Scholarlink

<p>With the rapid growth of user data and service demands, edge computing faces increasing challenges in ensuring both efficiency and security. Existing methods often fail to provide trustworthy services while mitigating risks such as malicious attacks and privacy breaches, leading to decreased system reliability. To address this, we propose a dual-metric reputation-based federated learning trustworthy service recommendation method, enhancing edge computing performance and data security. Next, we use a multi-weight subjective logic model to evaluate node opinions in quality and trust dimensions, improving the reliability of trust evaluation. We further integrate an opinion blockchain and design a TaQ consensus mechanism, utilizing dual verification to ensure data integrity, security, and user privacy. Simulation results demonstrate that, compared to traditional training-quality-based methods, our approach eliminates malicious nodes 10 percentage points earlier, significantly improving service reliability. Additionally, the TaQ consensus algorithm enhances federated learning accuracy to 88%, ensuring robust security while optimizing the user experience in edge environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reputation-based federated learning and blockchain for trustworthy service recommendations in edge computing

  • 期刊论文

  • 发表日期：

    2025年9月11日

查看全文
  • Shanchen Pang,
  • Zixuan Fan,
  • Haiyuan Gui,
  • Xiao He,
  • Zhi Lu,
  • Shengzhe Zhao,
  • Baoyun Chen

摘要

With the rapid growth of user data and service demands, edge computing faces increasing challenges in ensuring both efficiency and security. Existing methods often fail to provide trustworthy services while mitigating risks such as malicious attacks and privacy breaches, leading to decreased system reliability. To address this, we propose a dual-metric reputation-based federated learning trustworthy service recommendation method, enhancing edge computing performance and data security. Next, we use a multi-weight subjective logic model to evaluate node opinions in quality and trust dimensions, improving the reliability of trust evaluation. We further integrate an opinion blockchain and design a TaQ consensus mechanism, utilizing dual verification to ensure data integrity, security, and user privacy. Simulation results demonstrate that, compared to traditional training-quality-based methods, our approach eliminates malicious nodes 10 percentage points earlier, significantly improving service reliability. Additionally, the TaQ consensus algorithm enhances federated learning accuracy to 88%, ensuring robust security while optimizing the user experience in edge environments.