Detecção de ataques de negação de serviço em redes definidas por software com o uso de florestas aleatórias

Abstract

The advancement of technologies such as the Internet of Things (IoT), cloud computing, 5G networks, and streaming services has increased the complexity and security requirements of computer networks. In this context, Software-Defined Networking (SDN) emerges as a promising solution by enabling greater flexibility and centralized control. However, this centralization also introduces new attack vectors, especially Denial-of-Service (DoS) attacks, which can compromise the operation of the SDN controller. This dissertation proposes a DoS attack detection system based on Machine Learning algorithms, with an emphasis on the Random Forest model. The system was developed in a simulated environment and evaluated using synthetic data and the DARPA 2009 benchmark, achieving an accuracy of 99.93% after hyperparameter tuning. The results obtained demonstrate the feasibility of the approach and its potential for application in real-world environments, offering an effective and adaptable solution for security monitoring in SDN networks.