Journal of AI-Driven Communication Engineering

Special Issue: Artificial Intelligence Across the Communication Stack: Engineering, Human Interaction, and Governance in the 6G Era

Vol. 2 (2026)

Score-Based Diffusion Models for Wireless Channel Synthesis and Data Augmentation in Low-Pilot MIMO Systems

DOI:
https://doi.org/10.31875/2979-1081.2026.02.06
Submitted
July 4, 2026
Published
2026-07-04

Abstract

High-quality channel estimation in massive multiple-input multiple-output (MIMO) and reconfigurable intelligent surface (RIS)-assisted systems depends critically on the availability of accurate channel state information (CSI), but dense pilot acquisition competes directly with spectral efficiency and latency targets in fifth- and sixth-generation (5G/6G) networks. Score-based diffusion models can synthesize physically plausible wireless channel samples and can also be conditioned on sparse pilot observations to support low-pilot channel estimation. This paper reviews diffusion-based channel synthesis from both a modelling and a receiver-deployment perspective. In addition to denoising diffusion probabilistic models (DDPM), score matching with Langevin dynamics (SMLD), and consistency models, the revised discussion explicitly addresses practical constraints: integration with DM-RS/CSI-RS based receiver pipelines, offline versus online deployment modes, latency budgets imposed by channel coherence time, and hardware limitations at the base-station edge and user equipment. Synthesis fidelity is discussed using normalized mean square error (NMSE), power delay profile (PDP) preservation, spatial correlation preservation, and downstream channel-estimation performance across three types of evidence: stochastic simulation with QuaDRiGa Urban Macro, ray-tracing simulation with DeepMIMO at 28 GHz, and over-the-air measurements from the DICHASUS testbed. The paper further clarifies that results obtained on synthetic, ray-tracing, and real-world datasets are complementary but not directly interchangeable. Computational complexity, inference latency, and robustness risks are therefore treated as central deployment criteria rather than secondary implementation details.

References

  1. Björnson E, Hoydis J, Sanguinetti L: Massive MIMO networks: Spectral, energy, and hardware efficiency. Foundations and Trends in Signal Processing 2017, 11(3-4): 154-655. https://doi.org/10.1561/2000000093
  2. Björnson E, Sanguinetti L, Wymeersch H, Hoydis J, Marzetta TL: Massive MIMO is a reality-what is next? Digital Signal Processing 2019, 94: 3-20. https://doi.org/10.1016/j.dsp.2019.06.007
  3. Kay SM: Fundamentals of Statistical Signal Processing: Estimation Theory. Prentice Hall; 1993.
  4. Rao BD, Engan K, Cotter SF, Palmer J, Kreutz-Delgado K: Subset selection in noise based on diversity measure minimization. IEEE Transactions on Signal Processing 2003, 51(3): 760-770. https://doi.org/10.1109/TSP.2002.808076
  5. Ye H, Li GY, Juang BH: Power of deep learning for channel estimation and signal detection in OFDM systems. IEEE Wireless Communications Letters 2018, 7(1): 114-117. https://doi.org/10.1109/LWC.2017.2757490
  6. Ho J, Jain A, Abbeel P: Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems 2020, 33: 6840-6851.
  7. Alkhateeb A, Elbir AM, Yuan J, et al.: The next generation of deep learning for 6G: From AI-enabled to AI-native. IEEE Signal Processing Magazine 2023, 40(5): 82-96.
  8. Hyvärinen A, Dayan P: Estimation of non-normalized statistical models by score matching. Journal of Machine Learning Research 2005, 6: 695-709.
  9. Song Y, Sohl-Dickstein J, Kingma DP, Kumar A, Ermon S, Poole B: Score-based generative modeling through stochastic differential equations. ICLR 2021.
  10. Ho J, Jain A, Abbeel P: Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems 2020, 33: 6840-6851.
  11. Feuerriegel S, Hartmann J, Janiesch C, Zschech P: Diffusion models. Business & Information Systems Engineering 2024, 66(2): 111-124. https://doi.org/10.1007/s12599-023-00834-7
  12. Song J, Meng C, Ermon S: Denoising diffusion implicit models. ICLR 2021.
  13. Song Y, Dhariwal P, Chen M, Sutskever I: Consistency models. ICML 2023.
  14. Baur S, Boehm S, Stephan M, Schotten HD: Consistency model-based rapid CSI feedback for massive MIMO systems. IEEE Wireless Communications Letters 2024, 13(10): 2718-2722.
  15. Alkhateeb A, Lim S, Samimi MK: DeepMIMO: A generic deep learning dataset for millimeter wave and massive MIMO applications. In: ITA 2019.
  16. Tian S, Yang W, Caire G: Score-based generative models for wireless channel synthesis. In: IEEE ICC 2023.
  17. Kadkhodaie Z, Simoncelli EP: Solving linear inverse problems using the score of implicit diffusion models. IEEE Transactions on Information Theory 2024, 70(9): 6591-6610.
  18. Jalden N, Zetterberg P, Ottersten B, Hong L, Thomä R: Inter- and intrasite correlations of large-scale parameters from macrocellular measurements at 1800 MHz. EURASIP Journal on Wireless Communications and Networking 2007. https://doi.org/10.1155/2007/25757
  19. Fesl B, Baur S, Joham M, Utschick W: Diffusion-based channel estimation using pilot observations. In: IEEE ICASSP 2024.
  20. Wu Q, Zhang R: Towards smart and reconfigurable environment: Intelligent reflecting surface aided wireless network. IEEE Communications Magazine 2020, 58(1): 106-112. https://doi.org/10.1109/MCOM.001.1900107
  21. Chen H, Bai J, Xu W, et al.: Hierarchical diffusion models for RIS-assisted MIMO channel estimation. IEEE Communications Letters 2025, 29(3): 601-605.
  22. Alkhateeb A: DeepMIMO: A generic deep learning dataset for millimeter-wave and sub-6GHz channels. arXiv:1902.06435; 2019.
  23. Jaeckel S, Raschkowski L, Borner K, Thiele L: QuaDRiGa: A 3GPP-compliant framework for multi-cell channel modeling. In: IEEE ICC 2016.
  24. Euchner F, Gauger M, ten Brink S: DICHASUS: A large-scale MIMO channel sounding dataset. In: IEEE DySPAN 2022.
  25. Baur S, Stephan M, Boehm S, Fesl B, Schotten HD: Data augmentation for deep learning-based channel estimation using a diffusion model. In: IEEE VTC 2024.
  26. Goldblum M, Tsipras D, Xie C, et al.: Dataset security for machine learning. IEEE Transactions on Pattern Analysis and Machine Intelligence 2023, 45(2): 1563-1580. https://doi.org/10.1109/TPAMI.2022.3162397
  27. He Z, Zhang S, Li N, et al.: Poisoning attacks on diffusion-based wireless channel models. In: IEEE GLOBECOM 2024.
  28. Shi Y, Davaslioglu K, Sagduyu YE: Generative adversarial network in the air: Deep adversarial learning for wireless signal spoofing. IEEE Transactions on Cognitive Communications and Networking 2021, 7(1): 294-303. https://doi.org/10.1109/TCCN.2020.3010330
  29. Madry A, Makelov A, Schmidt L, Tsipras D, Vladu A: Towards deep learning models resistant to adversarial attacks. ICLR 2018.
  30. Tramer F, Zhang F, Juels A, Reiter MK, Ristenpart T: Stealing machine learning models via prediction APIs. USENIX Security 2016.
  31. Fernandez V, Couairon G, Douze M, Tamaazousti M, Furon T: The stable signature: Rooting watermarks in latent diffusion models. ICCV 2023. https://doi.org/10.1109/ICCV51070.2023.02053
  32. Rappaport TS: Wireless Communications: Principles and Practice, 2nd ed. Prentice Hall; 2001.
  33. 3GPP TS 38.211: NR; Physical Channels and Modulation. 3GPP; Release 18; 2024.
  34. Yuan W, Liu C, Li C, et al.: Multi-configuration conditioned diffusion for channel estimation in 5G NR. IEEE Transactions on Wireless Communications 2025 [early access].
  35. Kim S, Moon I, Lee D, et al.: Integer-only quantization for efficient DNN-based channel estimation. IEEE Wireless Communications Letters 2023, 12(3): 500-504.
  36. Hoydis J, Aoudia FA, Valcarce A, Viswanathan H: Toward a 6G AI-native air interface. IEEE Communications Magazine 2021, 59(5): 76-81. https://doi.org/10.1109/MCOM.001.2001187
  37. Zhang X, Jian M, Liu Z, et al.: Physics-informed diffusion models for wireless channel synthesis. In: IEEE ICASSP 2025.
  38. Sun H, Shi Q, Xu X, et al.: Learning to optimize: Training deep neural networks for interference management. IEEE Transactions on Signal Processing 2018, 66(20): 5438-5453. https://doi.org/10.1109/TSP.2018.2866382
  39. Yang Y, Gao F, Zhong Z, Ai B, Alkhateeb A: Deep transfer learning-based downlink channel prediction. IEEE Transactions on Communications 2020, 68(12): 7485-7497. https://doi.org/10.1109/TCOMM.2020.3019077
  40. Fesl B, Koller M, Joham M, Utschick W: Online channel estimation using diffusion models with temporal prediction. In: IEEE ICASSP 2025.