Ensemble Deep Learning Model based on Multi-Class Classification Technique to Detect Cyber Attacks in IoT Environment
Ahmed Alrefaei, Mohammad Ilyas · 2024
The Internet of Things (IoT) has seen a remarkable expansion, encompassing devices that connect to the internet, collect data, and share information. However, this growth has introduced a host of security challenges, leading to an increase in cyber-attacks. To address these challenges and ensure data integrity, researchers have been focusing on developing Intrusion Detection Systems (IDS) using deep ensemble models. This paper presents a multi-class classification algorithm approach for intrusion detection in IoT, employing a deep ensemble model incorporating Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM). Evaluation metrics such as F1 score, recall, and precision under the Receiver Operating Characteristics (ROC) curve are utilized. The proposed models’ proficiency is demonstrated using the UNSW 2018 IoT Botnet dataset, showing that the deep ensemble multi-class classification models achieve high accuracy in detecting IoT attacks with 98.4%.