An Efficient Framework for Real-Time Intrusion Detection in IoT Cyber Security Environments

Abstract

Computer networks are increasingly exposed to viruses, malicious activities,
and other hostile attacks. Intrusion detection serves as a critical component of network
security, acting as an active defence mech1`anism. However, traditional intrusion detection
systems often provide lower accuracy, suffer from high false-positive rates, and fail to
identify emerging intrusion patterns. To address these limitations, this study proposes a deep
learning based framework for detecting cyber security vulnerabilities and breaches in cyberphysical
systems. The framework contrasts unsupervised learning with discriminative deep
learning approaches and incorporates a generative adversarial network to detect cyber threats
within IoT-enabled industrial intelligent control (IIC) networks. The proposed method
demonstrates improved reliability and effectiveness in identifying diverse attack types while
maintaining the confidentiality and integrity of sensitive user and system information. State
of the art deep learning classifiers including RNNs, MLPs, and DNNs integrated into the
framework achieve strong true-negative and detection rates for attack categories such as
Brute Force XXS, Brute Force WEB, DoS Hulk, and DOS_LOIC_HTTP across benchmark
datasets like NSL-KDD, KDDCup99, and UNSW-NB15. Overall, the approach provides a
robust, adaptable, and high-accuracy solution for modern network defence