Enhancing resilience in mobile ad hoc networks: deluge attack mitigation via machine learning based anomaly detection
A. Sharath Kumar, Santanu Kumar Sahoo, Mithhil Arora, Ramachandran Thulasiram, Aniruddha Ram, Amit Kumar · Multidisciplinary Science Journal · 2025
Mobile Ad Hoc Networks (MANETs) face multiple security threats, with deluge attacks emerging as a particular threat to network performance. These conventional measures frequently fail in detecting or combating such attacks, considering the dynamic and decentralized nature of MANETs. A Deluge Attack on MANETs is a DoS attack in which an attacker overwhelms the network with high levels of routing or data packets and consumes bandwidth and battery resources. ML anomaly detection recognizes unusual traffic patterns, allowing for early detection and mitigation, and thus improving MANET resilience against such attacks and providing efficient network performance. In this research, an innovative technique can be proposed for ensuring enhanced resilience against deluge attacks: the Ensemble Red Fox Optimized - Scalable Random Forest Algorithm (ERFO-SRFA) that employs machine-learning-based anomaly detection techniques. The model proposed uses a tailored anomaly detection frame in real-time to identify and mitigate an ongoing deluge attack. A dataset is taken from the KDD Cup competition with the preprocessed data normalized by min-max normalization, creating a background for standardizing input features. The experimental trials demonstrate the efficiency of the ERFO-SRFA approach in drawing the line between normal network behavior and malicious activities associated with deluge attacks. The model demonstrates impressive results, showing an accuracy of 98.45%, precision of 98.36%, recall of 98.25%, and an F1-score of 98.44%. This indicates that the ERFO-SRFA method for deluge attack detection is reliable and presents few false positives. Thus, this research enhances the protection of MANETs by considerably improving deluge attack detection and mitigation, enabling the development of more robust mobile networking systems. The successful execution of this machine learning-based anomaly detection system represents a significant milestone in strengthening MANETs against the ever-changing landscape of security adversities. The research proposes a ML-based anomaly detection method to counter Deluge attacks in MANETs. The model is highly effective in identifying attack patterns, improving network security and resilience. Experimental results indicate remarkable improvements in detection accuracy and response time, validating the effectiveness of the model in protecting MANET communications.