BoostedEnML: Efficient Technique for Detecting Cyberattacks in IoT Systems Using Boosted Ensemble Machine Learning

Area of Science:

• Cybersecurity
• Machine Learning
• Network Intrusion Detection

Background:

• Internet of Things (IoT) systems face increasing security threats, necessitating robust intrusion detection systems (IDS).
• Existing machine learning-based IDS often struggle with the dynamic nature of IoT networks and require improved detection rates and lower computational overhead.
• Ensemble methods offer a promising approach to enhance IDS performance by combining multiple machine learning classifiers.

Purpose of the Study:

• To propose an efficient and accurate intrusion detection system (IDS) for Internet of Things (IoT) environments.
• To develop a novel ensemble model, BoostedEnML, leveraging boosted machine learning classifiers for enhanced cyberattack detection.
• To address the challenge of detecting multiple, sophisticated attacks in IoT networks with high precision and efficiency.

Main Methods:

• Trained six distinct machine learning classifiers (DT, RF, ET, LGBM, AD, XGB) and created ensembles using stacking and majority voting.
• Employed data balancing techniques including Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic (ADASYN) sampling.
• Constructed the BoostedEnML model using LightGBM and XGBoost, validated on two datasets containing diverse high-profile attacks (e.g., DDoS, DoS, botnets).

Main Results:

• The proposed BoostedEnML model demonstrated superior performance compared to existing ensemble models.
• Achieved 100% accuracy, precision, recall, F-score, and Area Under the Curve (AUC) for multiclass classification on the selected datasets.
• The combination of LightGBM and XGBoost resulted in a lightweight yet highly efficient intrusion detection model.

Conclusions:

• BoostedEnML offers a highly effective solution for detecting a wide range of cyberattacks in IoT networks.
• The model's high performance metrics indicate its potential for real-world application in securing IoT ecosystems.
• The research highlights the efficacy of boosted ensemble methods in creating efficient and accurate intrusion detection systems for resource-constrained IoT environments.