Enhancing Internet of Things security using performance gradient boosting for network intrusion detection systems

The rise of the Internet of things (IoT) changes inter-device communication to increase efficacy and livability conditions in all sectors. However, it is proportion of security vulnerabilities making IoT networks prime targets for sophisticated cyber threats. The deployment of solution methods is needed to embark on new grounds in exploring IoT cybersecurity with the current state-of-the-art machine learning (ML) techniques to fortify IoT networks from such malicious threats. This paper highlights critical vulnerabilities in IoT traffic through a detailed analysis and performance evaluation of the state-of-the art ensemble classifiers, eXtreme gradient boosting (XGBoost) and light gradient boosting machine (LGBM), to understand their detection capability for a diverse set of cyber-attacks. From the results, it can be seen that XGBoost and LGBM classifiers outperformed the conventional models with an extraordinary average accuracy of 99.553 % and 99.651 %, respectively in the definition of true threats. The performance metrics proved better detection capabilities of the classifiers with the potential that their use affords to minimize false positives and false negatives, which are preponderant considerations for the integrity of an IoT network. Further comparative analysis tries to furnish the strengths and limitations of these classifiers and propose a practical framework for their implementation to strengthen real-life IoT environments against cyber threats. This work delivers in-depth findings on IoT network behaviors and threat patterns using synthetic minority oversampling technique (SMOTE) for class balancing and analyzing the importance of features in IoT behavioral RT-IoT2022 dataset. This is also considered to be a foundational resource for creating cutting-edge AI-driven defense mechanisms to tackle ever-evolving cybersecurity threats.