Network Attack Detection in the PfSense Firewall via Deep Learning
Merzoug Soltane, Aimen Gasmi · 2024
The detection of network attacks is a crucial aspect in ensuring the sustainability and proper functioning of information systems. Complex threat patterns and malicious actors possess the ability to inflict significant damage to cyber systems. In this study, we propose novel approaches utilizing deep learning techniques to identify and respond to threats and alerts found within network logs acquired through pfSense, an open-source firewall software. pfSense incorporates various robust security services including firewall capabilities, URL filtering, and virtual private networking (VPN), among others. The primary objective of this research is to analyze the acquired logs from a local installation of pfSense software, in order to develop a potent and efficient solution that can effectively manage traffic flow by automatically recognizing patterns through the proposed deep learning architectures, which present a significant challenge. To accomplish this, we have devised an attack detection system based on the CICDDOS2019 dataset, leveraging a deep neural network (DNN) model that has been seamlessly integrated into pfSense, enabling automatic identification and prevention of attacks.