EUVIMEDEUVIMED entdecken
Crossref

Energy-Efficient Resource Allocation in Generative AI-Aided Secure Semantic Mobile Networks

Jie Zheng, Baoxia Du, Hongyang Du, Jiawen Kang, Dusit Niyato, Haijun Zhang

IEEE Transactions on Mobile Computing · 2024-12 · 23 (12) · 11422-11435

DOI: 10.1109/tmc.2024.3396860 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2024-12. Verfasst von Jie Zheng und 5 weiteren Autor:innen. Die Arbeit wurde bisher 34-mal zitiert (Crossref) und nennt 78 Literaturangaben. Gefördert von: National Natural Science Foundation of China, National Research Foundation Singapore, Info-communications Media Development Authority.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The integration of semantic communication with Internet of Things (IoT) technologies has advanced the development of Semantic IoT (SIoT), with edge mobile networks playing an increasingly vital role. This paper presents a framework for SIoT-based image retrieval services, focusing on the application in automotive market analysis. Here, semantic information in the form of textual representations is transmitted to users, such as automotive companies, and stored as knowledge graphs, instead of raw imagery. This approach reduces the amount of data transmitted, thereby lowering communication resource usage, and ensures user privacy. We explore potential adversarial attacks that could disrupt image transmission in SIoT and propose a defense mechanism utilizing Generative Artificial Intelligence (GAI), specifically the Generative Diffusion Models (GDMs). Unlike methods that necessitate adversarial training with specifically crafted adversarial example samples, GDMs adopt a strategy of adding and removing noise to negate adversarial perturbations embedded in images, offering a more universally applicable defense strategy. The GDM-based defense aims to protect image transmission in SIoT. Furthermore, considering mobile devices’ resource constraints, we employ GDM to devise resource allocation strategies, optimizing energy use and balancing between image transmission and defense-related energy consumption. Our numerical analysis reveals the efficacy of GDM in reducing energy consumption during adversarial attacks. For instance, in a scenario, GDM-based defense lowers energy consumption by 5.64%, decreasing the number of image retransmissions from 18 to 6, thus underscoring GDM's role in bolstering network security.

Förderung

  • National Natural Science Foundation of China
  • National Research Foundation Singapore
  • Info-communications Media Development Authority
  • Future Communications Research and Development Programme
  • Defence Science Organisation (DSO) National Laboratories
  • AI Singapore Programme AISG
  • Singapore Ministry of Education Tier 1

Zitieren

APA

Zheng, J., Du, B., Du, H., Kang, J., Niyato, D., & Zhang, H. (2024). Energy-Efficient Resource Allocation in Generative AI-Aided Secure Semantic Mobile Networks. IEEE Transactions on Mobile Computing, 23(12), 11422-11435. https://doi.org/10.1109/tmc.2024.3396860

Vancouver

Zheng J, Du B, Du H, Kang J, Niyato D, Zhang H. Energy-Efficient Resource Allocation in Generative AI-Aided Secure Semantic Mobile Networks. IEEE Transactions on Mobile Computing. 2024;23(12):11422-11435. doi:10.1109/tmc.2024.3396860

BibTeX
@article{Zheng2024,
  author = {Zheng, Jie and Du, Baoxia and Du, Hongyang and Kang, Jiawen and Niyato, Dusit and Zhang, Haijun},
  title = {Energy-Efficient Resource Allocation in Generative AI-Aided Secure Semantic Mobile Networks},
  journal = {IEEE Transactions on Mobile Computing},
  year = {2024},
  volume = {23},
  number = {12},
  pages = {11422--11435},
  doi = {10.1109/tmc.2024.3396860}
}
Kostenlos starten →