A Convolutional Neural Network/Random Forest Hybrid Model for DoS Attack Detection in IoT Networks
Elina Valentina Jaimes Bastidas, Germán A. Montoya, Carlos Lozano-Garzón · 2025
The Internet of Things (IoT) is a rapidly evolving technology field that has experienced significant growth in recent years. This expansion has introduced new security challenges, particularly regarding attacks that can spread within IoT networks, especially denial-of-service (DoS) attacks. To detect this attack effectively, this work proposes a novel integration of Convolutional Neural Network (CNN) and Random Forest (RF) techniques, demonstrating high performance by accurately identifying over 99% of attacks on the popular UNSW-NB15 dataset. The model improved the classification of previously mis-labeled traces by 41%, highlighting its effectiveness in enhancing detection accuracy. Moreover, considering the promising results achieved, our approach can be applied to address other types of attacks.