Scholarlink

<p>Internet of Things (IoT) devices are more and more connected, giving rise to smart homes that rely on these systems for energy management, security monitoring, and comfort. At the same time, when different heterogeneous devices are running, they can cause automation conflicts, unsafe interactions, and inconsistencies in policies, which are hard to manage in dynamic environments with conventional rule-based systems. Artificial intelligence (AI) and deep learning (DL) models are increasingly being employed by researchers for this purpose. Many researchers are employing AI-based models and DL based solutions for this purpose. To tackle these challenges, this paper introduces an Adaptive Spatio-Temporal Graph Intelligence Framework (AST-GIF) for intelligent conflict-aware analysis in a smart home IoT system. The proposed framework models the evolving interactions over time by defining the IoT devices, automation policies, and environmental contexts as nodes in a dynamic spatio-temporal graph. A Spatio-Temporal Graph Transformer Encoder (ST-GTE) is used to learn temporal interaction patterns and identify anomalous cross-device behaviors, and a Hierarchical Conflict Semantic Classifier (HCSC) based on a Graph Attention Network (GAT) and a Temporal Convolutional Network (TCN) is used for semantic analysis of conflict-related interaction patterns. Furthermore, the Aquila Optimization Algorithm (AOA) is embedded to optimize the model parameters and enhance the convergence performance. The framework is tested with benchmark datasets such as CASAS, IoT-23, and UCI Appliances Energy, which describe various aspects of the smart home IoT behaviour. Experimental results show that the proposed model has an accuracy of 97.2%, a precision of 96.8%, a recall of 96.1%, and an F1 of 96.4%, which are approximately 1.9% and 2.0% higher than the accuracy and F1 of the existing transformer-based approaches. Moreover, the framework has a good AUC-ROC value of 0.972 that reflects high discriminative power in imbalanced IoT environments. The results show the effectiveness of the proposed AST-GIF framework for real-time conflict-aware analysis and intelligent management in dynamic smart home IoT 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

Adaptive Spatio-temporal Graph Learning for Intelligent Conflict Detection in Secure Smart Home IoT Systems

  • 期刊论文

  • 发表日期：

    2026年7月25日

查看全文
  • Rami Baazeem

摘要

Internet of Things (IoT) devices are more and more connected, giving rise to smart homes that rely on these systems for energy management, security monitoring, and comfort. At the same time, when different heterogeneous devices are running, they can cause automation conflicts, unsafe interactions, and inconsistencies in policies, which are hard to manage in dynamic environments with conventional rule-based systems. Artificial intelligence (AI) and deep learning (DL) models are increasingly being employed by researchers for this purpose. Many researchers are employing AI-based models and DL based solutions for this purpose. To tackle these challenges, this paper introduces an Adaptive Spatio-Temporal Graph Intelligence Framework (AST-GIF) for intelligent conflict-aware analysis in a smart home IoT system. The proposed framework models the evolving interactions over time by defining the IoT devices, automation policies, and environmental contexts as nodes in a dynamic spatio-temporal graph. A Spatio-Temporal Graph Transformer Encoder (ST-GTE) is used to learn temporal interaction patterns and identify anomalous cross-device behaviors, and a Hierarchical Conflict Semantic Classifier (HCSC) based on a Graph Attention Network (GAT) and a Temporal Convolutional Network (TCN) is used for semantic analysis of conflict-related interaction patterns. Furthermore, the Aquila Optimization Algorithm (AOA) is embedded to optimize the model parameters and enhance the convergence performance. The framework is tested with benchmark datasets such as CASAS, IoT-23, and UCI Appliances Energy, which describe various aspects of the smart home IoT behaviour. Experimental results show that the proposed model has an accuracy of 97.2%, a precision of 96.8%, a recall of 96.1%, and an F1 of 96.4%, which are approximately 1.9% and 2.0% higher than the accuracy and F1 of the existing transformer-based approaches. Moreover, the framework has a good AUC-ROC value of 0.972 that reflects high discriminative power in imbalanced IoT environments. The results show the effectiveness of the proposed AST-GIF framework for real-time conflict-aware analysis and intelligent management in dynamic smart home IoT systems.