Hybrid C4.5 Decision Tree Model for Privacy-Preserving Cloud Intrusion Detection Systems
Year: 2026
Type: conference-paper
Abstract: This study introduces an extensive framework for a cooperative privacy-preserving cloud intrusion detection system designed to enhance the security of Internet of Things ecosystems. The study presents a hierarchical architecture that integrates edge, fog and cloud layers to facilitate scalable performance, employing lightweight detection methods underpinned by blockchain-based trust management to guarantee transparent and dependable communication across diverse devices. During the methods phase, the system implements preprocessing and feature selection before classification, utilising various techniques such as the C4.5 decision tree, Support Vector Machine, traditional decision tree, Convolutional Neural Network, ConvNet, and Fully Convolutional Network, while maintaining privacy in distributed cloud analytics and collaborative decision-making. The experimental assessment demonstrates that integrating preprocessing with the Synthetic Minority Oversampling Technique enhances threat detection, while the C4.5 decision tree attains approximately 96.33 percent accuracy on the CES CICIDS2018 dataset using a diminished set of flow-based features, enabling real-time functionality under high-speed network conditions. The final remarks highlight that the incorporation of blockchain technology, tiered computing resources, and efficient classification methods facilitates a more reliable, scalable, and adaptable intrusion detection system, offering a feasible pathway for robust security in dynamic IoT environments.
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204754977