Logistic Regression Ensemble Classifier for Intrusion Detection System in Internet of Things

Area of Science:

• Computer Science
• Cybersecurity
• Network Security

Background:

• The Internet of Things (IoT) connects devices globally, but introduces security vulnerabilities.
• Node security breaches in IoT systems can lead to attacks and anomalies.
• Effective Intrusion Detection Systems (IDS) are crucial for securing IoT technologies.

Purpose of the Study:

• To propose an effective Intrusion Detection System (IDS) for Internet of Things (IoT) environments.
• To develop a Logistic Regression based Ensemble Classifier (LREC) for enhanced IoT security.
• To address data imbalance and feature selection challenges in IDS development.

Main Methods:

• Proposed a Logistic Regression based Ensemble Classifier (LREC) combining AdaBoost and Random Forest (RF).
• Utilized Adaptive Synthetic Sampling (ADASYN) to handle data imbalance.
• Employed Recursive Feature Elimination (RFE) for feature selection.
• Evaluated the RFE-LREC method on BoT-IoT and TON-IoT datasets.

Main Results:

• The RFE-LREC achieved a classification accuracy of 99.99% on the BoT-IoT dataset.
• Performance was analyzed using metrics including accuracy, recall, precision, F1-score, FAR, ROC, TNR, and MCC.
• Outperformed existing methods like TL-IDS and LSTM in classification accuracy.

Conclusions:

• The proposed RFE-LREC method offers a highly effective solution for Intrusion Detection Systems in IoT.
• The approach successfully addresses key challenges like data imbalance and feature selection.
• Demonstrated superior performance compared to existing state-of-the-art methods.