Darknet traffic detection and characterization with models based on decision trees and neural networks

• DOI: 10.1016/j.iswa.2023.200199
• Publication Date: 2023-02-17T12:46:43Z
• Authors: Mateus Coutinho M...; Paulo Vitor Barbo...; Alex B. Vieira; Antonino Galletta; Massimo Villari; Roberto M. de Oli...; Edelberto Franco ...
• Keywords: Relevance; Feature (linguistics); Traffic classification

The Darknet is a set of networks and technologies, having as fundamental principles anonymity security. In many cases, they are associated with illicit activities, opening space for malware traffic attacks to legitimate services. To prevent misuse necessary classify characterize its existing traffic. this paper, we the real available from CIC-Darknet2020 dataset. that sense, performed feature extraction grouped possible subnets an n-gram approach. Furthermore, evaluated relevance best features selected by Recursive Feature Elimination method problem. Our results indicate simple models, like Decision Trees Random Forests, reach accuracy above 99% on classification. methodology represents gain up 13% in comparison state-of-the-art.