Date of Publication :8th May 2024

Abstract:Computer viruses, bad software, and other aggressive acts can damage a computer network. Intrusion monitoring, which is an active defence system, is a key part of network security. Problems with traditional intrusion detection systems include low accuracy, missed threats, a lot of false alarms, and not being able to handle new types of breaches. In order to address these issues, we propose a novel approach for identifying vulnerabilities in cyber-physical systems using deep learning. Our suggested framework highlights the differences between uncontrolled and DL -based methods. We demonstrate the effectiveness of a generative adversarial network in detecting cyber risks in IoT-powered IICs networks. The results show that this system is able to identify various types of threats with higher accuracy, reliability, and efficiency. State-of-the-art DL classifiers successfully detected the most common attacks on NSL-KDD, KDDCup99, and UNSW-NB15 datasets, while also protecting sensitive user and system data during training and testing.

Reference :