EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

TAR-DT: A Trusted and Attack-Resilient Mechanism for Distributed DNN Training in Agentic Edge Intelligence

Zhonghui Wu, Yunxiao Ma, Lu Lu, Han Xiao, Chao Liu

Future Internet · 17.08.2026 · 18 (8) · 439

DOI: 10.3390/fi18080439 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Future Internet, online erschienen am 17.08.2026. Verfasst von Zhonghui Wu und 4 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science and Technology Major Project.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As deep neural networks continue to scale and enable emerging applications such as agentic AI systems, training increasingly relies on distributed paradigms across heterogeneous edge devices. However, this shift introduces significant security challenges, particularly model poisoning attacks, which are largely underexplored in model-parallel settings. To address these challenges, we propose a trusted and attack-resilient mechanism for distributed DNN training that supports both data and model parallelism. The mechanism leverages a blockchain-enabled infrastructure to ensure the tamper-resistant and auditable execution of security-critical operations. It introduces a Loss-aware Credit Evaluation mechanism to assess agent reliability based on group-level training dynamics and a Shuffling-based Isolation Mechanism to progressively cluster and isolate malicious agents across training epochs. In addition, Byzantine-tolerant aggregation (BTA) is employed to further mitigate adversarial influence during model aggregation. Extensive experiments demonstrate that the proposed mechanism achieves superior robustness and efficiency compared with state-of-the-art methods under diverse poisoning attack scenarios.

Förderung

  • National Science and Technology Major Project

Zitieren

APA

Wu, Z., Ma, Y., Lu, L., Xiao, H., & Liu, C. (2026). TAR-DT: A Trusted and Attack-Resilient Mechanism for Distributed DNN Training in Agentic Edge Intelligence. Future Internet, 18(8), 439. https://doi.org/10.3390/fi18080439

Vancouver

Wu Z, Ma Y, Lu L, Xiao H, Liu C. TAR-DT: A Trusted and Attack-Resilient Mechanism for Distributed DNN Training in Agentic Edge Intelligence. Future Internet. 2026;18(8):439. doi:10.3390/fi18080439

BibTeX
@article{Wu2026,
  author = {Wu, Zhonghui and Ma, Yunxiao and Lu, Lu and Xiao, Han and Liu, Chao},
  title = {TAR-DT: A Trusted and Attack-Resilient Mechanism for Distributed DNN Training in Agentic Edge Intelligence},
  journal = {Future Internet},
  year = {2026},
  volume = {18},
  number = {8},
  pages = {439},
  doi = {10.3390/fi18080439}
}
Kostenlos starten →