The Space-Air-Ground-Sea Integrated Networks (SAGSIN) significantly enhance global communication by merging satellite, aviation, terrestrial, and marine networks. Crucial to SAGSIN’s functionality and security is spectrum monitoring using deep learning-based Automatic Modulation Classification (AMC), essential for processing and classifying complex modulation signals. However, these AMC models are susceptible to adversarial attacks. Thus, we introduce the Deep Time-Frequency Denoising Transformation (DTFDT) defense method to mitigate the impact of adversarial attacks. The DTFDT method is comprised of a deep denoising module and a transformation module. The denoising module maps signals into the time-frequency domain, amplifying the differences between benign and adversarial examples, aiding in the elimination of adversarial perturbations. Concurrently, the transformation module develops a learnable network, generating example-specific transformation matrices suited for signal data, which diminishes the effectiveness of attacks. Extensive evaluations on two datasets, RML2016.10a and DMRadio09.real, demonstrate the superior defense capabilities of DTFDT against various attacks.

In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning has become a key research component of the Sixth-Generation wireless systems (6G) with numerous regulatory and defense applications. In order to facilitate the application of deep learning in radio signal recognition, in this work, a large-scale real-world radio signal dataset is created based on a special aeronautical monitoring system - Automatic Dependent Surveillance-Broadcast (ADS-B). This paper makes two main contributions. First, an automatic data collection and labeling system is designed to capture over-the-air ADS-B signals in the open and real-world scenario without human participation. Through data cleaning and sorting, a high-quality dataset of ADS-B signals is created for radio signal recognition. Second, we conduct an in-depth study on the performance of deep learning models using the new dataset, as well as comparison with a recognition benchmark using machine learning and deep learning methods. Finally, we conclude this paper with a discussion of open problems in this area.