RETRACTED ARTICLE: A novel deep learning-based approach for detecting attacks in social IoT.

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• The Social Internet of Things (IoT) integrates IoT devices with social platforms, enabling device-to-device interactions.
• Customer adoption is hindered by privacy concerns and the risk of public data exposure.
• Ensuring trustworthy connections and data privacy is crucial for widespread IoT implementation.

Purpose of the Study:

• To address the limitations of existing methods in identifying fraudulent nodes and distinguishing attack types in IoT networks.
• To propose a novel approach for enhancing security and privacy in the Social Internet of Things.
• To develop a robust system for detecting and isolating malicious nodes.

Main Methods:

• A novel Multi-hop Convolutional Neural Network with an attention mechanism (MH-CNN-AM) is proposed.
• The model is designed to identify attacks from hostile nodes and separate them from the network.
• Performance is evaluated using metrics such as accuracy, precision, recall, F1-score, and Mean Absolute Error (MAE).

Main Results:

• The proposed MH-CNN-AM model demonstrates effectiveness in identifying and isolating malicious nodes.
• Comparative analysis shows superior performance against existing methodologies in detecting various forms of attacks.
• The study validates the model's capability in enhancing network security and trustworthiness.

Conclusions:

• The MH-CNN-AM model offers a significant advancement in securing Social IoT environments.
• Effective identification and isolation of malicious nodes are critical for user trust and data protection.
• Future research should focus on further refining detection techniques and ensuring comprehensive privacy in IoT ecosystems.