REA Press Null REA Press3042-31043042-3104 REA Press https://doi.org/10.48314/apem.v2i2.33 Research Article Internet of things, Traffic classification, Supervised learning, Distributed denial-of-service attack, Internet of things-23. A Fast Internet of Things DDoS Attack Detection Method Using Deep Feedforward NetworksA Fast Internet of Things DDoS Attack Detection Method Using Deep Feedforward Networks Babaiyan Vahide Computer Engineering and IT Department, Faculty of Electrical and Computer Engineering, Shiraz University of Technology, Shiraz, Iran. Bushehrian Omid Computer Engineering and IT Department, Faculty of Electrical and Computer Engineering, Shiraz University of Technology, Shiraz, Iran. Javidan Reza Computer Engineering and IT Department, Faculty of Electrical and Computer Engineering, Shiraz University of Technology, Shiraz, Iran. 05 2025 13 05 2025 2 2 © 2025 REA Press 2025

This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

 A Fast Internet of Things DDoS Attack Detection Method Using Deep Feedforward Networks

The increasing use of Internet of Things (IoT) devices has led to a surge in data traffic, which can be vulnerable to intentional denial-of-service (DoS) attacks that disrupt the intended Quality of Service (QoS). This paper presents a deep learning-based approach using Feedforward Neural Networks (FNNs) to detect Distributed Denial-of-Service (DDoS) attacks in IoT networks. We evaluated the performance of this approach on the IoT-23 dataset, which included captures of both malware-infected and benign IoT traffic. We conducted a comparative analysis between the FNN approach and three commonly used Machine Learning (ML) models, namely, Support Vector Machines (SVM), Random Forests (RFs), and Gradient Boosting (GRB). Our findings demonstrate that all methods achieve similar levels of accuracy. However, the FNN model distinguishes itself with significantly higher precision than the other models. Furthermore, our analysis revealed that FNN exhibits the shortest learning time among the considered models.

Null