Advancing IoT security: a comprehensive AI-based trust framework for intrusion detection

Over the years, the Internet of Things (IoT) devices have shown rapid proliferation and development in various domains. However, the widespread adoption of smart devices significantly ameliorates the possibility of several security challenges. To address these challenges, this research presents an a...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Peer-to-peer networking and applications 2024-09, Vol.17 (5), p.2737-2757
Hauptverfasser: Kaliappan, Chandra Prabha, Palaniappan, Kanmani, Ananthavadivel, Devipriya, Subramanian, Ushasukhanya
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Over the years, the Internet of Things (IoT) devices have shown rapid proliferation and development in various domains. However, the widespread adoption of smart devices significantly ameliorates the possibility of several security challenges. To address these challenges, this research presents an advanced AI-enhanced trust framework for IoT Intrusion detection to safeguard IoT environments from any potential intrusion attempts. The proposed framework integrates cutting-edge AI techniques for intrusion detection which identifies the anomalies based on the device behavior and responds dynamically to emerging threats. Initially, a robust Intrusion Detection System (IDS) is developed based on an Isolation Forest (IF) algorithm and Autoencoders (AE) to promptly identify anomalies in real-time. Then, behavioral Modeling is performed by employing Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNNs) for precise behavioral understanding of IoT devices. Additionally, the Bayesian Network is used to perform adaptive trust assessment and the Reinforcement Learning based Proximal Policy Optimization (PPO) for providing dynamic responses to the detected anomalies. The proposed framework is practically implemented and evaluated using IoTID20 and N-BaIoT datasets, and compared with baseline intrusion detection methods including, CNN-TSODE, cuLSTMGRU, ELETL-IDS, Fed-Inforce-Fusion, and Conv-LSTM. The results demonstrate that the proposed framework achieves high efficiency and outperformed other baseline methods by obtaining a detection accuracy of 98.25%, recall of 96.8%, and precision of 97.45%. Overall, the proposed AI-Enhanced Trust Framework offers a promising solution by identifying the intrusion endeavors effectively and contributing toward the attainment of secure and trustworthy IoT ecosystems.
ISSN:1936-6442
1936-6450
DOI:10.1007/s12083-024-01684-0