Detection of Malicious Domains Using Passive DNS with XGBoost

• Autor(a) principal: Silveira, Marcos Rogério [UNESP]
• Data de Publicação: 2020
• Outros Autores: Cansian, Adriano Mauro [UNESP], Kobayashi, Hugo Koji
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1109/ISI49825.2020.9280552; http://hdl.handle.net/11449/208294

The Domain Name System (DNS) has as its main function the mapping of domain names to IPs and vice versa. Because of its function combined with the exponential growth of the internet, it has become an essential component. Because of this, attackers use DNS for malicious activities, such as Phishing, Fast-Flux Domains, DGAs, in addition to the spread of malware. In this paper we present an approach for automatic detection of malicious domains using a Passive DNS dataset combined with machine learning techniques. One way to perform the detection of these malicious domains is by blocklists, which can take some time before someone reports and there is human analysis. The model presented in this work is capable of detecting malicious domains at an early stage through its Passive DNS traffic. 12 features were extracted exclusively from DNS traffic. Our model makes use of the XGBoost supervised machine learning algorithm, and obtains an average AUC of 0.976.