Optimized Ensemble Model with Genetic Algorithm for DDoS Attack Detection in IoT Networks
Makhduma F. Saiyed, Irfan Al‐Anbagi · 2024
The growth in Internet of Things (IoT) networks has made them more vulnerable to various cyber threats, in-cluding Distributed Denial of Service (DDoS) attacks. Addressing DDoS attacks in resource-constrained IoT environments demands advanced detection methods beyond traditional cybersecurity. Existing machine learning and deep learning models have a tradeoff between accuracy and complexity. Pruning and quan-tization techniques present challenges related to precision and customization, highlighting the need for more balanced solutions. In response to these challenges, this paper introduces a novel Optimized Ensemble Model with Genetic Algorithm (OMEGA) system designed to detect high- and low-volume DDoS attacks in resource-constrained IoT networks. The system employs a combination of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) networks in its ensemble model to detect these attacks. In addition, the system employs novel post-training GA-based pruning and Min-Max quantization techniques for optimization. This combination enhances the detection accu-racy of high- and low-volume DDoS and significantly reduces computational demands, making the OMEGA system suitable for deployment in edge devices with limited resources. The OMEGA system is tested using a real-world IoT testbed and various datasets, showing an accuracy of over 90%.