “Channel extrapolation for mimo systems with the assistance of multi-path information induced from channel state information,”arXiv preprint arXiv:2601.21524, Jan. 2026

    2026

  8. [15]

    AI-based time-, frequency-, and space-domain channel extrapolation for 6G: Opportunities and challenges,

    Z. Zhang, J. Zhang, Y . Zhang, L. Yu, and G. Liu, “AI-based time-, frequency-, and space-domain channel extrapolation for 6G: Opportunities and challenges,”IEEE V ehicular Technology Magazine, vol. 18, no. 1, pp. 29–39, Jan. 2023

    2023

  9. [16]

    Effective outdoor pathloss prediction: A multi-layer segmentation approach with weighting map,

    Y . Gao, T. Wen, W. Xie, J. Du, Y . Zeng, D. Niyato, and S. Xu, “Effective outdoor pathloss prediction: A multi-layer segmentation approach with weighting map,”arXiv preprint arXiv:2601.08436, Jan. 2026

    2026

  10. [17]

    Enabling 6g through multi-domain channel ex- trapolation: Opportunities and challenges of generative artificial intelligence,

    Y . Gao, Z. Lu, Y . Wu, Y . Jin, S. Zhang, X. Chu, S. Xu, and C.-X. Wang, “Enabling 6g through multi-domain channel ex- trapolation: Opportunities and challenges of generative artificial intelligence,”IEEE Communications Magazine, vol. 64, no. 1, pp. 222–228, Oct. 2026

    2026

  11. [18]

    Massive MIMO channel predic- tion in real propagation environments using tensor decomposi- tion and autoregressive models,

    W. Liu, Z. Chen, and X. Gao, “Massive MIMO channel predic- tion in real propagation environments using tensor decomposi- tion and autoregressive models,” in33rd Annual International Symposium on Personal, Indoor and Mobile Radio Communi- cations (PIMRC), Kyoto, Japan, Sep. 202...

    2022

  12. [19]

    Two-stage based partial online optimization channel prediction for massive MIMO in high-mobility,

    P. Wang, J. Li, A. Zhang, B. Ning, B. Sun, and P. Sun, “Two-stage based partial online optimization channel prediction for massive MIMO in high-mobility,”IEEE Communications Letters, vol. 28, no. 2, pp. 417–421, Jan. 2024

    2024

  13. [20]

    Basis expansion extrapolation based DL channel prediction with ul channel estimates for TDD MIMO-OTFS systems,

    Y . Zhang, X. Zhu, Y . Liu, Y . Jiang, R. Yin, Y . L. Guan, and D. Gonz´alez, “Basis expansion extrapolation based DL channel prediction with ul channel estimates for TDD MIMO-OTFS systems,” inIEEE International Conference on Communications (ICC), Rome, Italy, May 2023, pp. 2270–2275

    2023

  14. [21]

    A novel mobility induced channel prediction mechanism for vehicular communications,

    F. Peng, S. Zhang, Z. Jiang, X. Wang, and W. Chen, “A novel mobility induced channel prediction mechanism for vehicular communications,”IEEE Transactions on Wireless Communica- tions, vol. 22, no. 5, pp. 3488–3502, Nov. 2022

    2022

  15. [22]

    Sparse Bayesian learning based channel extrapolation for RIS assisted MIMO- OFDM,

    X. Xu, S. Zhang, F. Gao, and J. Wang, “Sparse Bayesian learning based channel extrapolation for RIS assisted MIMO- OFDM,”IEEE Transactions on Communications, vol. 70, no. 8, pp. 5498–5513, Jun. 2022

    2022

  16. [23]

    Nonparametric regres- sion for MU-MIMO channel prediction: From KNN to local linear regression,

    Z. Xiao, J. Sun, Z. Zhang, and Y . Liu, “Nonparametric regres- sion for MU-MIMO channel prediction: From KNN to local linear regression,”IEEE Transactions on Wireless Communica- tions, vol. 23, no. 4, pp. 2784–2795, Aug. 2023

    2023

  17. [24]

    AI-driven channel state information (CSI) extrapolation for 6G: Current situations, challenges and future research,

    Y . Gao, Z. Lu, X. Wu, W. Yu, S. Liu, J. Du, Y . Jin, S. Zhang, X. Chu, and S. Xu, “AI-driven channel state information (CSI) extrapolation for 6G: Current situations, challenges and future research,”IEEE Communications Surveys & Tutorials, vol. 28, pp. 4485–4518, Jan. 2026

    2026

  18. [25]

    Massive MIMO channel prediction: Kalman filtering vs. machine learn- 12 ing,

    H. Kim, S. Kim, H. Lee, C. Jang, Y . Choi, and J. Choi, “Massive MIMO channel prediction: Kalman filtering vs. machine learn- 12 ing,”IEEE Transactions on Communications, vol. 69, no. 1, pp. 518–528, Sep. 2020

    2020

  19. [26]

    Channel mapping based on interleaved learning with complex-domain MLP- mixer,

    Z. Chen, Z. Zhang, Z. Yang, and L. Liu, “Channel mapping based on interleaved learning with complex-domain MLP- mixer,”IEEE Wireless Communications Letters, vol. 13, no. 5, pp. 1369–1373, Feb. 2024

    2024

  20. [27]

    A comparison of neural networks for wireless channel prediction,

    O. Stenhammar, G. Fodor, and C. Fischione, “A comparison of neural networks for wireless channel prediction,”IEEE Wireless Communications, vol. 31, no. 3, pp. 235–241, Feb. 2024

    2024

  21. [28]

    Recurrent neural network- based frequency-domain channel prediction for wideband com- munications,

    W. Jiang and H. D. Schotten, “Recurrent neural network- based frequency-domain channel prediction for wideband com- munications,” inIEEE 89th vehicular technology conference (VTC2019-Spring), Kuala Lumpur , Malaysia, Apr. 2019, pp. 1– 6

    2019

  22. [29]

    LSTM-based channel prediction for secure massive MIMO communications under imperfect CSI,

    T. Peng, R. Zhang, X. Cheng, and L. Yang, “LSTM-based channel prediction for secure massive MIMO communications under imperfect CSI,” inIEEE International Conference on Communications (ICC), Virtual, Jun. 2020, pp. 1–6

    2020

  23. [30]

    Predicting future CSI feedback for highly-mobile massive MIMO systems,

    Y . Zhang, A. Alkhateeb, P. Madadi, J. Jeon, J. Cho, and C. Zhang, “Predicting future CSI feedback for highly-mobile massive MIMO systems,”arXiv preprint arXiv:2202.02492, May 2022

    2022 arXiv

  24. [31]

    E2Enet: An end-to-end channel prediction neural network based on uplink pilot for fdd systems,

    S. Fan, H. Li, X. Liang, Z. Liu, X. Gu, and L. Zhang, “E2Enet: An end-to-end channel prediction neural network based on uplink pilot for fdd systems,”IEEE Wireless Communications Letters, vol. 13, no. 5, pp. 1285–1289, Feb. 2024

    2024

  25. [32]

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

    Y . Jin, Y . Wu, Y . Gao, S. Zhang, S. Xu, and C.-X. Wang, “Linformer: A linear-based lightweight transformer architecture for time-aware MIMO channel prediction,”IEEE Transactions on Wireless Communications, vol. 24, no. 9, pp. 7177–7190, Apr. 2025

    2025

  26. [34]

    CSI-LLM: a novel downlink channel prediction method aligned with llm pre-training,

    S. Fan, Z. Liu, X. Gu, and H. Li, “CSI-LLM: a novel downlink channel prediction method aligned with llm pre-training,” in IEEE Wireless Communications and Networking Conference (WCNC), Milan, Italy, Mar. 2025, pp. 1–6

    2025

  27. [35]

    A generative pre-trained language model for channel prediction in wireless communications systems,

    B. Lin, H. Zhang, Y . Jiang, Y . Wang, T. Zhang, S. Yan, H. Li, Y . Liu, and F. Gao, “A generative pre-trained language model for channel prediction in wireless communications systems,” inConference on Empirical Methods in Natural Language Processing, Budapest, Hungary, Nov. 2...

    2025

  28. [36]

    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, Jan. 2025

    2025 arXiv

  29. [37]

    A wireless foundation model for multi-task prediction,

    Y . Sheng, J. Wang, X. Zhou, L. Liang, H. Ye, S. Jin, and G. Y . Li, “A wireless foundation model for multi-task prediction,” arXiv preprint arXiv:2507.05938, Aug. 2025

    2025 arXiv

  30. [38]

    A MIMO wireless channel foundation model via CIR-CSI consistency,

    J. Jiang, W. Yu, Y . Li, Y . Gao, and S. Xu, “A MIMO wireless channel foundation model via CIR-CSI consistency,”arXiv preprint arXiv:2502.11965, May 2025

    2025 arXiv

  31. [39]

    “3rd generation partnership project; technical specification group radio access network; study on artificial intelligence (AI)/machine learning (ML) for nr air interface (Release 18),” 3GPP, document TR 38.843, Mar. 2023

    2023

  32. [40]

    Building machines that learn and think like people,

    B. M. Lake, T. D. Ullman, J. B. Tenenbaum, and S. J. Gershman, “Building machines that learn and think like people,”Behavioral and brain sciences, vol. 40, p. 1851–1863, Oct. 2017

    2017

  33. [41]

    A survey on large language models with some insights on their capabilities and limitations,

    A. Matarazzo and R. Torlone, “A survey on large language models with some insights on their capabilities and limitations,” arXiv preprint arXiv:2501.04040, Nov. 2025

    2025 arXiv

  34. [42]

    The computational power of the human brain,

    P. J. Gebicke-Haerter, “The computational power of the human brain,”Frontiers in Cellular Neuroscience, vol. 17, pp. 1–26, Aug. 2023

    2023

  35. [43]

    Matching networks for one shot learning,

    O. Vinyals, C. Blundell, T. Lillicrap, and D. Wierstra, “Matching networks for one shot learning,”Advances in neural information processing systems, vol. 29, pp. 3637–3645, Dec. 2016

    2016

  36. [44]

    An energy budget for signaling in the grey matter of the brain,

    D. Attwell and S. B. Laughlin, “An energy budget for signaling in the grey matter of the brain,”Journal of Cerebral Blood Flow & Metabolism, vol. 21, no. 10, pp. 1133–1145, Oct. 2001

    2001

  37. [45]

    Modular deep learning,

    J. Pfeiffer, S. Ruder, I. Vuli ´c, and E. M. Ponti, “Modular deep learning,”arXiv preprint arXiv:2302.11529, May 2023

    2023 arXiv

  38. [46]

    Parameter-efficient fine- tuning of large-scale pre-trained language models,

    N. Ding, Y . Qin, G. Yang, F. Wei, Z. Yang, Y . Su, S. Hu, Y . Chen, C.-M. Chan, and W. Chen, “Parameter-efficient fine- tuning of large-scale pre-trained language models,”Nature machine intelligence, vol. 5, no. 3, pp. 220–235, Mar. 2023

    2023

  39. [47]

    Emergent modularity in pre- trained Transformers,

    Z. Zhang, Z. Zeng, Y . Lin, C. Xiao, X. Wang, X. Han, Z. Liu, R. Xie, M. Sun, and J. Zhou, “Emergent modularity in pre- trained Transformers,”arXiv preprint arXiv:2305.18390, May 2023

    2023 arXiv

  40. [48]

    L. H. Marshall and H. W. Magoun,Discoveries in the human brain: neuroscience prehistory, brain structure, and function. Springer Science & Business Media, Jan. 2013

    2013

  41. [49]

    Toward next-generation artificial intelli- gence: Catalyzing the NeuroAI revolution,

    A. Zador, S. Escola, B. Richards, Y . Bengio, K. Boahen, M. Botvinick, D. Chklovskii, A. Churchland, C. Clopath, and J. DiCarlo, “Toward next-generation artificial intelli- gence: Catalyzing the NeuroAI revolution,”arXiv preprint arXiv:2210.08340, Oct. 2022

    arXiv

  42. [50]

    Configurable foundation models: Building LLMs from a modular perspective,

    C. Xiao, Z. Zhang, C. Song, D. Jiang, F. Yao, X. Han, X. Wang, S. Wang, Y . Huang, and G. Lin, “Configurable foundation models: Building LLMs from a modular perspective,”arXiv preprint arXiv:2409.02877, Sep. 2024

    2024 arXiv

  43. [51]

    Masked autoencoders are scalable vision learners,

    K. He, X. Chen, S. Xie, Y . Li, P. Doll ´ar, and R. Gir- shick, “Masked autoencoders are scalable vision learners,” in IEEE/CVF conference on computer vision and pattern recogni- tion, LA, U.S., Jan. 2022, pp. 16 000–16 009

    2022

  44. [52]

    28 GHz in- door and outdoor propagation measurements and analysis at a regional airport,

    K. Du, O. Ozdemir, F. Erden, and I. Guvenc, “28 GHz in- door and outdoor propagation measurements and analysis at a regional airport,” inIEEE 32nd Annual International Sympo- sium on Personal, Indoor and Mobile Radio Communications (PIMRC), Virtual, Sep. 2021, pp. 866–872

    2021

  45. [53]

    Bringing it indoors: A review of narrowband radio propaga- tion modeling for enclosed spaces,

    M. E. Diago-Mosquera, A. Arag ´on-Zavala, and G. Casta ˜n´on, “Bringing it indoors: A review of narrowband radio propaga- tion modeling for enclosed spaces,”IEEE Access, vol. 8, pp. 103 875–103 899, Jan. 2020

    2020

  46. [54]

    Integrated communication, localization, and sensing in 6G D- MIMO networks,

    H. Guo, H. Wymeersch, B. Makki, H. Chen, Y . Wu, G. Durisi, M. F. Keskin, M. H. Moghaddam, C. Madapatha, and H. Yu, “Integrated communication, localization, and sensing in 6G D- MIMO networks,”IEEE Wireless Communications, vol. 32, no. 2, pp. 214–221, Mar. 2025

    2025

  47. [55]

    Statistical channel model of wide- band sub-THz radio propagation in indoor factories at 142 GHz: Towards 6G industrial wireless networks,

    S. Ju and T. S. Rappaport, “Statistical channel model of wide- band sub-THz radio propagation in indoor factories at 142 GHz: Towards 6G industrial wireless networks,”IEEE Transactions on Wireless Communications, vol. 23, no. 11, pp. 16 316– 16 331, Aug. 2024

    2024

  48. [56]

    Transformer masked autoencoders for next-generation wireless communications: Architecture and opportunities,

    A. Zayat, M. A. Hasabelnaby, M. Obeed, and A. Chaaban, “Transformer masked autoencoders for next-generation wireless communications: Architecture and opportunities,”IEEE Com- munications Magazine, vol. 62, no. 7, pp. 88–94, 2023

    2023

  49. [57]

    Wifo: Wireless foundation model for channel prediction,

    B. Liu, S. Gao, X. Liu, X. Cheng, and L. Yang, “Wifo: Wireless foundation model for channel prediction,”Science China Information Sciences, vol. 68, no. 6, p. 162302, 2025

    2025

  50. [58]

    Moefica- tion: Transformer feed-forward layers are mixtures of experts,

    Z. Zhang, Y . Lin, Z. Liu, P. Li, M. Sun, and J. Zhou, “Moefica- tion: Transformer feed-forward layers are mixtures of experts,” inFindings of the Association for Computational Linguistics (ACL), Dublin, Ireland, May 2022, pp. 877–890

    2022

  51. [59]

    Balanced k-means for clustering,

    M. I. Malinen and P. Fr¨anti, “Balanced k-means for clustering,” Proceedings of SSSPR, vol. 8621, pp. 32–41, Aug. 2014

    2014

  52. [60]

    DeepMIMO: A generic deep learning dataset for millimeter wave and massive MIMO applications,

    A. Alkhateeb, “DeepMIMO: A generic deep learning dataset for millimeter wave and massive MIMO applications,”arXiv preprint arXiv:1902.06435, Feb. 2019

    1902 arXiv

  53. [61]

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

    S. Jaeckel, L. Raschkowski, K. B ¨orner, and L. Thiele, “Quadriga: A 3-d multi-cell channel model with time evolution for enabling virtual field trials,”IEEE Transactions on Antennas 13 and Propagation, vol. 62, no. 6, pp. 3242–3256, 2014

    2014

  54. [62]

    Quadriga - quasi deterministic radio channel generator, user manual and documentation,

    S. Jaeckel, L. Raschkowksi, K. B ¨orner, L. Thiele, F. Burkhardt, and E. Eberlein, “Quadriga - quasi deterministic radio channel generator, user manual and documentation,” Fraunhofer Hein- rich Hertz Institute, Tech. Rep. v2.8.1, 2023. ACKNOWLEDGMENT This work was supported by...

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.