A Hybrid Approach to Defend Against Adversarial Evasion Attacks

This paper discusses the application of Automatic Modulation Classification (AMC) for addressing spectrum congestion in wireless communication and the vulnerability of deep learning models to adversarial evasion attacks. By focusing on Convolutional Neural Networks (CNN), this study proposes a robust design that employs the STFT algorithm and adversarial learning. Experimental results show an enhanced resilience against adversarial attacks, but the paper concludes that further research is needed to balance model performance with adversarial resistance.