Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. Advances in Computer, Signals and Systems
  4. Vol 10, Issue 2, 2026

Interference-Robust Automatic Modulation Recognition under Skewed Deployment Priors

Download as PDF

DOI: 10.23977/acss.2026.100218 | Downloads: 1 | Views: 29

Author(s)

Ruizhi Li 1, Ruisen Luo 1

Affiliation(s)

1 School of Computer Science and Artificial Intelligence, Southwest Minzu University, Chengdu, China

Corresponding Author

Ruizhi Li

ABSTRACT

Deep learning-based automatic modulation recognition (AMR) models are typically evaluated on idealized synthetic benchmarks that assume interference-free conditions and balanced class distributions. However, real-world non-cooperative receivers inevitably suffer from severe structural interference (e.g., localized bursts or full-band jamming) and heavily skewed deployment priors. To investigate these practical mismatches, this paper systematically evaluates a compact convolutional-recurrent backbone on RadioML 2016.10a under both interference and class-imbalance scenarios. We introduce a lightweight sequence gate that adaptively recalibrates temporal feature channels with only 850 additional parameters, combined with IQ-plane augmentation and post-hoc logit calibration. Extensive experiments show that burst-interference robustness is largely a property of this convolutional-recurrent family and that the proposed sequence gate adds up to 4.1 points on top of the plain backbone, whereas domain-specific augmentation dominates resilience against full-band jamming. Furthermore, when trained under a long-tailed prior (imbalance ratio of 50), the standard model exhibits an alarming gap of 22.3% between deployment-weighted accuracy and macro-averaged recall, revealing severe performance degradation on minority classes that conventional metrics conceal. By incorporating the gate, augmentation, and a calibrated decision rule, the macro recall is substantially restored by up to 13.5% without sacrificing overall accuracy. This work provides empirical insights and an efficient baseline for deploying robust AMR models in non-ideal electromagnetic environments.

KEYWORDS

Automatic Modulation Recognition; Interference Robustness; Class Imbalance; Macro-Averaged Recall; Deployment Prior

CITE THIS PAPER

Ruizhi Li, Ruisen Luo. Interference-Robust Automatic Modulation Recognition under Skewed Deployment Priors. Advances in Computer, Signals and Systems (2026). Vol. 10, No. 2, 159-166. DOI: http://dx.doi.org/10.23977/acss.2026.100218.

REFERENCES

[1] T. J. O’Shea, T. Roy, and T. C. Clancy, "Over-the-air deep learning based radio signal classification," IEEE Journal of Selected Topics in Signal Processing, vol. 12, no. 1, pp. 168–179, 2018. DOI: 10.1109/JSTSP.2018.2797022.
[2] G. Gao, X. Hu, B. Li, W. Wang, and F. M. Ghannouchi, "CrossTLNet: A multitask-learning-empowered neural network with temporal convolutional network-long short-term memory for automatic modulation classification," Electronics, vol. 12, no. 22, p. 4668, 2023. DOI: 10.3390/electronics12224668.
[3] W. Zhang, K. Xue, A. Yao, and Y. Sun, "CTRNet: An automatic modulation recognition based on Transformer-CNN neural network," Electronics, vol. 13, no. 17, p. 3408, 2024. DOI: 10.3390/electronics13173408.
[4] Y. Huo, C. Wang, J. Liang, K. Yue, and W. Li, "SMTrans: An efficient automatic modulation recognition network based on the scale-aware modulation transformer," Physical Communication, vol. 74, p. 102966, 2026. DOI: 10.1016/j.phycom.2025.102966.
[5] D. Liu, P. Liu, J. Chen, K. Xiong, B. Jing, and D. Liao, "Automatic modulation recognition based on a lightweight Transformer network under strong interference conditions," Engineering Reports, vol. 7, no. 9, 2025. DOI: 10.1002/eng2.70405.
[6] O. Kwon, H. Lee, M. Debbah, and I. Lee, "Deep learning-based anti-jamming beamforming designs against adversarial jamming attacks," IEEE Transactions on Wireless Communications, vol. 25, pp. 20900–20912, 2026. DOI: 10.1109/TWC.2026.3713089.
[7] Y. Cui, M. Jia, T.-Y. Lin, Y. Song, and S. Belongie, "Class-balanced loss based on effective number of samples," in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2019, pp. 9260–9269. DOI: 10.1109/CVPR.2019.00949.
[8] T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollár, "Focal loss for dense object detection," in Proc. IEEE International Conference on Computer Vision (ICCV), 2017, pp. 2999–3007. DOI: 10.1109/ICCV.2017.324.
[9] M. Buda, A. Maki, and M. A. Mazurowski, "A systematic study of the class imbalance problem in convolutional neural networks," Neural Networks, vol. 106, pp. 249–259, 2018. DOI: 10.1016/j.neunet.2018.07.011.
[10] Y. Zhang, B. Kang, B. Hooi, S. Yan, and J. Feng, "Deep long-tailed learning: A survey," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 9, pp. 10795–10816, 2023. DOI: 10.1109/TPAMI.2023.3268118.
[11] A. K. Menon, S. Jayasumana, A. S. Rawat, H. Jain, A. Veit, and S. Kumar, "Long-tail learning via logit adjustment," in International Conference on Learning Representations (ICLR), 2021. arXiv:2007.07314.
[12] J. Hu, L. Shen, and G. Sun, "Squeeze-and-excitation networks," in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2018, pp. 7132–7141. DOI: 10.1109/CVPR.2018.00745.

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 51411
Visits: 1328628

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.