Rank and Wormhole Attack Detection Model for RPL-Based Internet of Things Using Machine Learning.

Area of Science:

• Computer Science
• Cybersecurity
• Network Engineering

Background:

• The rapid expansion of Internet of Things (IoT) technology presents significant security vulnerabilities, particularly within its communication infrastructure.
• The Routing Protocol for Low-Power and Lossy Networks (RPL), commonly used in IoT, lacks robust security features due to its lightweight design, making it susceptible to various attacks.
• Attacks targeting RPL resources can lead to the collapse of IoT systems, necessitating advanced security solutions.

Purpose of the Study:

• To propose a novel, lightweight multiclass classification model for detecting both RPL-specific and sensor-network-inherited attacks in IoT environments.
• To address the lack of suitable datasets for training and evaluating IoT attack detection models by generating a new, comprehensive dataset.
• To optimize the model's performance through careful feature selection and the application of a light gradient boosting machine algorithm.

Main Methods:

• Development of a novel dataset by constructing diverse network models to simulate various attack scenarios.
• Implementation of optimal feature selection techniques to enhance the efficiency and accuracy of the detection model.
• Design and application of a light gradient boosting machine (LightGBM) algorithm tailored for multiclass classification of IoT security threats.

Main Results:

• The proposed MC-MLGBM model demonstrated high performance in detecting a wide range of IoT attacks, as evidenced by metrics such as accuracy, precision, and recall.
• Extensive experimental validation using confusion matrices confirmed the model's effectiveness in distinguishing between different attack types.
• Further evaluation using multiclass-specific metrics like cross-entropy, Cohen's kappa, and Matthews correlation coefficient validated the model's robustness and reliability compared to existing benchmarks.

Conclusions:

• The MC-MLGBM model offers an effective and lightweight solution for enhancing the security of IoT networks against complex cyber threats.
• The generated dataset and optimized feature selection provide a valuable resource for future research in IoT security.
• The study highlights the potential of machine learning approaches in securing resource-constrained IoT environments and the RPL protocol.