Enhancing IoT Security Through the Integration of Explainable AI and Ensemble Xgboost for Improved Intrusion Detection

The rapid growth of the Internet of Things (IoT) has significantly impacted industries such as smart cities, but it has also introduced new cybersecurity challenges. Traditional intrusion detection systems (IDS) often struggle to adapt to the complex and changing nature of IoT networks, creating a need for more advanced and transparent solutions. This study introduces an Explainable Ensemble XGBoost Model designed to enhance intrusion detection in IoT environments. By combining the high accuracy of the XGBoost algorithm with Shapley Additive Explanations (SHAP) values, the model offers both strong detection capabilities and clear interpretability. Preprocessing of the RTIoT2022 dataset included addressing missing values, removing duplicates, and standardizing features. The model produced exceptional results, achieving accuracy exceeding 99.9 %. SHAP analysis indicated that packet-level metrics and flow timing are the most important features in identifying different types of intrusions. This study addresses a critical gap in IoT security by integrating accurate detection with explainable AI, offering a solution that enhances both security and transparency. The results emphasize the importance of integrating explainability into cybersecurity systems to strengthen IoT networks against changing cyber threats.