Malicious Traffic Detection in DNS Over HTTPS (DoH): Edge Prediction with Graph Convolutional Network
Pongsarun Boonyopakorn, Ukid Changsan · 2024
The widespread use of the internet on various devices has become crucial in daily life. However, internet connectivity raises significant concerns regarding privacy and security. One important aspect of internet usage is the Domain Name Server (DNS), which operates without encryption. To address these vulnerabilities, DNS over HTTPS has been introduced, encrypting DNS requests within HTTPS connections to enhance privacy and security. Nevertheless, attackers exploit these vulnerabilities using DNS Tunneling Tools for malicious purposes, compromising user security. Therefore, it becomes a challenge to implement Malicious Traffic Detection in DNS over HTTPS (DoH), conducting Edge Prediction with Graph Convolutional Network, achieving an accuracy rate of up to 96.29%. This will lead to its utilization with traffic capture devices to prevent attacks.