Scholarlink

<p>Centralized machine learning frameworks require large-scale data aggregation, which raises significant concerns regarding privacy, security, and data ownership, and limits scalability in resource-constrained Edge-IoT and vehicular networks. Federated Learning (FL) enables decentralized model training without sharing raw data, whereas semi-asynchronous FL improves system efficiency through flexible client participation and periodic synchronization. However, existing methods remain vulnerable to insecure client selection, inconsistent updates, and adversarial manipulation. To address these challenges, in this study, we propose a hybrid blockchain-enabled semi-asynchronous FL framework with optimized client selection. A blockchain is incorporated as a decentralized and tamper-resistant mechanism to validate model updates, ensure data integrity, and maintain transparent and immutable records. Furthermore, an Optimized Deep Neural Network (O-DNN) integrated with an Enhanced Nuclear Reaction Optimization (E-NRO) algorithm is developed to perform intelligent client selection, improving convergence speed and model performance in heterogeneous environments. The framework enhances consistency by securely verifying model updates within the blockchain network. Experimental results demonstrate that the proposed E-NRO-O-DNN model achieves an accuracy of 87.99%, outperforming QBC-ZKPAF (84.67%), HBEoT (87.89%), and PP-ABE (80.21%), while achieving faster convergence and improved robustness. The proposed framework provides a scalable, secure, and privacy-preserving solution for distributed learning in Edge-IoT vehicular systems. The source code and implementation details supporting the findings of this study are publicly available at the following GitHub repository: <a href="https://github.com/mrekhasreephd/Federated-Learning-and-Blockchain-enabled-Edge-IoT-Environment/tree/main">https://github.com/mrekhasreephd/Federated-Learning-and-Blockchain-enabled-Edge-IoT-Environment/tree/main</a>.</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

Optimized client selection for hybrid blockchain-enabled semi-asynchronous federated learning in edge-IoT vehicular networks

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Rekhasree Manthu,
  • Rakesh Tripathi,
  • Tirath Prasad Sahu,
  • Balaprakasa Rao Killi

摘要

Centralized machine learning frameworks require large-scale data aggregation, which raises significant concerns regarding privacy, security, and data ownership, and limits scalability in resource-constrained Edge-IoT and vehicular networks. Federated Learning (FL) enables decentralized model training without sharing raw data, whereas semi-asynchronous FL improves system efficiency through flexible client participation and periodic synchronization. However, existing methods remain vulnerable to insecure client selection, inconsistent updates, and adversarial manipulation. To address these challenges, in this study, we propose a hybrid blockchain-enabled semi-asynchronous FL framework with optimized client selection. A blockchain is incorporated as a decentralized and tamper-resistant mechanism to validate model updates, ensure data integrity, and maintain transparent and immutable records. Furthermore, an Optimized Deep Neural Network (O-DNN) integrated with an Enhanced Nuclear Reaction Optimization (E-NRO) algorithm is developed to perform intelligent client selection, improving convergence speed and model performance in heterogeneous environments. The framework enhances consistency by securely verifying model updates within the blockchain network. Experimental results demonstrate that the proposed E-NRO-O-DNN model achieves an accuracy of 87.99%, outperforming QBC-ZKPAF (84.67%), HBEoT (87.89%), and PP-ABE (80.21%), while achieving faster convergence and improved robustness. The proposed framework provides a scalable, secure, and privacy-preserving solution for distributed learning in Edge-IoT vehicular systems. The source code and implementation details supporting the findings of this study are publicly available at the following GitHub repository: https://github.com/mrekhasreephd/Federated-Learning-and-Blockchain-enabled-Edge-IoT-Environment/tree/main.