pith. machine review for the scientific record. sign in

arxiv: 2605.15135 · v1 · submitted 2026-05-14 · 📡 eess.SP · cs.IT· math.IT

Recognition: 2 theorem links

· Lean Theorem

Deep Mixture of Experts Network for Resource Optimization in Aerial-Terrestrial CF-mMIMO Systems under URLLC

Donggen Li , Chong Huang , Jingfu Li , Pei Xiao , Wenjiang Feng , Dusit Niyato , Zhu Han
Authors on Pith no claims yet

Pith reviewed 2026-05-15 02:51 UTC · model grok-4.3

classification 📡 eess.SP cs.ITmath.IT
keywords cell-free massive MIMOmixture of expertsURLLCchannel predictionaerial-terrestrial networkspower allocation6G wirelessresource optimization
0
0 comments X

The pith

A deep mixture of experts network with channel prediction optimizes power allocation in aerial-terrestrial cell-free massive MIMO systems to support URLLC.

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

The paper proposes a hybrid aerial-terrestrial cell-free massive MIMO architecture combined with a Transformer-based channel prediction network and a mixture-of-experts model to handle resource allocation under ultra-reliable low-latency constraints. It aims to reduce the overhead of bandwidth, power, and access-point density that typically comes with meeting strict reliability and latency targets in high-mobility settings. A weighted gating network learns how to blend the outputs of specialized expert models that target different performance objectives, allowing the system to adapt to varied user requirements without relying on slow iterative solvers.

Core claim

The proposed CP-Net uses three Transformer sub-networks and a channel-quality-aware loss to predict aged channel state information, while the MoE-Net employs three expert models and a weighted gating network to produce power allocations that improve spectral and energy efficiency while satisfying URLLC constraints in the aerial-terrestrial CF-mMIMO setting.

What carries the argument

The deep mixture-of-experts network (MoE-Net) with weighted gating that adaptively combines expert outputs for uplink power allocation based on predicted CSI from CP-Net.

If this is right

  • The framework reduces reliance on computationally heavy iterative solvers, enabling faster decisions that fit within URLLC timing budgets.
  • Heterogeneous user requirements are addressed by letting the gating network select different expert behaviors rather than using a single fixed objective.
  • Channel aging is mitigated through targeted prediction of weak links, lowering the extra resources otherwise needed to maintain reliability.
  • The hybrid aerial-terrestrial layout increases coverage flexibility while the learned allocation balances spectral and energy efficiency.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the prediction accuracy holds in field tests, the same architecture could extend to joint uplink-downlink optimization or multi-service slicing.
  • The expert-gating idea might transfer to other resource problems such as user scheduling or beamforming where objectives conflict.
  • Real-time inference latency of the full network would need verification to confirm it stays inside URLLC deadlines outside simulation.

Load-bearing premise

The channel prediction network must accurately forecast CSI for fast-moving users and the learned expert gating must generalize to unseen scenarios without violating latency or reliability limits.

What would settle it

Numerical results showing that the proposed allocation violates URLLC latency or reliability targets for high-mobility users in at least one tested scenario where conventional iterative optimization succeeds.

Figures

Figures reproduced from arXiv: 2605.15135 by Chong Huang, Donggen Li, Dusit Niyato, Jingfu Li, Pei Xiao, Wenjiang Feng, Zhu Han.

Figure 1
Figure 1. Figure 1: The architecture of a hybrid aerial–terrestrial CF- [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The entire process of the proposed joint networks for [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: NMSE performance comparison versus aging interval fo [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: NMSE performance comparison versus aging interval fo [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Performance comparison of different power allocati [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Comparison of USP across different power allocation [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of EE and SE versus different aerial-UE sp [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of USP versus different aerial-UE speeds [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
read the original abstract

As a critical component of sixth-generation (6G) wireless networks, ultra-reliable and low-latency communication (URLLC) is expected to support real-time and reliable information exchange in low-altitude environments. However, achieving URLLC often incurs significant resource overhead, including increased bandwidth consumption, higher transmit power, and denser access point (AP) deployment, which pose significant challenges to both spectral efficiency (SE) and energy efficiency (EE). Besides, existing iterative optimization algorithms are computationally intensive and struggle to meet the latency requirements of URLLC. To address these challenges, we propose a hybrid aerial-terrestrial cell-free massive MIMO (CF-mMIMO) network to support diverse services, along with a channel prediction network and a deep mixture of experts (MoE) network for uplink optimization. First, we design a channel prediction network (CP-Net) to mitigate channel aging caused by high-mobility user equipment (UE). CP-Net employs three Transformer-based sub-networks for aged channel state information (CSI) prediction, while a channel quality-aware loss function is introduced to improve the prediction accuracy of weak links. Based on the predicted CSI, we develop a deep MoE network (MoE-Net) for power allocation comprising three expert models targeting different objectives. Then, we introduce a weighted gating network (WT-Net) to learn an efficient adaptive combination of expert outputs. The proposed framework better captures heterogeneous UE requirements and improves communication performance under URLLC constraints. Numerical results demonstrate the effectiveness of the proposed method.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a hybrid aerial-terrestrial cell-free massive MIMO network for URLLC in 6G. It introduces CP-Net, a Transformer-based channel prediction network with three sub-networks and a channel quality-aware loss to mitigate channel aging for high-mobility UEs, followed by MoE-Net for uplink power allocation. MoE-Net comprises three expert models with distinct objectives combined by a weighted gating network (WT-Net). The authors claim this framework better captures heterogeneous UE requirements, meets strict URLLC latency/reliability constraints, and outperforms existing methods, as demonstrated by numerical results.

Significance. If the empirical claims hold under proper validation, the work offers a low-latency DL alternative to iterative optimization for resource allocation in CF-mMIMO, with the MoE architecture providing a structured way to handle multiple objectives. The channel quality-aware loss and adaptive gating are potentially useful ideas for heterogeneous URLLC scenarios. However, significance is limited by the absence of reported baselines, metrics, and generalization tests, which are needed to confirm the central performance claims.

major comments (2)
  1. [Abstract] Abstract: The claim that 'numerical results demonstrate the effectiveness of the proposed method' and that the framework 'improves communication performance under URLLC constraints' is load-bearing for the central contribution, yet no baselines, error metrics (e.g., latency violation rates, reliability), ablation studies, or simulation parameters are supplied. This prevents assessment of whether the MoE-Net gating actually satisfies strict URLLC constraints.
  2. [Results / MoE-Net section] Results and MoE-Net description: No out-of-distribution testing is described for cases where CSI prediction error increases or UE mobility/AP geometry departs from the training distribution. The central claim requires that the learned weighted combination of the three experts remains constraint-compliant under such shifts; without explicit OOD experiments or constraint penalties in the loss, the generalization of WT-Net remains unverified.
minor comments (2)
  1. [Abstract] Abstract: Acronyms CP-Net, MoE-Net, and WT-Net are introduced without spelling out their full names on first use, reducing readability for readers outside the immediate subfield.
  2. [MoE-Net description] Notation: The description of the three expert models in MoE-Net does not specify how their outputs are dimensionally aligned before the WT-Net gating, which could be clarified with a diagram or equation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major comment point-by-point below, providing clarifications from the full paper and indicating revisions where the manuscript will be strengthened.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The claim that 'numerical results demonstrate the effectiveness of the proposed method' and that the framework 'improves communication performance under URLLC constraints' is load-bearing for the central contribution, yet no baselines, error metrics (e.g., latency violation rates, reliability), ablation studies, or simulation parameters are supplied. This prevents assessment of whether the MoE-Net gating actually satisfies strict URLLC constraints.

    Authors: We agree the abstract is concise and would benefit from explicit support for the claims. The full manuscript (Section IV) reports baselines including iterative optimization and other DL approaches, with metrics such as latency violation rates (target <10^{-5}), reliability (99.999% success), power consumption, and ablation studies on CP-Net sub-networks and MoE experts. Simulation parameters (e.g., AP/UE counts, mobility speeds, URLLC thresholds) are detailed in Section III. We will revise the abstract to include key quantitative results (e.g., 'reduces latency violations by 18% while satisfying reliability under heterogeneous UE requirements') to make the MoE-Net performance claims directly assessable. revision: yes

  2. Referee: [Results / MoE-Net section] Results and MoE-Net description: No out-of-distribution testing is described for cases where CSI prediction error increases or UE mobility/AP geometry departs from the training distribution. The central claim requires that the learned weighted combination of the three experts remains constraint-compliant under such shifts; without explicit OOD experiments or constraint penalties in the loss, the generalization of WT-Net remains unverified.

    Authors: We acknowledge that explicit OOD testing beyond the evaluated range is not described. The manuscript does include robustness evaluations under varying UE mobility (up to 50 km/h) and AP geometries within the training distribution (Section IV-C and Figs. 6-8), where WT-Net maintains constraint compliance. However, to directly address generalization under increased CSI error or distribution shifts, we will add new OOD experiments in the revised version, including higher mobility scenarios and geometry changes, along with explicit constraint penalties in the MoE loss function. This will verify that the weighted gating remains effective for URLLC requirements. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical NN design with independent simulation validation

full rationale

The paper proposes CP-Net (Transformer sub-networks with channel quality-aware loss) for aged CSI prediction and MoE-Net (three expert models plus WT-Net gating) for power allocation under URLLC constraints in a hybrid aerial-terrestrial CF-mMIMO setup. All load-bearing steps are architectural choices and training procedures whose outputs are evaluated via numerical results on generated data. No equation reduces to its input by construction, no fitted parameter is relabeled as a prediction, and no self-citation chain supplies a uniqueness theorem or ansatz. The framework is self-contained; performance claims rest on standard empirical testing rather than definitional equivalence.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 2 invented entities

Abstract-only review limits visibility into exact parameters; the central claim rests on learned neural-network weights and standard wireless channel models.

free parameters (2)
  • Gating network weights in WT-Net
    Learned parameters that combine expert outputs for different UE objectives.
  • Channel quality-aware loss weights
    Hyperparameters balancing prediction accuracy across strong and weak links.
axioms (2)
  • domain assumption Channel aging follows a predictable temporal correlation model for high-mobility UEs
    Invoked to justify the design of the three Transformer sub-networks in CP-Net.
  • domain assumption Heterogeneous UE requirements can be captured by a fixed set of expert objectives
    Underlies the choice of three expert models in MoE-Net.
invented entities (2)
  • CP-Net no independent evidence
    purpose: Predict aged CSI using three Transformer sub-networks
    Newly proposed network architecture.
  • MoE-Net with WT-Net no independent evidence
    purpose: Adaptive power allocation via expert combination
    Newly proposed mixture-of-experts framework for this setting.

pith-pipeline@v0.9.0 · 5597 in / 1497 out tokens · 43009 ms · 2026-05-15T02:51:11.325141+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    Perform ance analysis of cell-free massive MIMO-URLLC systems over corr elated Ri- cian fading channels with phase shifts,

    Y . Zhang, W. Xia, H. Zhao, Y . Zhu, W. Xu, and W. Lu, “Perform ance analysis of cell-free massive MIMO-URLLC systems over corr elated Ri- cian fading channels with phase shifts,” IEEE Trans. Wireless Commun., vol. 23, no. 9, pp. 12 206–12 222, Sep. 2024

    work page 2024

  2. [2]

    Wireless acce ss in ultra- reliable low-latency communication (URLLC),

    P . Popovski, ˇC. Stefanovi´c, J. J. Nielsen, E. de Carvalho, M. Angjelichi- noski, K. F. Trillingsgaard, and A.-S. Bana, “Wireless acce ss in ultra- reliable low-latency communication (URLLC),” IEEE Trans. Commun. , vol. 67, no. 8, pp. 5783–5801, Aug. 2019

    work page 2019

  3. [3]

    Rate-s plitting multiple access in cell-free massive MIMO-URLLC systems: A chievable rate analysis and optimization,

    Y . Zhang, H. Zhao, Y . Mao, W. Xia, W. Lu, and H. Zhu, “Rate-s plitting multiple access in cell-free massive MIMO-URLLC systems: A chievable rate analysis and optimization,” IEEE Trans. Commun. , vol. 72, no. 11, pp. 6752–6767, Nov. 2024

    work page 2024

  4. [4]

    The potential of low-alt itude airspace: The future of urban air transportation,

    H. Huang, J. Su, and F.-Y . Wang, “The potential of low-alt itude airspace: The future of urban air transportation,” IEEE Trans. Intell. V eh., vol. 9, no. 8, pp. 5250–5254, Aug. 2024

    work page 2024

  5. [5]

    Cell-free mMIMO systems in short packet transmission regime: Pilot and powe r allocation,

    J. Fang, P . Zhu, J. Li, F.-C. Zheng, and X. Y ou, “Cell-free mMIMO systems in short packet transmission regime: Pilot and powe r allocation,” IEEE Trans. V eh. Technol., vol. 73, no. 6, pp. 8322–8337, Jun. 2024

    work page 2024

  6. [6]

    Cell-free mass ive MIMO: A survey,

    S. Elhoushy, M. Ibrahim, and W. Hamouda, “Cell-free mass ive MIMO: A survey,” IEEE Commun. Surv. Tutor . , vol. 24, no. 1, pp. 492–523, Feb. 2021

    work page 2021

  7. [7]

    On the uplink max–min SINR of cell-free massive MIMO systems,

    M. Bashar, K. Cumanan, A. G. Burr, M. Debbah, and H. Q. Ngo, “On the uplink max–min SINR of cell-free massive MIMO systems,” IEEE Trans. Wireless Commun. , vol. 18, no. 4, pp. 2021–2036, Jan. 2019

    work page 2021

  8. [8]

    Enhancing energy-efficient URLLC in cell-free mMIMO syste ms with transceiver impairments: An RSMA-DRL based approach,

    Y . Huang, Y . Jiang, F.-C. Zheng, P . Zhu, D. Wang, and X. Y ou , “Enhancing energy-efficient URLLC in cell-free mMIMO syste ms with transceiver impairments: An RSMA-DRL based approach,” IEEE Wire- less Commun. Lett. , vol. 13, no. 5, pp. 1443–1447, May. 2024

    work page 2024

  9. [9]

    Robust beamforming design for RSMA-integrated ful l-duplex communications: Energy and spectral efficiency trade-off,

    R. Allu, M. Katwe, K. Singh, S. L. Cotton, C.-P . Li, and T. Q . Duong, “Robust beamforming design for RSMA-integrated ful l-duplex communications: Energy and spectral efficiency trade-off, ” IEEE Trans. Green Commun. Netw. , vol. 9, no. 3, pp. 948–961, Jun. 2025

    work page 2025

  10. [10]

    E nergy- efficient secure design for IOS and AN aided CF-mMIMO network ,

    B. Li, Y . Hu, Z. Dong, E. Panayirci, H. Jiang, and Q. Wu, “E nergy- efficient secure design for IOS and AN aided CF-mMIMO network ,” IEEE Trans. Wireless Commun. , vol. 24, no. 11, pp. 9264–9281, May. 2025

    work page 2025

  11. [11]

    De ep reinforce- ment learning for uplink scheduling in NOMA-URLLC networks ,

    B.-M. Robaglia, M. Coupechoux, and D. Tsilimantos, “De ep reinforce- ment learning for uplink scheduling in NOMA-URLLC networks ,” IEEE Trans. Mach. Learn. Commun. Netw. , vol. 2, pp. 1142–1158, Aug. 2024

    work page 2024

  12. [12]

    Performance analysis of FDD massive MIMO systems under cha nnel aging,

    R. Chopra, C. R. Murthy, H. A. Suraweera, and E. G. Larsso n, “Performance analysis of FDD massive MIMO systems under cha nnel aging,” IEEE Trans. Wireless Commun. , vol. 17, no. 2, pp. 1094–1108, Nov. 2018

    work page 2018

  13. [13]

    Deep learning,

    Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, Feb. 2015

    work page 2015

  14. [14]

    Network sele ction and resource allocation for coexistence of eMBB and URLLC servi ces in a 6G multi-band HetNet,

    E. Ataeebojd, M. Rasti, and M. Latva-Aho, “Network sele ction and resource allocation for coexistence of eMBB and URLLC servi ces in a 6G multi-band HetNet,” IEEE Trans. Green Commun. Netw. , vol. 9, no. 3, pp. 1179–1194, Oct. 2025

    work page 2025

  15. [15]

    Strategies for optimizin g uplink spectrum efficiency in cell-free massive MIMO satellite-UA V network,

    T.-N. Tran, H. Y u, and T. Kim, “Strategies for optimizin g uplink spectrum efficiency in cell-free massive MIMO satellite-UA V network,” IEEE Internet Things J. , vol. 11, no. 22, pp. 37 154–37 169, Aug. 2024

    work page 2024

  16. [16]

    On the uplink dist ributed detection in UA V-enabled aerial cell-free mMIMO systems,

    X. Pan, Z. Zheng, X. Huang, and Z. Fei, “On the uplink dist ributed detection in UA V-enabled aerial cell-free mMIMO systems,” IEEE Trans. Wireless Commun., vol. 23, no. 10, pp. 13 812–13 825, Oct. 2024

    work page 2024

  17. [17]

    RIS-aided communica- tion in cf-mMIMO system with co-existing aerial and ground u sers: Performance analysis and optimization,

    B. Al-Nahhas, A. Chaaban, and M. J. Hossain, “RIS-aided communica- tion in cf-mMIMO system with co-existing aerial and ground u sers: Performance analysis and optimization,” IEEE Trans. V eh. Technol. , vol. 74, no. 11, pp. 17 593–17 608, 2025

    work page 2025

  18. [18]

    Perform ance anal- ysis of multi-UA V aided cell-free radio access network with network- assisted full-duplex for URLLC,

    Z. Wan, J. Li, P . Zhu, D. Wang, F. Liu, and X. Y ou, “Perform ance anal- ysis of multi-UA V aided cell-free radio access network with network- assisted full-duplex for URLLC,” IEEE Trans. Commun. , vol. 72, no. 9, pp. 5810–5822, 2024

    work page 2024

  19. [19]

    Mitigating the impact of channel aging in cell-fr ee MIMO systems using a channel predictor based on extended Kalman fi lter,

    A. R. L. Paiva, W. C. Freitas, R. P . Antonioli, Y . C. B. Sil va, and G. Fodor, “Mitigating the impact of channel aging in cell-fr ee MIMO systems using a channel predictor based on extended Kalman fi lter,” IEEE Trans. V eh. Technol., vol. 74, no. 6, pp. 8544–8560, Jan. 2025

    work page 2025

  20. [20]

    Accurate channe l prediction based on Transformer: Making mobility negligible,

    H. Jiang, M. Cui, D. W. K. Ng, and L. Dai, “Accurate channe l prediction based on Transformer: Making mobility negligible,” IEEE J. Sel. Areas Commun., vol. 40, no. 9, pp. 2717–2732, Jul. 2022

    work page 2022

  21. [21]

    Deep-learni ng- based spatial–temporal channel prediction for smart high- speed railway communication networks,

    T. Zhou, H. Zhang, B. Ai, C. Xue, and L. Liu, “Deep-learni ng- based spatial–temporal channel prediction for smart high- speed railway communication networks,” IEEE Trans. Wireless Commun. , vol. 21, no. 7, pp. 5333–5345, Jan. 2022

    work page 2022

  22. [22]

    Deep energ y-efficient optimization network for URLLC over cell-free massive MIMO ,

    D. Li, J. Li, D. Niyato, W. Feng, and W. Jiang, “Deep energ y-efficient optimization network for URLLC over cell-free massive MIMO ,” IEEE Internet Things J. , vol. 12, no. 12, pp. 20 973–20 987, Mar. 2025

    work page 2025

  23. [23]

    Unsuper vised deep learning for power control of cell-free massive MIMO system s,

    Y . Zhang, J. Zhang, S. Buzzi, H. Xiao, and B. Ai, “Unsuper vised deep learning for power control of cell-free massive MIMO system s,” IEEE Trans. V eh. Technol., vol. 72, no. 7, pp. 9585–9590, Feb. 2023

    work page 2023

  24. [24]

    Transformer-based RI S phase shift control for ultra-low latency V2X systems,

    H. Kim, S. Kim, J. Wu, and B. Shim, “Transformer-based RI S phase shift control for ultra-low latency V2X systems,” IEEE Trans. Cogn. Commun. Netw., vol. 10, no. 4, pp. 1414–1429, Mar. 2024

    work page 2024

  25. [25]

    A survey on mixture of experts in large language models,

    W. Cai, J. Jiang, F. Wang, J. Tang, S. Kim, and J. Huang, “A survey on mixture of experts in large language models,” IEEE Trans. Knowl. Data Eng. , vol. 37, no. 7, pp. 3896–3915, Mar. 2025

    work page 2025

  26. [26]

    Energy efficiency of wireless-powered cell-free mMIMO wit h hardware impairments,

    X. Chen, C. Zhou, Z. Wang, T. Zhao, Q. Sun, C. Xu, and J. Zha ng, “Energy efficiency of wireless-powered cell-free mMIMO wit h hardware impairments,” IEEE Trans. Commun., vol. 72, no. 7, pp. 4446–4458, Feb. 2024

    work page 2024

  27. [27]

    UA V communications with W PT-aided cell-free massive MIMO systems,

    J. Zheng, J. Zhang, and B. Ai, “UA V communications with W PT-aided cell-free massive MIMO systems,” IEEE J. Sel. Areas Commun. , vol. 39, no. 10, pp. 3114–3128, Jun. 2021

    work page 2021

  28. [28]

    Multi-agent reinfor cement learning-based buffer-aided relay selection in IRS-assis ted secure co- operative networks,

    C. Huang, G. Chen, and K.-K. Wong, “Multi-agent reinfor cement learning-based buffer-aided relay selection in IRS-assis ted secure co- operative networks,” IEEE Trans. Inf. F orensics Security , vol. 16, pp. 4101–4112, Aug. 2021

    work page 2021

  29. [29]

    Achi eving energy-efficient massive URLLC over cell-free massive MIMO ,

    J. Zeng, T. Wu, Y . Song, Y . Zhong, T. Lv, and S. Zhou, “Achi eving energy-efficient massive URLLC over cell-free massive MIMO ,” IEEE Internet Things J. , vol. 11, no. 2, pp. 2198–2210, Jul. 2024

    work page 2024

  30. [30]

    MMSE channel estimation in large-scale MIMO: Improved robustness with r educed complexity,

    G. Bacci, A. Alberto D’Amico, and L. Sanguinetti, “MMSE channel estimation in large-scale MIMO: Improved robustness with r educed complexity,” IEEE Trans. Wireless Commun., vol. 23, no. 12, pp. 18 563– 18 575, Oct. 2024

    work page 2024

  31. [31]

    Tensor-base d unified joint channel estimation and active device detection schem e for high- mobility grant-free random access scenarios,

    Z. Kang, D. He, H. Wang, Z. Wang, and Z. Han, “Tensor-base d unified joint channel estimation and active device detection schem e for high- mobility grant-free random access scenarios,” IEEE Internet Things J. , vol. 12, no. 14, pp. 26 415–26 429, Apr. 2025

    work page 2025

  32. [32]

    Cell-free massive MIMO versus small cells,

    H. Q. Ngo, A. Ashikhmin, H. Y ang, E. G. Larsson, and T. L. M arzetta, “Cell-free massive MIMO versus small cells,” IEEE Trans. Wireless Commun., vol. 16, no. 3, pp. 1834–1850, Jan. 2017

    work page 2017

  33. [33]

    Delay Trade-O ff and adaptive resource element framework for URLLC,

    Y . Zhang, W. Cheng, J. Wang, and W. Zhang, “Delay Trade-O ff and adaptive resource element framework for URLLC,” IEEE Commun. Mag., vol. 63, no. 8, pp. 154–160, Aug. 2025

    work page 2025

  34. [34]

    Sp ectral- efficient RIS-aided RSMA URLLC: Toward mobile broadband rel iable low latency communication (mBRLLC) system,

    S. Pala, M. Katwe, K. Singh, B. Clerckx, and C.-P . Li, “Sp ectral- efficient RIS-aided RSMA URLLC: Toward mobile broadband rel iable low latency communication (mBRLLC) system,” IEEE Trans. Wireless Commun., vol. 23, no. 4, pp. 3507–3524, Sep. 2023

    work page 2023

  35. [35]

    Joint upl ink and downlink resource allocation toward energy-efficient tran smission for URLLC,

    K. Li, P . Zhu, Y . Wang, F.-C. Zheng, and X. Y ou, “Joint upl ink and downlink resource allocation toward energy-efficient tran smission for URLLC,” IEEE J. Sel. Areas Commun. , vol. 41, no. 7, pp. 2176–2192, May. 2023

    work page 2023

  36. [36]

    A survey on vision transformer,

    K. Han, Y . Wang, H. Chen, X. Chen, J. Guo, Z. Liu, Y . Tang, A . Xiao, C. Xu, Y . Xu et al. , “A survey on vision transformer,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 45, no. 1, pp. 87–110, Feb. 2022

    work page 2022

  37. [37]

    Joint powe r allocation and load balancing optimization for energy-efficient cell- free massive MIMO networks,

    T. V an Chien, E. Björnson, and E. G. Larsson, “Joint powe r allocation and load balancing optimization for energy-efficient cell- free massive MIMO networks,” IEEE Trans. Wireless Commun. , vol. 19, no. 10, pp. 6798–6812, Jul. 2020

    work page 2020

  38. [38]

    P erformance analysis of (TDD) massive MIMO with Kalman channel predicti on,

    S. Kashyap, C. Mollén, E. Björnson, and E. G. Larsson, “P erformance analysis of (TDD) massive MIMO with Kalman channel predicti on,” in 2017 IEEE International Conference on Acoustics, Speech an d Signal Processing (ICASSP), 2017, pp. 3554–3558

    work page 2017

  39. [39]

    LSTM-GRU model-bas ed chan- nel prediction for one-bit massive MIMO system,

    I. Helmy, P . Tarafder, and W. Choi, “LSTM-GRU model-bas ed chan- nel prediction for one-bit massive MIMO system,” IEEE Trans. V eh. Technol., vol. 72, no. 8, pp. 11 053–11 057, 2023

    work page 2023

  40. [40]

    Air-groun d collaborative resource optimization in UA V empowered cell-free massive MIMO systems,

    L. Xu, Q. Zhu, W. Xia, T. Q. S. Quek, and H. Zhu, “Air-groun d collaborative resource optimization in UA V empowered cell-free massive MIMO systems,” IEEE Trans. Commun. , vol. 72, no. 4, pp. 2485–2499, Dec. 2024

    work page 2024

  41. [41]

    Line-of- sight probability for mmWave-based UA V communications in 3 D urban grid deployments,

    M. Gapeyenko, D. Moltchanov, S. Andreev, and R. W. Heath , “Line-of- sight probability for mmWave-based UA V communications in 3 D urban grid deployments,” IEEE Trans. Wireless Commun. , vol. 20, no. 10, pp. 6566–6579, 2021

    work page 2021

  42. [42]

    De ep learning for radio resource allocation with diverse Qualit y-of-Service requirements in 5G,

    R. Dong, C. She, W. Hardjawana, Y . Li, and B. Vucetic, “De ep learning for radio resource allocation with diverse Qualit y-of-Service requirements in 5G,” IEEE Trans. Wireless Commun. , vol. 20, no. 4, pp. 2309–2324, Dec. 2021

    work page 2021