Boundary-Aware Margin Shaping for Joint Open-Set Modulation Recognition and Adversarial Robustness
- 23 January 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications Letters
- Vol. 15 (21622337) , 1543-1547
Abstract
Robust automatic modulation recognition (AMR) under open-set conditions and adversarial attacks remains a critical challenge for secure and reliable wireless communication. This letter proposes a Boundary-aware Margin Shaping (BMS) framework that unifies open-set recognition and adversarial robustness through adaptive boundary modeling. BMS introduces a margin-aware training scheme that jointly stabilizes feature representations, calibrates confidence, and smooths decision boundaries under adversarial conditions. For inference, we employ a standard KNN detector to perform out-of-distribution (OOD) separation. Comprehensive experiments on the RML2016.10a dataset demonstrate that BMS consistently achieves superior classification and OOD detection performance across various SNRs and adversarial attacks.Keywords
Funding Information
- National Natural Science Foundation of China (62471486)
- National Natural Science Foundation of China (62501637)
This publication has 10 references indexed in Scilit:
- S3AT: Self-Paced, Self-Distilled, and Self-Finetuned Adversarial Training for Robust Automatic Modulation RecognitionIEEE Transactions on Cognitive Communications and Networking, 2025
- The Importance of Expert Knowledge for Automatic Modulation Open Set RecognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2023
- Adversarial Machine Learning in Wireless Communications Using RF Data: A ReviewIEEE Communications Surveys & Tutorials, 2022
- Proceed From Known to Unknown: Jamming Pattern Recognition Under Open-Set SettingIEEE Wireless Communications Letters, 2022
- Channel-Aware Adversarial Attacks Against Deep Learning-Based Wireless Signal ClassifiersIEEE Transactions on Wireless Communications, 2021
- Adversarial Reciprocal Points Learning for Open Set RecognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2021
- Open-Set Adversarial Defense2020
- Towards Deep Learning Models Resistant to Adversarial AttacksarXiv, 2019
- Adversarial Attacks on Deep-Learning Based Radio Signal ClassificationIEEE Wireless Communications Letters, 2018
- Towards Open Set Deep Networks2016