2611005550
  • Open Access
  • Review

Foundation Models for Wireless Communication: Fundamentals, Challenges, and Future Directions, with Case Studies in Modulation Signal Constellation Diagram Analysis

  • Mohammad Rostami 1,†,   
  • Atik Faysal 1,†,   
  • Reihaneh Gh. Roshan 2,   
  • Nikhil Muralidhar 2,   
  • Yu-Dong Yao 3 and 
  • Huaxia Wang 1,*

Received: 16 March 2026 | Revised: 29 October 2026 | Accepted: 03 November 2026 | Published: 18 November 2026

Abstract

Modern wireless networks must cope with far more devices, and a far wider mix of traffic, than the task-specific models they were built around were designed to handle. Foundation models offer a single large network, pre-trained on unlabeled data with self-supervised objectives, can be reused across many downstream wireless tasks instead of being retrained from scratch for each one. This paper introduces the idea for the wireless setting, with an emphasis on modulation signals generated from their constellation diagrams. We walk through the full pipeline, from what to pre-train on and with which objectives when the inputs are RF signals rather than text or images, to how the pre-trained model is adapted afterward through full fine-tuning, linear probing, or parameter-efficient methods. Two of our own frameworks, NMformer and DenoMAE, serve as concrete case studies of what these choices look like in practice. We give insights about where the models still falls short today, namely its computational cost, the scarcity of labeled wireless data, and the difficulty of adapting in real time. Looking ahead, we argue that lightweight and edge-native architectures, physics-informed synthetic data, and energy-efficient computing are the directions most likely to make foundation models practical in deployed, resource-constrained wireless systems.

References 

  • 1.

    Vishwakarma, A.K.; Chaurasia, S.; Kumar, K.; et al. Internet of things technology, research, and challenges: A survey. Multimed. Tools Appl. 2025, 84, 8455–8490. https://doi.org/10.1007/s11042-024-19278-6.

  • 2.

    Rawat, D.B.; Song, M.; Shetty, S. An Overview of Cognitive Radio Networks. In Dynamic Spectrum Access for Wireless Networks; Springer International Publishing: Cham, Switzerland, 2015; pp. 1–11. https://doi.org/10.1007/978-3-319-15299-8-1.

  • 3.

    Devlin, J.; Chang, M.W.; Lee, K.; et al. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Minneapolis, MN, USA, 2–7 June 2019; Volume 1, pp. 4171–4186. https://doi.org/10.18653/v1/n19-1423.

  • 4.

    Kotaru, M. Adapting Foundation Models for Information Synthesis of Wireless Communication Specifications. arXiv 2023, arXiv:2308.04033v2.

  • 5.

    Adesina, D.; Bassey, J.; Qian, L. Robust Deep Radio Frequency Spectrum Learning for Future Wireless Communications Systems. IEEE Access 2020, 8, 148528–148540. https://doi.org/10.1109/access.2020.3015939.

  • 6.

    Fontaine, J.; Shahid, A.; De Poorter, E. Towards a Wireless Physical-Layer Foundation Model: Challenges and Strategies. arXiv 2024, arXiv:2403.12065v2.

  • 7.

    Faysal, A.; Rostami, M.; Roshan, R.G.; et al. NMformer: A Transformer for Noisy Modulation Classification in Wireless Communication. In Proceedings of the 2024 33rd Wireless and Optical Communications Conference (WOCC), Hsinchu, Taiwan, 25–26 October 2024; pp. 103–108. https://doi.org/10.1109/wocc61718.2024.10786062.

  • 8.

    Faysal, A.; Boushine, T.; Rostami, M.; et al. DenoMAE: A Multimodal Autoencoder for Denoising Modulation Signals. IEEE Commun. Lett. 2025, 29, 1659–1663. https://doi.org/10.1109/lcomm.2025.3570602.

  • 9.

    Wolf, D.; Payer, T.; Lisson, C.S.; et al. Self-supervised pre-training with contrastive and masked autoencoder methods for dealing with small datasets in deep learning for medical imaging. Sci. Rep. 2023, 13, 20260. https://doi.org/10.1038/s41598-023-46433-0.

  • 10.

    Tamizhelakkiya, K.; Gauni, S.; Chandhar, P. Modulation classification analysis of CNN model for wireless communication systems. AIMS Electron. Electr. Eng. 2023, 7, 337–353. https://doi.org/10.3934/electreng.2023018.

  • 11.

    O’Shea, T.; Hoydis, J. An Introduction to Deep Learning for the Physical Layer. IEEE Trans. Cogn. Commun. Netw. 2017, 3, 563–575. https://doi.org/10.1109/tccn.2017.2758370.

  • 12.

    Alikhani, S.; Charan, G.; Alkhateeb, A. Large Wireless Model (LWM): A Foundation Model for Wireless Channels. arXiv 2024, arXiv:2411.08872.

  • 13.

    Ott, J.; Stahlke, M.; Feigl, T.; et al. Simplicity is Key: An Unsupervised Pretraining Approach for Sparse Radio Channels. arXiv 2025, arXiv:2505.13055.

  • 14.

    Palhares, V.; Taner, S.; Studer, C. CSI2Vec: Towards a Universal CSI Feature Representation for Positioning and Channel Charting. arXiv 2025, arXiv:2506.05237.

  • 15.

    Liu, Y.; Liu, M.; Xie, W.; et al. RIS-MAE: A Self-Supervised Modulation Classification Method Based on Raw IQ Signals and Masked Autoencoder. arXiv 2025, arXiv:2508.00274.

  • 16.

    Mashaal, O.; Abou-Zeid, H. IQFM—A Wireless Foundational Model for I/Q Streams in AI-Native 6G. arXiv 2025, arXiv:2506.06718.

  • 17.

    Li, Z.; Li, S.; Yan, X. Time series as images: Vision transformer for irregularly sampled time series. Adv. Neural Inf. Process. Syst. 2023, 36, 49187–49204.

  • 18.

    Faysal, A.; Rostami, M.; Boushine, T.; et al. DenoMAE2.0: Improving Denoising Masked Autoencoders by Classifying Local Patches. arXiv 2025, arXiv:2502.18202.

  • 19.

    Xie, L.; Li, C.; Zhang, X.; et al. TRLS: A Time Series Representation Learning Framework via Spectrogram for Medical Signal Processing. arXiv 2024, arXiv:2401.05431.

  • 20.

    Parker, P.A.; Holan, S.H.; Ravishanker, N. Nonlinear time series classification using bispectrum-based deep convolutional neural networks. Appl. Stoch. Model. Bus. Ind. 2020, 36, 877–890. https://doi.org/10.1002/asmb.2536.

  • 21.

    Jiang, J.; Yu, W.; Li, Y.; et al. A MIMO Wireless Channel Foundation Model via CIR-CSI Consistency. arXiv 2025, arXiv:2502.11965.

  • 22.

    Radford, A.; Kim, J.W.; Hallacy, C.; et al. Learning Transferable Visual Models From Natural Language Supervision. In Proceedings of the International Conference on Machine Learning, Virtual, 18–24 July 2021; pp. 8748–8763.

  • 23.

    Behzad, K.; Zandi, R.; Motamedi, E.; et al. RoboMNIST: A Multimodal Dataset for Multi-Robot Activity Recognition Using WiFi Sensing, Video, and Audio. arXiv 2025, arXiv:2408.16703v2.

  • 24.

    Chong, D.; Wang, H.; Zhou, P.; et al. Masked Spectrogram Prediction For Self-Supervised Audio Pre-Training. arXiv 2022, arXiv:2204.12768.

  • 25.

    He, K.; Chen, X.; Xie, S.; et al. Masked Autoencoders Are Scalable Vision Learners. In Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA, 18–24 June 2022; pp. 15979–15988. https://doi.org/10.1109/cvpr52688.2022.01553.

  • 26.

    Guler, B.; Geraci, G.; Jafarkhani, H. A Multi-Task Foundation Model for Wireless Channel Representation Using Contrastive and Masked Autoencoder Learning. arXiv 2025, arXiv:2505.09160v2.

  • 27.

    Mu, S.; Chen, S.; Zu, Y.; et al. Contrastive Learning with Feature-Level Augmentation for Wireless Signal Representation. Electronics 2025, 14, 2728. https://doi.org/10.3390/electronics14132728.

  • 28.

    Tiras, F.E.; Altinoluk, H.S. CrossRF: A Domain-Invariant Deep Learning Approach for RF Fingerprinting. arXiv 2025, arXiv:2505.18200.

  • 29.

    Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; et al. Generative Adversarial Networks. arXiv 2014, arXiv:1406.2661.

  • 30.

    Kingma, D.P.; Welling, M. Auto-Encoding Variational Bayes. arXiv 2013, arXiv:1312.6114.

  • 31.

    Böck, B.; Oeldemann, A.; Mayer, T.; et al. Physics-Informed Generative Modeling of Wireless Channels. arXiv 2025, arXiv:2502.10137.

  • 32.

    Gupta, C.; Das, R.K.; Barik, R.K.; et al. GANCE: Generative Adversarial Network Assisted Channel Estimation for Unmanned Aerial Vehicles Empowered 5G and beyond Wireless Networks. IEEE Access 2025, 13, 198–213. https://doi.org/10.1109/access.2024.3522847.

  • 33.

    Hu, E.J.; Shen, Y.; Wallis, P.; et al. LoRA: Low-rank adaptation of large language models. In Proceedings of the International Conference on Learning Representations, Virtual, 25–29 April 2022; Volume 1; p. 3.

  • 34.

    Zhang, H.; Zhou, F.; Wu, Q.; et al. FSOS-AMC: Few-Shot Open-Set Learning for Automatic Modulation Classification. In Proceedings of the 2024 16th International Conference on Wireless Communications and Signal Processing (WCSP), Hefei, China, 24–26 October 2024; pp. 746–751. https://doi.org/10.1109/wcsp62071.2024.10826902.

  • 35.

    Wang, Y.; Gao, Z.; Zheng, D.; et al. Transformer-Empowered 6G Intelligent Networks: From Massive MIMO Processing to Semantic Communication. IEEE Wirel. Commun. 2023, 30, 127–135. https://doi.org/10.1109/mwc.008.2200157.

  • 36.

    Stratton, J. An Introduction to Microsoft Copilot. In Copilot for Microsoft 365; Apress: Berkeley, CA, USA, 2024; pp. 19–35. https://doi.org/10.1007/979-8-8688-0447-2-2.

  • 37.

    Aboulfotouh, A.; Eshaghbeigi, A.; Abou-Zeid, H. Building 6G Radio Foundation Models with Transformer Architectures. arXiv 2024, arXiv:2411.09996v1.

  • 38.

    Chin, W.L. On the Noise Uncertainty for the Energy Detection of OFDM Signals. IEEE Trans. Veh. Technol. 2019, 68, 7593–7602. https://doi.org/10.1109/tvt.2019.2920142.

  • 39.

    Herrera-Loera, C.; Del-Valle-Soto, C.; Valdivia, L.J.; et al. A Systematic Review and Energy-Centric Taxonomy of Jamming Attacks and Countermeasures in Wireless Sensor Networks. Sensors 2026, 26, 579. https://doi.org/10.3390/s26020579.

  • 40.

    Raffel, C.; Shazeer, N.; Roberts, A.; et al. Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer. J. Mach. Learn. Res. 2020, 21, 1–67.

Share this article:
How to Cite
Rostami, M.; Faysal, A.; Roshan, R. Gh.; Muralidhar, N.; Yao, Y.-D.; Wang, H. Foundation Models for Wireless Communication: Fundamentals, Challenges, and Future Directions, with Case Studies in Modulation Signal Constellation Diagram Analysis. AI Engineering 2026, 2 (2), 23. https://doi.org/10.53941/aieng.2026.100023.
RIS
BibTex
Copyright & License
article copyright Image
Copyright (c) 2026 by the authors.
Article Metrics
63
Article Views
0
Citations