Close
Metadata

%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi/2020/10.01.17.17
%2 sid.inpe.br/sibgrapi/2020/10.01.17.17.22
%A Moreira, Rodrigo,
%A Rodrigues, Larissa Ferreira,
%A Rosa, Pedro Frosi,
%A Aguiar, Rui Luis Andrade,
%A Silva, Flávio de Oliveira,
%@affiliation Federal University of Uberlândia - Faculty of Computing (FACOM)
%@affiliation Federal University of Viçosa - Institute of Exact and Technological Sciences (IEP)
%@affiliation Federal University of Uberlândia - Faculty of Computing (FACOM)
%@affiliation University of Aveiro - Telecommunications Institute (IT)
%@affiliation Federal University of Uberlândia - Faculty of Computing (FACOM)
%T Packet Vision: a convolutional neural network approach for network traffic classification
%B Conference on Graphics, Patterns and Images, 33 (SIBGRAPI)
%D 2020
%E Musse, Soraia Raupp,
%E Cesar Junior, Roberto Marcondes,
%E Pelechano, Nuria,
%E Wang, Zhangyang (Atlas),
%S Proceedings
%8 Nov. 7-10, 2020
%J Los Alamitos
%I IEEE Computer Society
%C Virtual
%K Network traffic classification, convolutional neural networks, SDN, Network Slicing, data augmentation, fine-tuning.
%X Network traffic classification can improve the management and network service offer, taking into account the kind of application. The future network architectures, mainly mobile networks, foresee intelligent mechanisms in their architectural frameworks to deliver application-aware network requirements. The potential of convolutional neural networks capabilities, widely exploited in several contexts, can be used in network traffic classification. Thus, it is necessary to develop methods based on the content of packets which can transform them into a suitable input for CNN technologies. Hence, we implemented and evaluated the Packet Vision, a method capable of building images from packets raw-data, considering both header and payload. Our approach surpasses those found in the state-of-the-art, considering classification performance and regarding the fully-packet structure characteristic, delivering security and privacy by transforming the raw-data packet into images. Besides, we built a dataset with four traffic classes and evaluated three CNNs architectures, considering performance and the exploitation of training from scratch, fine-tuning and hyperparameter optimization. Experiments showcase applicability and suitability when combining Packet Vision with CNNs, which seemed to be a promising approach to deliver outstanding performance in the classification of network traffic.
%@language en
%3 17.pdf


Close