Encrypted Network Traffic Classification Using Deep and Parallel Network-in-Network Models
Классификация зашифрованного сетевого трафика с использованием глубоких и параллельных моделей «сеть в сети»
2020-01-01
SCID: 54.1/y3wpz4f9
Discuss with AI
F1 scoreISCX VPN-nonVPN datasetencrypted network traffic classificationnetwork-in-network modelsparallel decision strategy
Figures from the paper
Abstract (AI)
Network traffic classification aims to recognize different application or traffic types by analyzing received data packets. This paper presents a neural network model with deep and parallel network-in-network (NIN) structures for classifying encrypted network traffic. Comparing with standard convolutional neural networks (CNN), NIN adopts a micro network after each convolution layer to enhance local modeling. Besides, NIN utilizes a global average pooling instead of traditional fully connected layers before final classification, which reduces the number of model parameters significantly. In our proposed method, deep NIN models with multiple MLP convolutional layers are built to map fixed-length packet vectors towards application or traffic labels. Furthermore, a parallel decision strategy of building two sub-networks to process packet header and packet body separately is designed considering that they may carry different kinds of clues for classification. The results of our experiments on the “ISCX VPN-nonVPN” encrypted traffic dataset show that NIN models can achieve a better balance between classification accuracy and model complexity than conventional CNNs. The parallel decision strategy can further improve the accuracy of using single NIN model for encrypted network traffic classification. Finally, the test set F1 scores of 0.983 and 0.985 are achieved for traffic characterization and application identification respectively.
Key Findings
1
A deep Network-in-Network (NIN) model classifies encrypted network traffic by mapping fixed-length packet vectors to application or traffic labels.
2
A parallel architecture processes packet headers and bodies separately, exploiting their different classification clues and improving accuracy over a single NIN model.
3
NIN’s micro-networks after convolution layers improve local modeling, while global average pooling substantially reduces parameters compared with fully connected layers.
4
On the ISCX VPN-nonVPN dataset, NIN models provide a better balance between classification accuracy and model complexity than conventional CNNs.
5
The proposed approach achieves test-set F1 scores of 0.983 for traffic characterization and 0.985 for application identification.
Research Object
encrypted network traffic, represented by fixed-length packet vectors and separated packet headers and bodies
Research Subject
application and traffic-type classification performance, including the accuracy–model-complexity trade-off and the contribution of separate header/body processing
Publication Details
Publication Date
2020-01-01
Journal
Publisher
ISSN
Cited by
98
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai9
Deep Residual Learning for Image Recognition2016
ImageNet classification with deep convolutional neural networks2017
Reducing the Dimensionality of Data with Neural Networks2006
Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research Groups2012
End-to-end encrypted traffic classification with one-dimensional convolution neural networks2017
Deep Learning for Encrypted Traffic Classification: An Overview2019
Mobile Encrypted Traffic Classification Using Deep Learning: Experimental Evaluation, Lessons Learned, and Challenges2019
$Deep-Full-Range$ : A Deep Learning Based Network Encrypted Traffic Classification and Intrusion Detection Framework2019
A Survey of Techniques for Mobile Service Encrypted Traffic Classification Using Deep Learning2019