Covert Communications: A Generative Spectrum Map Paradigm

Spectrum map is emerging as an effective tool to characterize and visualize the radio environment for covert communications. However, current approaches for spectrum map acquisition still face significant challenges in terms of accuracy and adaptability in highly complex and rapidly changing environments. Generative artificial intelligence (GAI), with its powerful capability in data learning and generation, can provide a promising solution. In this article, we demonstrate a comprehensive study of GAI-based spectrum maps for covert communications. Specifically, we first classify spectrum maps, detail their construction methods from conventional to emerging approaches, and present a comparative analysis. Then, we review how GAI generates spectrum maps, illustrate their applications in covert communications, and highlight research gaps. To this end, we propose a novel hybrid framework with diffusion models to enable the rapid generation of fine-grained spectrum maps. We further present a case study to demonstrate its effectiveness in enhancing the covertness. Finally, three future directions for applying GAI-based spectrum maps to covert communications are outlined to further promote the research progress.