Intrusion detection in IoT networks using machine learning and deep learning approaches for MitM attack mitigation
Mohammed Ansar Ali, Salah Alawi Hussein Al-Sharafi
Abstract
The Internet of Things (IoT) revolutionizes connectivity but introduces significant security challenges, particularly concerning man-in-the-middle (MitM) attacks. These attacks, where adversaries intercept or alter communications, threaten the integrity and privacy of IoT networks due to the limited computational resources and security features of IoT devices. This study investigates machine learning and deep learning techniques to detect and mitigate MitM attacks in IoT environments. Three approaches, Long Short-Term Memory (LSTM), Random Forest, and Support Vector Machine (SVM), were assessed using a pre-prepared dataset that simulates network traffic and various attack scenarios. Random Forest achieved the highest accuracy (94%), combining robustness and balanced performance. LSTM excelled in analyzing sequential data with 92% accuracy, while SVM struggled with high false positive rates despite 85.7% accuracy. These findings highlight the potential of machine learning, particularly Random Forest and LSTM, to enhance IoT security. They pave the way for more resilient and effective frameworks for detecting and mitigating MitM attacks.