Segmentation and reconstruction of wireless signals under interference using the FoT-UNet neural network
DOI:
https://doi.org/10.54939/1859-1043.j.mst.110.2026.45-54Keywords:
RF jamming; Spectrum segmentation; Unet; Signal reconstruction; Time–frequency analysis; Deep learning.Abstract
The increasing prevalence of intentional radio frequency (RF) jamming poses significant challenges to spectrum monitoring and signal reconstruction in complex interference environments. Accordingly, this paper proposes FoT-UNet, a two-stage deep learning framework for time–frequency spectrum segmentation and reconstruction of jammed wireless signals. In the first stage, FoT-UNet-Seg performs RF interference segmentation on 256×256×1 short-time Fourier transform spectrograms, employing the Focal-Tversky loss to enhance sensitivity to small jamming regions. In the second stage, FoT-UNet-Recon, leverages the cleaned spectral features to reconstruct the original spectrum, optimized using a mean absolute error (MAE) objective to minimize amplitude distortion. Experiments on a dataset of 30,000 samples demonstrate that FoT-UNet-Seg achieves IoU = 0.9933, F1 = 0.9966, and Precision = 0.9971, with an average processing time of 13.9 ms per image and 10.9 million parameters. The reconstruction network attains MAE = 0.0801, MSE = 0.0125, PSNR = 20.04 dB, and SSIM = 0.7357, indicating accurate recovery of spectral patterns even under broadband burst interference. In comparison, our approach outperforms several other works in the same task.
References
[1]. Malik, T. A. Alghamdi, M. A. Al-Harthi, and H. Alshahrani. “Radio Frequency Interference, Its Mitigation and Its Propagation in GNSS”. Electronics, vol. 14, no. 12, p. 2483, (2025). DOI: 10.3390/electronics14122483.
[2]. H. Pirayesh and H. Zeng. “Jamming Attacks and Anti-Jamming Strategies in Wireless Networks: A Comprehensive Survey”. IEEE Communications Surveys & Tutorials, vol. 24, no. 2, pp. 767–809, (2022). DOI: 10.1109/COMST.2022.3155640.
[3]. A. S. Aldhaeebi, M. A. Altahrawi, and H. M. Alghamdi. “Mitigation of Chirp Jamming with Three Types of Neural Networks”. Proc. Int. Conf. on Computer Engineering and Systems, (2021).
[4]. S. H. Kiani, N. S. Khan, and M. Qureshi. “Taxonomy of Physical Layer Jamming Techniques”. TechRxiv preprint, (2023).
[5]. H. Calatrava, E. Fernandez, and M. Martin. “Robust Interference Mitigation in GNSS Snapshot Receivers”. Navigation: Journal of the Institute of Navigation, vol. 72, no. 2, pp. 201–216, (2025). DOI: 10.33012/navi.699.
[6]. Z. Guan, J. Zhang, and H. Wang. “Retrospection of Nonlinear Adaptive Algorithm-Based Signal Enhancement”. Sensors, vol. 22, no. 5, p. 1885, (2022). DOI: 10.3390/s22051885.
[7]. C.-L. Liao and J.-J. Ding. “Enhanced Adaptive Wiener Filtering for Frequency-Varying Noise”. Applied Sciences, vol. 9, no. 2, p. 47, (2025). DOI: 10.3390/app9020047.
[8]. J. Akeret, C. Chang, A. Lucchi, and A. Refregier. “Radio Frequency Interference Mitigation Using Deep Convolutional Neural Networks”. Astronomy and Computing, vol. 18, pp. 35–39, (2017). DOI: 10.1016/j.ascom.2016.10.001.
[9]. A. Vafaei Sadr, A. Sadr, M. Heydari, and A. Refregier. “Deep Learning Improves Identification of Radio Frequency Interference”. arXiv preprint, arXiv:2005.08992, (2020).
[10]. S. Siddharth. “Real-World Wireless Communication Dataset”. Kaggle Open Data Platform, [Online] Available: https://www.kaggle.com/datasets/siddharthsahoo/real-world-wireless-communication-dataset, (2023).
[11]. N. Abraham and N. Mefraz Khan. “A Novel Focal Tversky Loss Function with Improved Attention U-Net for Lesion Segmentation”. Proc. IEEE Int. Symp. on Biomedical Imaging (ISBI), pp. 683–687, (2019). DOI: 10.1109/ISBI.2019.8759329.
[12]. R. Q. Yan, C. Dai, W. Liu, J. X. Li, S. Y. Chen, X. C. Yu, S. F. Zuo, and X. L. Chen. “Radio-frequency interference detection based on the AC-U-Net network model with atrous convolution”. Research in Astronomy and Astrophysics, vol. 21, no. 5, pp. 1–13, (2021). DOI: 10.1088/1674-4527/21/5/119.
[13]. Z. Wang, J. Zhao, and X. Li. “Smart Interference Segmentation Network for Space-Borne SAR Images”. Remote Sensing, vol. 15, no. 23, article 5462, (2023). DOI: 10.3390/rs15235462.
[14]. F. Zheng, Z. Zhang, and D. Zhang. “Lightweight deep neural network for radio frequency interference detection and segmentation in synthetic aperture radar”. Scientific Reports, vol. 14, article 20685, (2024). DOI: 10.1038/s41598-024-71775-8.
[15]. F. Gu, L. Hao, B. Liang, S. Feng, S. Wei, W. Dai, Y. Xu, Z. Li, and Y. Dao. “Radio Frequency Interference Detection Using Efficient Multi-Scale Convolutional Attention UNet”. arXiv preprint arXiv:2404.00277, (2024), available: https://arxiv.org/abs/2404.00277.
