Scholarlink

<p>The emergence of cyber-physical threats focused on smart home settings requires a multi-faceted, secure, and real-time threat detection framework. In this paper, we propose a hybrid deep learning model that combines Graph Attention Networks (GAT) and Temporal Convolutional Networks (TCN) to improve anomaly detection performance. The structure makes use of 6G communication, blockchain, and deep learning in order to ensure that there is secure, real-time anomaly detection in smart homes. The architecture includes multi-modal feature extraction with GAT to model spatial dependencies and TCN to capture temporal attack patterns. A lightweight blockchain layer guarantees transaction immutability and alert validation via automated smart contracts to provide authentication and event traceability, all without trust. Experimental evaluation on a synthetic smart home dataset showed a significant improvement over baseline models (CNN + RNN, Random Forest, and ensemble meta-models), achieving 99.74% accuracy, a 99.74% F1-score, and an inference time of just 38&#xa0;ms/sample.</p><p>According to a thorough evaluation, users benefitted from a reduction in false alarm rates, improved energy efficiency on Edge TPU, and the quickest latencies for real time workloads due to Deep Learning. The greatest benefit derived from CustomChain blockchain technology is the near-finality at ~ 1.2&#xa0;s elapsed time on account of high speed and verifiable responses. These signals confirm that our approach is reliable, scalable, and ready to be integrated into future smart home ecosystems.</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

A secure and scalable home security system using IoT, 6G networks, AI-based hybrid deep learning models, blockchain, and smart contracts

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • Azath Mubarakali,
  • S. Sharavanan,
  • Hashim Elshafie,
  • Suresh Babu Changalasetty

摘要

The emergence of cyber-physical threats focused on smart home settings requires a multi-faceted, secure, and real-time threat detection framework. In this paper, we propose a hybrid deep learning model that combines Graph Attention Networks (GAT) and Temporal Convolutional Networks (TCN) to improve anomaly detection performance. The structure makes use of 6G communication, blockchain, and deep learning in order to ensure that there is secure, real-time anomaly detection in smart homes. The architecture includes multi-modal feature extraction with GAT to model spatial dependencies and TCN to capture temporal attack patterns. A lightweight blockchain layer guarantees transaction immutability and alert validation via automated smart contracts to provide authentication and event traceability, all without trust. Experimental evaluation on a synthetic smart home dataset showed a significant improvement over baseline models (CNN + RNN, Random Forest, and ensemble meta-models), achieving 99.74% accuracy, a 99.74% F1-score, and an inference time of just 38 ms/sample.

According to a thorough evaluation, users benefitted from a reduction in false alarm rates, improved energy efficiency on Edge TPU, and the quickest latencies for real time workloads due to Deep Learning. The greatest benefit derived from CustomChain blockchain technology is the near-finality at ~ 1.2 s elapsed time on account of high speed and verifiable responses. These signals confirm that our approach is reliable, scalable, and ready to be integrated into future smart home ecosystems.