Scholarlink

<p>The growth in the Internet of Things (IoT) demands the development of a secure and efficient authentication framework for safeguarding Machine-to-Machine (M2M) communication. Traditional centralized authentication systems face vulnerabilities, including single points of failure and data tampering, while the heterogeneity of IoT devices creates challenges in energy consumption and complicates anomaly detection. This research proposes MABF-IoT, a Blockchain-Integrated Graph Neural Network (GNN) framework that addresses authentication challenges in M2M communication within IoT environments. MABF-IoT leverages digital device fingerprinting for mutual authentication and employs a GNN-based anomaly detection model with a weighted aggregation function that adapts to evolving threats. The framework incorporates a Proof-of-Stake Consensus mechanism to reduce computational overhead. The comprehensive evaluation demonstrates that MABF-IoT outperforms existing frameworks (SAF, BAS, PuF, BPPS) in terms of recall, accuracy, and precision, with values ranging from 92% to 95%. Quality of Service (QoS) metrics indicate reduced energy consumption, authentication, and identification delay, while throughput and packet delivery ratio increase by 8.5% and 4.9%, respectively, compared to other models. The framework offers a scalable, efficient, and secure authentication solution for resource-constrained IoT environments, thereby enhancing the security of M2M communication.</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

Secure and efficient mutual authentication framework for IoT: a blockchain-integrated graph neural network approach

  • 期刊论文

  • 发表日期：

    2025年11月11日

查看全文
  • Mahendra Kumar Jhariya,
  • Vasudev Dehalwar,
  • Jyoti Bharti,
  • R. K. Pateriya,
  • Yogendra Kumar

摘要

The growth in the Internet of Things (IoT) demands the development of a secure and efficient authentication framework for safeguarding Machine-to-Machine (M2M) communication. Traditional centralized authentication systems face vulnerabilities, including single points of failure and data tampering, while the heterogeneity of IoT devices creates challenges in energy consumption and complicates anomaly detection. This research proposes MABF-IoT, a Blockchain-Integrated Graph Neural Network (GNN) framework that addresses authentication challenges in M2M communication within IoT environments. MABF-IoT leverages digital device fingerprinting for mutual authentication and employs a GNN-based anomaly detection model with a weighted aggregation function that adapts to evolving threats. The framework incorporates a Proof-of-Stake Consensus mechanism to reduce computational overhead. The comprehensive evaluation demonstrates that MABF-IoT outperforms existing frameworks (SAF, BAS, PuF, BPPS) in terms of recall, accuracy, and precision, with values ranging from 92% to 95%. Quality of Service (QoS) metrics indicate reduced energy consumption, authentication, and identification delay, while throughput and packet delivery ratio increase by 8.5% and 4.9%, respectively, compared to other models. The framework offers a scalable, efficient, and secure authentication solution for resource-constrained IoT environments, thereby enhancing the security of M2M communication.