A Novel Hybrid Method Using Grey Wolf Algorithm and Genetic Algorithm for IoT Botnet DDoS Attacks Detection

• Published in: International journal of computational intelligence systems Vol. 18; no. 1; pp. 1 - 61
• Main Authors: Maazalahi, Mahdieh, Hosseini, Soodeh
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 18.03.2025; Springer Nature B.V; Springer
• Subjects: Algorithms; Artificial Intelligence; Botnet attack; Computational Intelligence; Control; Cybersecurity; Datasets; DDOS; Denial of service attacks; Engineering; Evolutionary algorithms; Feature selection; Genetic algorithms; Genetics algorithm; Grey Wolf algorithm; Heuristic methods; Internet of Things; Intrusion; Intrusion detection systems; Machine learning; Malware; Mathematical Logic and Foundations; Mechatronics; Meta-heuristic algorithm; Research Article; Robotics; Machine learning algorithms; Meta-heuristic algorithm; Grey Wolf algorithm; Genetics algorithm; Evolutionary algorithms; DDOS; Botnet attack
• ISSN: 1875-6883; 1875-6891; 1875-6883
• DOI: 10.1007/s44196-025-00774-y

The Internet of Things (IoT) is a vast network of interconnected physical objects that has improved the conditions for a computer-based physical world and improved efficiency. With the increase in communication in an IoT system, Internet security has decreased, and the most dangerous and sophisticated attacks in the IoT have emerged, i.e., DDoS and Botnet attacks. DDoS attacks are a serious threat to the availability of Internet services, especially since botnets can now be launched by almost anyone. In this situation, the use of an intrusion detection system (IDS) is essential to detect intruders and maintain the security of IoT networks. In this paper, a new IDS is proposed to detect IoT-Botnet DDoS attacks. This IDS is a new three-phase system, the first phase is related to preprocessing on the dataset and the second phase includes a new hybrid method for feature selection using filter and wrapper methods based on the Grey Wolf (GW) algorithm and genetics called GW-GA. In this method, the initial population is randomly selected and then at each stage, feature selection is done by both algorithms simultaneously and the final answer is compared and the best solutions are given as a new population to both algorithms and the third phase includes the use of machine learning and metaheuristic algorithms as classifiers. In the proposed method and to verify the performance, it is evaluated using the large BOT-IoT dataset. The results show that the proposed method significantly reduces the feature and also increases the classification accuracy compared to other methods, and the RF and Bagging algorithms have achieved a maximum recognition accuracy of 0.999. The dimensions of BOT-IoT have been reduced from 46 features to 12.