Leveraging Generative AI for Security Defense of Mobile Edge-Enabled Digital Twins: A Survey
Yaoqi Yang, Geng Sun, Hongyang Du, Changyuan Zhao, Zehui Xiong, Dusit Niyato, Abbas Jamalipour, Zhu Han, Khaled B. Letaief · IEEE Transactions on Cognitive Communications and Networking · 2026
Addressing rising security threats that hinder the deployment of mobile edge-enabled digital twins (MDTs), this paper investigates the transformative potential of generative artificial intelligence (GAI) for enhancing MDT security frameworks. We propose a comprehensive five-layer MDT architecture and analyze various malicious security threats across physical and digital twin spaces. Focusing on GAI’s unique capabilities in data quality and diversity, we explore the application of variational autoencoders, generative adversarial networks, generated diffusion models, and transformer-based models within different layers of the MDT architecture. We also provide a summary of performance metrics, evaluation methodologies, and implementation considerations, concluding with key future research directions aimed at realizing real-time and complex attack defense strategies. Our findings show that GAI can significantly improve security measures for MDT systems, effectively addressing critical challenges in this rapidly evolving field.