Scholarlink

<p>Ensuring data reliability in Internet of Things (IoT) environments is critical for enabling trustworthy AI systems. This paper introduces RFoT, a layered framework that integrates Blockchain, Smart Contracts, and Federated Learning (FL) to guarantee data confidentiality, integrity, availability, and authenticity across the IoT data lifecycle. Confidentiality is addressed via privacy-preserving automation with Fernet encryption and Smart Contracts, while traceability and immutability are ensured through a dual-layer Blockchain architecture. Experimental results, based on a thermal comfort classification task using FL, demonstrate that conventional IoT setups propagate corrupted data, impairing model accuracy. In contrast, RFoT successfully blocks compromised data, preserving classification performance. These findings validate RFoT as a reliable data source for edge-based learning systems.</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

Enhancing IoT data reliability with blockchain and federated learning in edge environments

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月25日

查看全文
  • Eliabe N. Silva,
  • George P. Pinto,
  • Claudio Jr N. Silva,
  • Maycon L. M. Peixoto,
  • Gustavo B. Figueiredo,
  • Bruno P. Santos,
  • Cassio V. S. Prazeres

摘要

Ensuring data reliability in Internet of Things (IoT) environments is critical for enabling trustworthy AI systems. This paper introduces RFoT, a layered framework that integrates Blockchain, Smart Contracts, and Federated Learning (FL) to guarantee data confidentiality, integrity, availability, and authenticity across the IoT data lifecycle. Confidentiality is addressed via privacy-preserving automation with Fernet encryption and Smart Contracts, while traceability and immutability are ensured through a dual-layer Blockchain architecture. Experimental results, based on a thermal comfort classification task using FL, demonstrate that conventional IoT setups propagate corrupted data, impairing model accuracy. In contrast, RFoT successfully blocks compromised data, preserving classification performance. These findings validate RFoT as a reliable data source for edge-based learning systems.