Intrusion Detection System – IIDS

Abstract

The swift evolution of the Internet of Things (IoT) is generating significant cybersecurity challenges, as securing interlinked devices is of the utmost importance. The suggested smart system offers a method that could contribute to improved detection of cyberattacks aimed at IoT devices. This system leverages Conditional Generative Adversarial Networks (CGANs) to produce synthetic attack data, and then the LightGBM algorithm extracts patterns of malicious activity. The smart system accounts for a data preparation stage for IoT devices, and LightGBM's important feature selection step optimizes performance. The smart system is capable of differentiating between multiple attack types that include denial of service (DoS) attacks, address resolution protocol (ARP) spoofing, and data leakage. Using a gradient boosting framework, the smart system provides a reasonable trade-off between computational efficiency and accurate detection performance; furthermore, the model improves upon existing intrusion detection systems using the RT-IoT2022 dataset, attaining 87% accuracy while detecting attacks against the RT-IoT2022 dataset that contained synthetic data created by a GAN. This research demonstrates the capability of the tree-guided gradient boosting method to have effective performance and sound application in resource-constrained environments such as the IoT. It enhances detection capabilities and reduces computational costs. The model exhibited robust effectiveness for intrusion detection, in a time where our digital world has become more connected and more vulnerable, suggesting a more scalable approach was examined with an 80-20 training-test split toward improving cybersecurity for enterprise. The research will further continue to adapt to evolving malware attack strategies and to encourage the actual implementation of results on the Internet of Things.