Leveraging ensemble learning with metaheuristic optimization algorithms for an intelligent cyberattack defense framework in an IoT environment
Hend Khalid Alkahtani, Mashael M. Asiri, Rakan Khalid Marzouq Alanazi, Mohammed Mujib Alshahrani, Fahad Ahmed Al-Zahrani, Shaymaa E. Sorour, Mesfer Al Duhayyim · Alexandria Engineering Journal · 2025
Cybersecurity continues to be a significant problem for some industries on the Internet, as the number of security breaches is increasing over time. It is identified that many zero-day threats are continuously developing due to the addition of several protocols, mainly from the Internet of Things (IoT). The majority of these attacks are smaller versions of formerly known cyberattacks. IoT cybersecurity aims to decrease cybersecurity risk for companies and users by securing privacy and assets. The expansion of automatic devices for cyber threat classification and detection utilizing artificial intelligence (AI) and deep learning (DL) devices has become necessary for accomplishing security in IoT environments. Due to their notable performance, DL-based methods are essential to successfully reducing security problems associated with IoT gadgets. This article presents a Leveraging Ensemble Learning and Metaheuristic Optimization Algorithms for Intelligent Cyber Attack and Defense Framework (LELMOA-ICADF) model in IoT networks. The main intention of the LELMOA-ICADF model is to deliver an efficient method using advanced ensemble models for enhancing IoT cybersecurity. Initially, the min-max normalization is employed in the data pre-processing stage for transforming input data into a structured format. The cat swarm optimization (CSO) technique is utilized for the feature selection process to choose the most relevant and significant features from the dataset. Furthermore, ensemble models such as the bidirectional long short-term memory (BiLSTM), bidirectional gated recurrent unit (BiGRU), and deep belief network (DBN) are employed for the attack classification process. Finally, the artificial bee colony (ABC) method is used for parameter tuning to improve the classification performance of ensemble classifiers. The experimental assessment of the LELMOA-ICADF approach is performed under Edge-IIoT and ToN-IoT datasets. The performance validation of the LELMOA-ICADF approach portrayed a superior accuracy value of 99.40 % and 99.50 % under the dual datasets.