AI and Deep Learning for Terahertz Ultra-Massive MIMO: From Model-Driven Approaches to Foundation Models

Engineering ›› DOI: 10.1016/j.eng.2025.07.032

review-article

Author information +

History +

PDF (2528KB)

Abstract

This study explored the transformative potential of artificial intelligence (AI) in addressing the challenges posed by terahertz ultra-massive multiple-input multiple-output (UM-MIMO) systems. It begins by outlining the characteristics of terahertz UM-MIMO systems and identifies three primary challenges for transceiver design: computational complexity, modeling difficulty, and measurement limitations. The study posits that AI provides a promising solution to these challenges. Three systematic research roadmaps are proposed for developing AI algorithms tailored to terahertz UM-MIMO systems. The first roadmap, model-driven deep learning (DL), emphasizes the importance of leveraging available domain knowledge and advocates the adoption of AI only to enhance bottleneck modules within an established signal processing or optimization framework. Four essential steps are discussed: algorithmic frameworks, basis algorithms, loss-function design, and neural architecture design. The second roadmap presents channel state information (CSI) foundation models, aimed at unifying the design of different transceiver modules by focusing on their shared foundation, that is, the wireless channel. The training of a single compact foundation model is proposed to estimate the score function of wireless channels, which serve as a versatile prior for designing a wide variety of transceiver modules. Four essential steps are outlined: general frameworks, conditioning, site-specific adaptation, joint design of CSI foundation models, and model-driven DL. The third roadmap aims to explore potential directions for applying pretrained large language models (LLMs) to terahertz UM-MIMO systems. Several application scenarios are envisioned, including LLM-based estimation, optimization, search, network management, and protocol understanding. Finally, the study highlights open problems and future research directions.

Keywords

Terahertz communications / Ultra-massive multiple-input multiple-output / Model-driven deep learning / Foundation models / Large language models

Highlight

Cite this article

Download citation ▾

Wentao Yu, Hengtao He, Shenghui Song, Jun Zhang, Linglong Dai, Lizhong Zheng, Khaled B. Letaief. AI and Deep Learning for Terahertz Ultra-Massive MIMO: From Model-Driven Approaches to Foundation Models. Engineering DOI:10.1016/j.eng.2025.07.032

登录浏览全文

4963

注册一个新账户忘记密码

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	.Cisco.Cisco annual internet report (2018–2023). Report. San Jose: Cisco Systems, Inc.; 2023.

[2]	Letaief KB, Chen W, Shi Y, Zhang J, Zhang YJA.The roadmap to 6G: AI empowered wireless networks.IEEE Commun Mag 2019; 57(8):84-90.

[3]	Shenzhen: Huawei Technologies Co., Ltd. (2017)

[4]	Letaief KB, Shi Y, Lu J, Lu J.Edge artificial intelligence for 6G: vision, enabling technologies, and applications.IEEE J Sel Areas Commun 2021; 40(1):5-36.

[5]	You X, Wang CX, Huang J, Gao X, Zhang Z, Wang M, et al.Towards 6G wireless communication networks: vision, enabling technologies, and new paradigm shifts.Sci China Inf Sci 2021; 64(1):110301.

[6]	Dang S, Amin O, Shihada B, Alouini MS.What should 6G be?.Nat Electron 2020; 3(1):20-29.

[7]	.Future technology trends of terrestrial international mobile telecommunications systems towards 2030 and beyond.Report. Geneva: International Telecommunication Union; 2022.

[8]	5G Americas.ITU’s IMT-2030 vision: navigating towards 6G in the Americas.Bellevue: 5G Americas; 2024.

[9]	IM T-2030 (6G) Promotion Group.White paper on 6G vision and candidate technologies.Report. Shenzhen: IM T-2030 (6G) Promotion Group; 2021.

[10]	Rappaport TS, Xing Y, Kanhere O, Ju S, Madanayake A, Mandal S, et al.Wireless communications and applications above 100 GHz: opportunities and challenges for 6G and beyond.IEEE Access 2019; 7:78729-78757.

[11]	Neil: Federal Communications Commission (2018)

[12]	Petrov V, Kurner T, Hosako I.IEEE 802.15.3d: first standardization efforts for sub-terahertz band communications toward 6G.IEEE Commun Mag 2020; 58(11):28-33.

[13]	Marzetta TL.Noncooperative cellular wireless with unlimited numbers of base station antennas.IEEE Trans Wirel Commun 2010; 9(11):3590-3600.

[14]	Björnson E, Chae CB, Heath RWJr, Marzetta TL, Mezghani A, Sanguinetti L, et al. Towards 6G MIMO: massive spatial multiplexing, arXiv:240102844 (2024)

[15]	Akyildiz IF, Jornet JM.Realizing ultra-massive MIMO (1024 × 1024) communication in the (0.06–10) terahertz band. Nano Commun Netw 2016;8:46–54.

[16]	Faisal A, Sarieddeen H, Dahrouj H, Al-Naffouri TY, Alouini MS.Ultramassive MIMO systems at terahertz bands: prospects and challenges.IEEE Veh Technol Mag 2020; 15(4):33-42.

[17]	Ning B, Tian Z, Mei W, Chen Z, Han C, Li S, et al.Beamforming technologies for ultra-massive MIMO in terahertz communications.IEEE Open J Commun Soc 2023; 4:614-658.

[18]	Sarieddeen H, Alouini MS, Al-Naffouri TY.An overview of signal processing techniques for terahertz communications.Proc IEEE 2021; 109(10):1628-1665.

[19]	Chen H, Sarieddeen H, Ballal T, Wymeersch H, Alouini MS, Al-Naffouri TY.A tutorial on terahertz-band localization for 6G communication systems.IEEE Commun Surv Tutor 2022; 24(3):1780-1815.

[20]	Elbir AM, Mishra KV, Chatzinotas S, Bennis M.Terahertz-band integrated sensing and communications: challenges and opportunities.IEEE Aerosp Electron Syst Mag 2024; 39(12):38-49.

[21]	Han C, Wu Y, Chen Z, Chen Y, Wang G.THz ISAC: a physical-layer perspective of terahertz integrated sensing and communication.IEEE Commun Mag 2024; 62(2):102-108.

[22]	Lemic F, Abadal S, Tavernier W, Stroobant P, Colle D, Alarcon E, et al.Survey on terahertzx nanocommunication and networking: a top–down perspective.IEEE J Sel Areas Commun 2021; 39(6):1506-1543.

[23]	Jornet JM, Akyildiz IF.Graphene-based plasmonic nano-antenna for terahertz band communication in nanonetworks.IEEE J Sel Areas Commun 2013; 31(12):685-694.

[24]	Song HJ, Lee N.Terahertz communications: challenges in the next decade.IEEE Trans Terahertz Sci Technol 2021; 12(2):105-117.

[25]	Do H, Cho S, Park J, Song HJ, Lee N, Lozano A.Terahertz line-of-sight MIMO communication: theory and practical challenges.IEEE Commun Mag 2021; 59(3):104-109.

[26]	Yang N, Shafie A.Terahertz communications for massive connectivity and security in 6G and beyond era.IEEE Commun Mag 2024; 62(2):72-78.

[27]	Shafie A, Yang N, Han C, Jornet JM, Juntti M, Kürner T.Terahertz communications for 6G and beyond wireless networks: challenges, key advancements, and opportunities.IEEE Netw 2023; 37(3):162-169.

[28]	Han C, Chen Y, Yan L, Chen Z, Dai L.Cross far-and near-field wireless communications in terahertz ultra-large antenna array systems.IEEE Wirel Commun 2024; 31(3):148-154.

[29]	Yu W, Shen Y, He H, Yu X, Zhang J, Letaief KB.Hybrid far-and near-field channel estimation for THz ultra-massive MIMO via fixed point networks.In: Proceedings of IEEEGlobal Communications Conference; 2022 Dec 4–8; Rio de Janeiro, Brazil. New York City: IEE E; 2022. p. 5384–9.

[30]	Wang J, Wang CX, Huang J, Wang H, Gao X.A general 3D space–time–frequency non-stationary THz channel model for 6G ultra-massive MIMO wireless communication systems.IEEE J Sel Areas Commun 2021; 39(6):1576-1589.

[31]	Carvalho ED, Ali A, Amiri A, Angjelichinoski M, Heath RW.Non-stationarities in extra-large-scale massive MIMO.IEEE Wirel Commun 2020; 27(4):74-80.

[32]	Wang B, Jian M, Gao F, Li GY, Lin H.Beam squint and channel estimation for wideband mmWave massive MIMO-OFDM systems.IEEE Trans Signal Process 2019; 67(23):5893-5908.

[33]	Zhang J, Yu X, Letaief KB.Hybrid beamforming for 5G and beyond millimeter-wave systems: a holistic view.IEEE Open J Commun Soc 2020; 1:77-91.

[34]	Lin C, Li GY.Terahertz communications: an array-of-subarrays solution.IEEE Commun Mag 2016; 54(12):124-131.

[35]	Wu Y, Koch JD, Vossiek M, Schober R, Gerstacker W.ML detection without CSI for constant-weight codes in THz communications with strong phase noise.In: Proceedings of IEEEGlobal Communications Conference; 2022 Dec 4–8; Rio de Janeiro, Brazil. New York City: IEE E; 2022. p. 831–6.

[36]	Bicais S, Dore JB.Design of digital communications for strong phase noise channels.IEEE Open J Veh Technol 2020; 1:227-243.

[37]	Sha Z, Wang Z.Channel estimation and equalization for terahertz receiver with RF impairments.IEEE J Sel Areas Commun 2021; 39(6):1621-1635.

[38]	Yu W, Shen Y, He H, Yu X, Song S, Zhang J, et al.An adaptive and robust deep learning framework for THz ultra-massive MIMO channel estimation.IEEE J Sel Top Signal Process 2023; 17(4):761-776.

[39]	Wang Z, Zhang J, Du H, Niyato D, Cui S, Ai B, et al.A tutorial on extremely large-scale MIMO for 6G: fundamentals, signal processing, and applications.IEEE Commun Surv Tutor 2024; 26(3):1560-1605.

[40]	Björnson E, Sanguinetti L, Wymeersch H, Hoydis J, Marzetta TL.Massive MIMO is a reality–what is next? Five promising research directions for antenna arrays.Digit Signal Process 2019; 94:3-20.

[41]	Marinello JC, Abr Tão, Amiri A, de ECarvalho, Popovski P.Antenna selection for improving energy efficiency in XL-MIMO systems.IEEE Trans Veh Technol 2020; 69(11):13305-13318.

[42]	Ye S, Xiao M, Kwan MW, Ma Z, Huang Y, Karagiannidis G, et al.Extremely large aperture array (ELAA) communications: foundations, research advances and challenges.IEEE Open J Commun Soc 2024; 5:7075-7120.

[43]	Amiri A, Angjelichinoski M, de Carvalho E, Heath RW.Extremely large aperture massive MIMO: low complexity receiver architectures.In: Proceedings of the 2018 IEEE Globecom Workshops; 2018 Dec 9–13; Abu Dhabi, United Arab Emirates. New York City: IEE E; 2018. p. 1–6.

[44]	Björnson E, Kara F, Kolomvakis N, Kosasih A, Ramezani P, Salman MB. Enabling 6G performance in the upper mid-band by transitioning from massive to gigantic MIMO., arXiv:240705630 (2024)

[45]	O T’Shea, Hoydis J.An introduction to deep learning for the physical layer.IEEE Trans Cogn Commun Netw 2017; 3(4):563-575.

[46]	Ye N, Miao S, Pan J, Ouyang Q, Li X, Hou X.Artificial intelligence for wireless physical-layer technologies (AI4PHY): a comprehensive survey.IEEE Trans Cogn Commun Netw 2024; 10(3):729-755.

[47]	Shi Y, Lian L, Shi Y, Wang Z, Zhou Y, Fu L.Machine learning for large-scale optimization in 6G wireless networks.IEEE Commun Surv Tutor 2023; 25(4):2088-2132.

[48]	Bjornson E, Giselsson P.Two applications of deep learning in the physical layer of communication systems.IEEE Signal Process Mag 2020; 37(5):134-140.

[49]	Bengio Y, Lodi A, Prouvost A.Machine learning for combinatorial optimization: a methodological tour d’horizon.Eur J Oper Res 2021; 290(2):405-421.

[50]	He H, Jin S, Wen CK, Gao F, Li GY, Xu Z.Model-driven deep learning for physical layer communications.IEEE Wirel Commun 2019; 26(5):77-83.

[51]	Qin Z, Ye H, Li GY, Juang BHF.Deep learning in physical layer communications.IEEE Wirel Commun 2019; 26(2):93-99.

[52]	Van N Huynh, Wang J, Du H, Hoang DT, Niyato D, Nguyen DN, et al.Generative AI for physical layer communications: a survey.IEEE Trans Cogn Commun Netw 2024; 10(3):706-728.

[53]	Andrews JG, Humphreys TE, Ji T.6G takes shape.IEEE BITS Inf Theory Mag 2024; 4(1):2-24.

[54]	Yu W, Sohrabi F, Jiang T.Role of deep learning in wireless communications.IEEE BITS Inf Theory Mag 2022; 2(2):56-72.

[55]	He H, Wen CK, Jin S, Li GY.Model-driven deep learning for MIMO detection.IEEE Trans Signal Process 2020; 68:1702-1715.

[56]	Shafin R, Liu L, Chandrasekhar V, Chen H, Reed J, Zhang JC.Artificial intelligence-enabled cellular networks: a critical path to beyond-5G and 6G.IEEE Wirel Commun 2020; 27(2):212-217.

[57]	Raviv T, Shlezinger N.Data augmentation for deep receivers.IEEE Trans Wirel Commun 2023; 22(11):8259-8274.

[58]	Soltani N, Sankhe K, Dy J, Ioannidis S, Chowdhury K.More is better: data augmentation for channel-resilient RF fingerprinting.IEEE Commun Mag 2020; 58(10):66-72.

[59]	Akyildiz IF, Han C, Hu Z, Nie S, Jornet JM.Terahertz band communication: an old problem revisited and research directions for the next decade.IEEE Trans Commun 2022; 70(6):4250-4285.

[60]	Bodet D, Hall J, Masihi A, Thawdar N, Melodia T, Restuccia F, et al.Data signals for deep learning applications in terahertz communications.Comput Netw 2024; 254:110800.

[61]	Hall J, Jornet JM, Thawdar N, Melodia T, Restuccia F.Deep learning at the physical layer for adaptive terahertz communications.IEEE Trans Terahertz Sci Technol 2023; 13(2):102-112.

[62]	Tarboush S, Sarieddeen H, Chen H, Loukil MH, Jemaa H, Alouini MS, et al.TeraMIMO: a channel simulator for wideband ultra-massive MIMO terahertz communications.IEEE Trans Veh Technol 2021; 70(12):12325-12341.

[63]	Han C, Gao W, Yang N, Jornet JM.Molecular absorption effect: a double-edged sword of terahertz communications.IEEE Wirel Commun 2022; 30(4):140-146.

[64]	Ayach OE, Rajagopal S, Abu-Surra S, Pi Z, Heath RW.Spatially sparse precoding in millimeter wave MIMO systems.IEEE Trans Wirel Commun 2014; 13(3):1499-1513.

[65]	Yu X, Shen JC, Zhang J, Letaief KB.Alternating minimization algorithms for hybrid precoding in millimeter wave MIMO systems.IEEE J Sel Top Signal Process 2016; 10(3):485-500.

[66]	Huang Y, Li Y, Ren H, Lu J, Zhang W.Multi-panel MIMO in 5G.IEEE Commun Mag 2018; 56(3):56-61.

[67]	Dai J, Liu A, Lau VKN.FDD massive MIMO channel estimation with arbitrary 2D-array geometry.IEEE Trans Signal Process 2018; 66(10):2584-2599.

[68]	Gao F, Wang B, Xing C, An J, Li GY.Wideband beamforming for hybrid massive MIMO terahertz communications.IEEE J Sel Areas Commun 2021; 39(6):1725-1740.

[69]	Tan J, Dai L.Delay-phase precoding for THz massive MIMO with beam split.In: Proceedings of the 2019 IEEE Global Communications Conference; 2019 Dec 9–13; Big Island, H I, USA. New York City: IEE E; 2019. p. 1–6.

[70]	Tan J, Dai L.THz precoding for 6G: challenges, solutions, and opportunities.IEEE Wirel Commun 2023; 30(4):132-138.

[71]	Ratnam VV, Mo J, Alammouri A, Ng BL, Zhang J, Molisch AF.Joint phase-time arrays: a paradigm for frequency-dependent analog beamforming in 6G.IEEE Access 2022; 10:73364-73377.

[72]	Zheng T, Cui M, Wu Z, Dai L.Near-field wideband beam training based on distance-dependent beam split.IEEE Trans Wirel Commun 2024; 24(2):1-14.

[73]	Tan J, Dai L.Wideband beam tracking in THz massive MIMO systems.IEEE J Sel Areas Commun 2021; 39(6):1693-1710.

[74]	Serghiou D, Khalily M, Brown TW, Tafazolli R.Terahertz channel propagation phenomena, measurement techniques and modeling for 6G wireless communication applications: a survey, open challenges and future research directions.IEEE Commun Surv Tutor 2022; 24(4):1957-1996.

[75]	Han C, Wang Y, Li Y, Chen Y, Abbasi NA, Kurner T, et al.Terahertz wireless channels: a holistic survey on measurement, modeling, and analysis.IEEE Commun Surv Tutor 2022; 24(3):1670-1707.

[76]	Petrov V, Jornet JM, Singh A.Near-field 6G Networks: why mobile terahertz communications MUST operate in the near field.In: Proceedings of IEEEGlobal Communications Conference; 2023 Dec 4–8; Kuala Lumpur, Malaysia. New York City: IEE E; 2023. p. 3983–9.

[77]	Jiang JS, Ingram MA.Spherical-wave model for short-range MIMO.IEEE Trans Commun 2005; 53(9):1534-1541.

[78]	Wei X, Dai L.Channel estimation for extremely large-scale massive MIMO: far-field, near-field, or hybrid-field?.IEEE Commun Lett 2022; 26(1):177-181.

[79]	Tarboush S, Ali A, Al-Naffouri TY.Cross-field channel estimation for ultra massive-MIMO THz systems.IEEE Trans Wirel Commun 2024; 23(8):8619-8653.

[80]	Yan L, Han C, Yuan J.A dynamic array-of-subarrays architecture and hybrid precoding algorithms for terahertz wireless communications.IEEE J Sel Areas Commun 2020; 38(9):2041-2056.

[81]	Gao W, Chen Y, Han C, Chen Z.Distance-adaptive absorption peak modulation (DA-APM) for terahertz covert communications.IEEE Trans Wirel Commun 2021; 20(3):2064-2077.

[82]	Heng Y, Mo J, Andrews JG.Learning site-specific probing beams for fast mmWave beam alignment.IEEE Trans Wirel Commun 2022; 21(8):5785-5800.

[83]	Hoydis J, Cammerer S, Aoudia FA, Vem A, Binder N, Marcus G, et al. Sionna: an open-source library for next-generation physical layer research., arXiv:220311854 (2022)

[84]	Cui M, Wu Z, Lu Y, Wei X, Dai L.Near-field MIMO communications for 6G: fundamentals, challenges, potentials, and future directions.IEEE Commun Mag 2023; 61(1):40-46.

[85]	Yu W, Ma Y, He H, Song S, Zhang J, Letaief KB.Deep learning for near-field XL-MIMO transceiver design: principles and techniques.IEEE Commun Mag 2024; 63(1):52-58.

[86]	Forsch C, Alrabadi O, Brueck S, Gerstacker W. Phase noise robust terahertz communications, IEEE, Helsinki, Finland. New York City (2022), pp. 1-6

[87]	Chan WL, Moravec ML, Baraniuk RG, Mittleman DM.Terahertz imaging with compressed sensing and phase retrieval.Opt Lett 2008; 33(9):974-976.

[88]	Cao R, He H, Yu X, Song S, Huang K, Zhang J, et al. Joint channel estimation and cooperative localization for near-field ultra-massive MIMO., arXiv:231213683 (2023)

[89]	Liu S, Yu X, Gao Z, Xu J, Ng DWK, Cui S.Sensing-enhanced channel estimation for near-field XL-MIMO systems.IEEE J Sel Areas Commun 2024; 43(3):628-643.

[90]	Cao R, Yu W, He H, Yu X, Song S, Zhang J. Newtonized near-field channel estimation for ultra-massive MIMO Systems, IEEE, Dubai, United Arab Emirates. New York City (2024), pp. 1-6

[91]	Lee T, Park J, Kim H, Andrews JG. Generating high dimensional user-specific wireless channels using diffusion models., arXiv:240903924 (2024)

[92]	Chi G, Yang Z, Wu C, Xu J, Gao Y, Liu Y, et al. RF-diffusion: radio signal generation via time–frequency diffusion, Association for Computing Machinery (ACM), Washington, DC, USA. New York City (2024), pp. 77-92

[93]	Arvinte M, Tamir JI.MIMO channel estimation using score-based generative models.IEEE Trans Wirel Commun 2023; 22(6):3698-3713.

[94]	Yu W, He H, Yu X, Song S, Zhang J, Murch R, et al.Bayes-optimal unsupervised learning for channel estimation in near-field holographic MIMO.IEEE J Sel Top Signal Process 2024; 18(4):714-729.

[95]	Olutayo T, Champagne B. Score-based generative modeling for MIMO detection without knowledge of noise statistics, IEEE, Toronto, ON, Canada. New York City (2023), pp. 1-7

[96]	He K, He L, Fan L, Deng Y, Karagiannidis GK, Nallanathan A.Learning-based signal detection for MIMO systems with unknown noise statistics.IEEE Trans Commun 2021; 69(5):3025-3038.

[97]	Balevi E, Andrews JG.Unfolded hybrid beamforming with GAN compressed ultra-low feedback overhead.IEEE Trans Wirel Commun 2021; 20(12):8381-8392.

[98]	Jayashankar T, Lee GCF, Lancho A, Weiss A, Polyanskiy Y, Wornell G. Score-based source separation with applications to digital communication signals, Curran Associates Inc., New Orleans, LA, USA. Red Hook (2023), pp. 5092-5125

[99]	Ma Y, Shen Y, Yu X, Zhang J, Song SH, Letaief KB.Learn to communicate with neural calibration: scalability and generalization.IEEE Trans Wirel Commun 2022; 21(11):9947-9961.

[100]

Nguyen NT, Ma M, Lavi O, Shlezinger N, Eldar YC, Lee SA.Deep unfolding hybrid beamforming designs for THz massive MIMO systems.IEEE Trans Signal Process 2023; 71:3788-3804.

[101]

He H, Wang R, Jin W, Jin S, Wen CK, Li GY.Beamspace channel estimation for wideband millimeter-wave MIMO: a model-driven unsupervised learning approach.IEEE Trans Wirel Commun 2022; 22(3):1808-1822.

[102]

He H, Yu X, Zhang J, Song S, Letaief KB.Message passing meets graph neural networks: a new paradigm for massive MIMO systems.IEEE Trans Wirel Commun 2024; 23(5):4709-4723.

[103]

Liu G, Hu Z, Wang L, Zhang H, Xue J, Matthaiou M.A hypernetwork based framework for non-stationary channel prediction.IEEE Trans Veh Technol 2024; 73(6):8338-8351.

[104]

Ding Y, Rao BD.Dictionary learning-based sparse channel representation and estimation for FDD massive MIMO systems.IEEE Trans Wirel Commun 2018; 17(8):5437-5451.

[105]

Wen C, Tong J, Hu Y, Lin Z, Zhang J. WRF-GS: wireless radiation field reconstruction with 3D Gaussian splatting, IEEE, London, UK. New York City (2025), pp. 1-10

[106]

Meng X, Kabashima Y.Quantized compressed sensing with score-based generative models.In: Proceedings of the Eleventh International Conference on Learning Representations ICLR 2023; 2023 May 1–5; Kigali, Rwanda. New York City: IEE E; 2023.

[107]

Elata N, Michaeli T, Elad M. Adaptive compressed sensing with diffusion-based posterior sampling, Springer, Milan, Italy. Berlin (2025), pp. 290-308

[108]

Zilberstein N, Dick C, Doost-Mohammady R, Sabharwal A, Segarra S.Annealed Langevin dynamics for massive MIMO detection.IEEE Trans Wirel Commun 2022; 22(6):3762-3776.

[109]

Wang Z, Zhou Y, Shi Y, Letaief KB. Federated fine-tuning for pre-trained foundation models over wireless networks., arXiv:240702924 (2024)

[110]

Liu B, Liu X, Gao S, Cheng X, Yang L.LLM4CP: adapting large language models for channel prediction.J Commun Inf Netw 2024; 9(2):113-125.

[111]

Shao J, Tong J, Wu Q, Guo W, Li Z, Lin Z, et al.WirelessLLM: empowering large language models towards wireless intelligence.J Commun Inf Netw 2024; 9(2):99-112.

[112]

Romera-Paredes B, Barekatain M, Novikov A, Balog M, Kumar MP, Dupont E, et al.Mathematical discoveries from program search with large language models.Nature 2024; 625(7995):468-475.

[113]

Shen Y, Shao J, Zhang X, Lin Z, Pan H, Li D, et al.Large language models empowered autonomous edge AI for connected intelligence.IEEE Commun Mag 2024; 62(10):140-146.

[114]

Zou H, Zhao Q, Tian Y, Bariah L, Bader F, Lestable T, et al. TelecomGPT: a framework to build telecom-specfic large language models., arXiv:2407.09424 (2024)

[115]

Hornik K, Stinchcombe M, White H.Multilayer feedforward networks are universal approximators.Neural Netw 1989; 2(5):359-366.

[116]

Ye H, Li GY, Juang BH.Power of deep learning for channel estimation and signal detection in OFDM systems.IEEE Wirel Commun Lett 2018; 7(1):114-117.

[117]

Sun H, Chen X, Shi Q, Hong M, Fu X, Sidiropoulos ND.Learning to optimize: training deep neural networks for interference management.IEEE Trans Signal Process 2018; 66(20):5438-5453.

[118]

He H, Wen CK, Jin S, Li GY.Deep learning-based channel estimation for beamspace mmWave massive MIMO systems.IEEE Wirel Commun Lett 2018; 7(5):852-855.

[119]

Zhang J, Wen CK, Liang L, Jin S.Universal model-driven deep learning for MIMO transceiver design.IEEE Commun Mag 2023; 62(4):74-80.

[120]

Ma Y, Yu W, Yu X, Zhang J, Song S, Letaief KB. Lightweight and flexible deep equilibrium learning for CSI feedback in FDD massive MIMO, IEEE, Stockholm, Sweden. New York City (2024), pp. 299-304

[121]

Bauschke H, Combettes P.Convex analysis and monotone operator theory in Hilbert spaces. Springer, Berlin (2011)

[122]

Fung SW, Heaton H, Li Q, McKenzie D, Osher S, Yin WJFB.Jacobian-free backpropagation for implicit networks.online, The Association for the Advancement of Artificial Intelligence (AAAI), Washington, DC 2022; 6648-6656.

[123]

Gao J, Chen X, Li GY.Deep unfolding based channel estimation for wideband terahertz near-field massive MIMO systems.Front Inf Technol Electron Eng 2024; 25(8):1162-1172.

[124]

Metzler CA, Maleki A, Baraniuk RG.From denoising to compressed sensing.IEEE Trans Inf Theory 2016; 62(9):5117-5144.

[125]

Shi Q, Razaviyayn M, Luo ZQ, He C.An iteratively weighted MMSE approach to distributed sum-utility maximization for a MIMO interfering broadcast channel.IEEE Trans Signal Process 2011; 59(9):4331-4340.

[126]

Zhao W, Han C, Song HJ, Björnson E. DNN based two-stage compensation algorithm for THz hybrid beamforming with imperfect hardware., arXiv:2411.14699 (2024)

[127]

He H, Wen CK, Jin S. Generalized expectation consistent signal recovery for nonlinear measurements, IEEE, Aachen, Germany. New York City (2017), pp. 2333-2337

[128]

Huang H, Xia W, Xiong J, Yang J, Zheng G, Zhu X.Unsupervised learning-based fast beamforming design for downlink MIMO.IEEE Access 2018; 7:7599-7605.

[129]

Stein CM.Estimation of the mean of a multivariate normal distribution.Ann Stat 1981; 9(6):1135-1151.

[130]

Eldar YC.Generalized SURE for exponential families: applications to regularization.IEEE Trans Signal Process 2009; 57(2):471-481.

[131]

Yu W, He H, Yu X, Song S, Zhang J, Letaief KB. Blind performance prediction for deep learning based ultra-massive MIMO channel estimation, IEEE, Rome, Italy. New York City (2023), pp. 2613-2618

[132]

Tian H, Lian L.GSURE-based unsupervised deep equilibrium model learning for large-scale channel estimation.Cape Town, South Africa, IEEE, New York City 2024; 1-6.

[133]

Raphan M, Simoncelli EP.Least squares estimation without priors or supervision.Neural Comput 2011; 23(2):374-420.

[134]

Shen Y, Shi Y, Zhang J, Letaief KB.Graph neural networks for scalable radio resource management: architecture design and theoretical analysis.IEEE J Sel Areas Commun 2021; 39(1):101-115.

[135]

Shen Y, Zhang J, Song SH, Letaief KB.Graph neural networks for wireless communications: from theory to practice.IEEE Trans Wirel Commun 2023; 22(5):3554-3569.

[136]

He H, Kosasih A, Yu X, Zhang J, Song H, Hardjawana W. GNN-enhanced approximate message passing for massive/ultra-massive MIMO detection, IEEE, Glasgow, UK. New York City (2023), pp. 1-6

[137]

He S, Xiong S, Ou Y, Zhang J, Wang J, Huang Y, et al.An overview on the application of graph neural networks in wireless networks.IEEE Open J Commun Soc 2021; 2:2547-2565.

[138]

Ha D, Dai A, Le QV.Hypernetworks.2016. ar Xiv:160909106.

[139]

Jin W, He H, Wen CK, Jin S, Li GY. Adaptive channel estimation based on model-driven deep learning for wideband mmWave systems, IEEE, Madrid, Spain. New York City (2021), pp. 1-6

[140]

Xie S, He H, Li H, Song S, Zhang J, Zhang YJA, et al. Deep learning-based adaptive joint source-channel coding using hypernetworks, IEEE, Madrid, Spain. New York City (2024), pp. 191-196

[141]

Cui M, Dai L.Channel estimation for extremely large-scale MIMO: far-field or near-field?.IEEE Trans Commun 2022; 70(4):2663-2677.

[142]

Zhao X, An Z, Pan Q, Yang L. Nerf2: neural radio-frequency radiance fields, Association for Computing Machinery (ACM), Madrid, Spain. New York City (2023), pp. 1-15

[143]

Zhang H, Shlezinger N, Guidi F, Dardari D, Imani MF, Eldar YC.Beam focusing for near-field multiuser MIMO communications.IEEE Trans Wirel Commun 2022; 21(9):7476-7490.

[144]

Wan Z, Gao Z, Gao F, Di MRenzo, Alouini MS.Terahertz massive MIMO with holographic reconfigurable intelligent surfaces.IEEE Trans Commun 2021; 69(7):4732-4750.

[145]

Elbir AM, Mishra KV, Chatzinotas S.Terahertz-band joint ultra-massive MIMO radar-communications: model-based and model-free hybrid beamforming.IEEE J Sel Top Signal Process 2021; 15(6):1468-1483.

[146]

Ma J, Ping L.Orthogonal AMP.IEEE Access 2017; 5:2020-2033.

[147]

Cespedes J, Olmos PM, Sanchez-Fernandez M, Perez-Cruz F.Expectation propagation detection for high-order high-dimensional MIMO systems.IEEE Trans Commun 2014; 62(8):2840-2849.

[148]

Wu S, Kuang L, Ni Z, Lu J, Huang D, Guo Q.Low-complexity iterative detection for large-scale multiuser MIMO-OFDM systems using approximate message passing.IEEE J Sel Top Signal Process 2014; 8(5):902-915.

[149]

Xu X, Zheng L.Neural feature learning in function space.J Mach Learn Res 2024; 25(142):1-76.

[150]

Xu X, Zheng L. Multiuser detection with neural feature learning, IEEE, Washington, DC, USA. New York City (2024), pp. 715-720

[151]

Xu X, Zheng L, Agrawal I. Neural feature learning for engineering problems, IEEE, Monticello, IL, USA. New York City (2023), pp. 1-8

[152]

Song Y, Sohl-Dickstein J, Kingma DP, Kumar A, Ermon S, Poole B.Score-based generative modeling through stochastic differential equations.online, The dblp Computer Science Bibliography, Trier 2021; 37799-37812.

[153]

Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models, Curran Associates Inc., Vancouver, BC, Canada. Red Hook (2020), pp. 6840-6851

[154]

Alikhani S, Charan G, Alkhateeb A. Large wireless model (LWM): a foundation model for wireless channels., arXiv:241108872 (2024)

[155]

Song Y, Ermon S. Generative modeling by estimating gradients of the data distribution, Curran Associates Inc., Vancouver, BC, Canada. Red Hook (2019), pp. 11918-11930

[156]

Vincent P.A connection between score matching and denoising autoencoders.Neural Comput 2011; 23(7):1661-1674.

[157]

Cai C, Yuan X, Zhang YJA. Score-based turbo message passing for plug-and-play compressive image recovery., arXiv:2503.22140 (2025)

[158]

Rombach R, Blattmann A, Lorenz D, Esser P, Ommer B. High-resolution image synthesis with latent diffusion models, IEEE, Vancouver, BC, Canada. New York City (2022), pp. 10684-10695

[159]

Yu W, He H, Yu X, Song S, Zhang J, Murch RD. Learning Bayes-optimal channel estimation for holographic MIMO in unknown EM environments, IEEE, Denver, CO, USA. New York City (2024), pp. 3592-3597

[160]

Aali A, Arvinte M, Kumar S, Tamir JI. Solving inverse problems with score-based generative priors learned from noisy data, IEEE, Grove, CA, USA. New York City (2023), pp. 837-843

[161]

Du H, Zhang R, Liu Y, Wang J, Lin Y, Li Z, et al.Enhancing deep reinforcement learning: a tutorial on generative diffusion models in network optimization.IEEE Commun Surv Tutor 2024; 26(4):2611-2646.

[162]

Kim M, Fritschek R, Schaefer RF. Learning end-to-end channel coding with diffusion models, IEEE, Braunschweig, Germany. New York City (2023), pp. 1-6

[163]

Letafati M, Ali S, Latva-aho M.Diffusion models for wireless communications.2023. arXiv: 2310.07312.

[164]

Meng X, Kabashima Y. Diffusion model based posterior sampling for noisy linear inverse problems., arXiv:2211.12343 (2024)

[165]

Zhou X, Liang L, Zhang J, Jiang P, Li Y, Jin S.Generative diffusion models for high dimensional channel estimation.2024. ar Xiv:240810501.

[166]

Guo J, Wen CK, Jin S, Li GY.Overview of deep learning-based CSI feedback in massive MIMO systems.IEEE Trans Commun 2022; 70(12):8017-8045.

[167]

Zilberstein N, Swami A, Segarra S.Joint channel estimation and data detection in massive MIMO systems based on diffusion models.In: Proceedings of IEEEInternational Conference on Acoustics, Speech and Signal Processing; 2024 Apr 14–19; Seoul, Republic of Korea. New York City: IEE E; 2024. p. 13291–5.

[168]

Ho J, Salimans T. Classifier-free diffusion guidance., arXiv:220712598 (2022)

[169]

Dhariwal P, Nichol A.Diffusion models beat GANs on image synthesis.online, Curran Associates Inc., Red Hook 2021; 8780-8794.

[170]

Xie H, Qin Z, Li GY, Juang BH.Deep learning enabled semantic communication systems.IEEE Trans Signal Process 2021; 69:2663-2675.

[171]

Li H, Yu W, He H, Shao J, Song S, Zhang J.Task-oriented communication with out-of-distribution detection: an information bottleneck framework.In: Proceedings of IEEEGlobal Communications Conference; 2023 Dec 4–8; Kuala Lumpur, Malaysia. New York City: IEE E; 2023. p. 3136–41.

[172]

Shao J, Mao Y, Zhang J.Learning task-oriented communication for edge inference: an information bottleneck approach.IEEE J Sel Areas Commun 2022; 40(1):197-211.

[173]

Houlsby N, Giurgiu A, Jastrzebski S, Morrone B, De Laroussilhe Q, Gesmundo A, et al.Parameter-efficient transfer learning for NLP.In: Proceedings of the 36th International Conference on Machine Learning; 2019 Jun 10–15;Long Beach, C A, USA. New York City: MLResearch Press;2019. p.2790–9.

[174]

Bu Z, Wang YX, Zha S, Karypis G. Differentially private bias-term fine-tuning of foundation models, JMLR.org, Vienna, Austria. New York City (2024), pp. 4730-4751

[175]

Hu EJ, Shen Y, Waills P, Allen-Zhu Z, Li Y, Wang S, et al.LoRA: low-rank adaptation of large language models.2021. ar Xiv:210609685.

[176]

Sun H, Tian H, Ni W, Zheng J, Niyato D, Zhang P.Federated low-rank adaptation for large models fine-tuning over wireless networks.IEEE Trans Wirel Commun 2025; 24(1):659-675.

[177]

Kang T, Wang Z, He H, Zhang J, Song S, Letaief KB. Federated low-rank adaptation with differential privacy over wireless networks., arXiv:241107806 (2024)

[178]

Zhang K, He H, Song S, Zhang J, Letaief KB. Communication-efficient distributed on-device LLM inference over wireless networks., arXiv:2503.14882 (2025)

[179]

Zou Q, Yang H. A concise tutorial on approximate message passing., arXiv:220107487 (2022)

[180]

Efron B.Tweedie’s formula and selection bias.J Am Stat Assoc 2011; 106(496):1602-1614.

[181]

Wang K, Gao Z, Chen S, Ning B, Chen G, Su Y.Knowledge and data dual-driven channel estimation and feedback for ultra-massive MIMO systems under hybrid field beam squint effect.IEEE Trans Wirel Commun 2024; 23(9):11240-11259.

[182]

Open AI, Achiam J, Adler S, Agarwai S, Ahmad L, Akkaya I, et al. GPT-4 technical report., arXiv:2303.08774 (2024)

[183]

DeepSeek-AI GD, Yang D, Zhang H, Song J, Zhang R, et al. DeepSeek-R1: incentivizing reasoning capability in LLMs via reinforcement learning., arXiv:2501.12948 (2025)

[184]

Zhao WX, Zhou K, Li J, Tang T, Wang X, Hou Y, et al. A survey of large language models., arXiv:2303.18223 (2025)

[185]

Liu B, Gao S, Liu X, Cheng X, Yang L. WiFo: wireless foundation model for channel prediction., arXiv:2412.08908 (2025)

[186]

Yu L, Shi L, Zhang J, Wang J, Zhang Z, Zhang Y, et al. ChannelGPT: a large model to generate digital twin channel for 6G environment intelligence., arXiv:2410.13379 (2024)

[187]

Zheng T, Dai L. Large language model enabled multi-task physical layer network., arXiv:2412.20772 (2025)

[188]

Liu X, Gao S, Liu B, Cheng X, Yang L. LLM4WM: adapting LLM for wireless multi-tasking., arXiv:2501.12983 (2025)

[189]

Yang T, Zhang P, Zheng M, Shi Y, Jing L, Huang J, et al. WirelessGPT: a generative pre-trained multi-task learning framework for wireless communication., arXiv:2502.06877 (2025)

[190]

Zhang E, Goto R, Sagan N, Mutter J, Phillips N, Alizadeh A, et al. LLM-Lasso: a robust framework for domain-informed feature selection and regularization., arXiv:2502.10648 (2025)

[191]

Lee W, Park J. LLM-empowered resource allocation in wireless communications systems., arXiv:2408.02944 (2024)

[192]

Zhou H, Hu C, Yuan D, Yuan Y, Wu D, Chen X, et al. Large language models for wireless networks: an overview from the prompt engineering perspective., arXiv:2411.04136 (2024)

[193]

Nazar AM, Celik A, Selim MY, Abdallah A, Qiao D, Eltawil AM. ENWAR: a RAG-empowered multi-modal LLM framework for wireless environment perception., arXiv:2410.18104 (2024)

[194]

Weindel F, Heckel R. LLM-guided search for deletion-correcting codes., arXiv:2504.00613 (2025)

[195]

Tong J, Shao J, Wu Q, Guo W, Li Z, Lin Z, et al. WirelessAgent: large language model agents for intelligent wireless networks., arXiv:2409.07964 (2024)

[196]

Shao J, Li X.AI flow at the network edge. IEEE Netw. In press.

[197]

Nikbakht R, Benzaghta M, Geraci G. TSpec-LLM: an open-source dataset for LLM understanding of 3GPP specifications., arXiv:2406.01768 (2024)

[198]

Bariah L, Zou H, Zhao Q, Mouhouche B, Bader F, Debbah M.Understanding telecom language through large language models.In: Proceedings of IEEEGlobal Communications Conference; 2023 Dec 8–12; Kuala Lumpur, Malaysia. New York City: IEE E; 2023. p. 6542–7.

[199]

Bornea AL, Ayed F, De ADomenico, Piovesan N, Maatouk A. Telco-RAG: navigating the challenges of retrieval-augmented language models for telecommunications., arXiv:2404.15939 (2024)

[200]

Yilma GM, Ayala-Romero JA, Garcia-Saavedra A, Costa-Perez X.TelecomRAG: taming telecom standards with retrieval augmented generation and LLMs.ACM SIGCOMM Comput Commun Rev 2024; 54(3):18-23.