Intrusion Detection and Classification in Mobile Ad Hoc Networks Using Machine Learning

Abstract

Mobile Ad Hoc Networks (MANETs) are decentralized, self-organizing networks that operate without fixed infrastructure, making them highly flexible and vulnerable to various security threats. Ensuring security in MANETs is challenging due to their dynamic topology, lack of centralized monitoring, and resource constraints. Traditional security mechanisms, such as encryption and authentication, often fail to detect sophisticated intrusion attacks in these networks. Existing intrusion detection systems (IDS) struggle with high false alarm rates, adaptability to evolving attacks, and computational overhead, creating a significant research gap in MANET security. Motivated by the increasing security threats and the inadequacy of existing methods, this study aims to develop an adaptive and efficient machine learning-based intrusion detection and classification system for MANETs. The research investigates the effectiveness of multiple machine learning classifiers, including Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), and an ensemble-based approach, in detecting and classifying network intrusions. The study utilizes the UNSW-NB15 dataset, which contains diverse attack types, and employs feature selection techniques to optimize model performance while minimizing computational costs. The proposed ensemble model integrates multiple classifiers to improve detection accuracy and robustness. Experimental results demonstrate that the ensemble model outperforms individual classifiers, achieving an accuracy of 98.53%, a precision of 98.79%, a recall of 98.78%, and an F1-score of 98.78%. This improvement significantly reduces false positives while maintaining high detection capability, addressing key limitations of traditional IDS. This study's contributions include the development of an efficient, adaptive ML-based IDS tailored for MANETs, a comparative analysis of machine learning models for intrusion detection, and the application of feature selection to enhance detection efficiency. The findings provide valuable insights for securing MANET environments and lay the foundation for future real-time adaptive intrusion detection research.