AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement
With the continuous expansion of the darknet and the increase in various criminal activities in the darknet, darknet traffic identification has become increasingly essential. However, existing darknet traffic identification methods rely on all traffic characteristics, which require a long computing...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-08-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/13/16/9353 |
_version_ | 1797585579710349312 |
---|---|
author | Tao Yang Rui Jiang Hongli Deng Qinru Li Ziyu Liu |
author_facet | Tao Yang Rui Jiang Hongli Deng Qinru Li Ziyu Liu |
author_sort | Tao Yang |
collection | DOAJ |
description | With the continuous expansion of the darknet and the increase in various criminal activities in the darknet, darknet traffic identification has become increasingly essential. However, existing darknet traffic identification methods rely on all traffic characteristics, which require a long computing time and a large amount of system resources, resulting in low identification efficiency. To this end, this paper proposes an autoencoder-based darknet traffic identification method (AE-DTI). First, AE-DTI maps the feature values to pixels of a two-dimensional grayscale image after deduplication and denoising of the darknet traffic dataset. Then, AE-DTI designs a new feature selection algorithm (AE-FS) to downscale the grayscale graph, and AE-FS trains a feature scoring network, which globally scores all the features based on the reconstruction error to select the features with scores greater than or equal to a set threshold value. Finally, AE-DTI uses a one-dimensional convolutional neural network with a dropout layer to identify darknet traffic on the basis of alleviating overfitting. Experimental results on the ISCXTor2016 dataset show that, compared with other dimensionality reduction methods (PCA, LLE, ISOMAP, and autoencoder), the classification model trained with the data obtained from AE-FS has a significant improvement in classification accuracy and classification efficiency. Moreover, AE-DTI also shows significant improvement in recognition accuracy compared with other models. Experimental results on the CSE-CIC-IDS2018 dataset and CIC-Darknet2020 dataset show that AE-DTI has strong generalization. |
first_indexed | 2024-03-11T00:09:12Z |
format | Article |
id | doaj.art-06688a419f8e4fe3858995a0112ab7e6 |
institution | Directory Open Access Journal |
issn | 2076-3417 |
language | English |
last_indexed | 2024-03-11T00:09:12Z |
publishDate | 2023-08-01 |
publisher | MDPI AG |
record_format | Article |
series | Applied Sciences |
spelling | doaj.art-06688a419f8e4fe3858995a0112ab7e62023-11-19T00:08:25ZengMDPI AGApplied Sciences2076-34172023-08-011316935310.3390/app13169353AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder ImprovementTao Yang0Rui Jiang1Hongli Deng2Qinru Li3Ziyu Liu4Education Information Technology Center, China West Normal University, Nanchong 637001, ChinaSchool of Computer Science, China West Normal University, Nanchong 637001, ChinaEducation Information Technology Center, China West Normal University, Nanchong 637001, ChinaSchool of Computer Science, China West Normal University, Nanchong 637001, ChinaSchool of Electronic and Information Engineering, China West Normal University, Nanchong 637001, ChinaWith the continuous expansion of the darknet and the increase in various criminal activities in the darknet, darknet traffic identification has become increasingly essential. However, existing darknet traffic identification methods rely on all traffic characteristics, which require a long computing time and a large amount of system resources, resulting in low identification efficiency. To this end, this paper proposes an autoencoder-based darknet traffic identification method (AE-DTI). First, AE-DTI maps the feature values to pixels of a two-dimensional grayscale image after deduplication and denoising of the darknet traffic dataset. Then, AE-DTI designs a new feature selection algorithm (AE-FS) to downscale the grayscale graph, and AE-FS trains a feature scoring network, which globally scores all the features based on the reconstruction error to select the features with scores greater than or equal to a set threshold value. Finally, AE-DTI uses a one-dimensional convolutional neural network with a dropout layer to identify darknet traffic on the basis of alleviating overfitting. Experimental results on the ISCXTor2016 dataset show that, compared with other dimensionality reduction methods (PCA, LLE, ISOMAP, and autoencoder), the classification model trained with the data obtained from AE-FS has a significant improvement in classification accuracy and classification efficiency. Moreover, AE-DTI also shows significant improvement in recognition accuracy compared with other models. Experimental results on the CSE-CIC-IDS2018 dataset and CIC-Darknet2020 dataset show that AE-DTI has strong generalization.https://www.mdpi.com/2076-3417/13/16/9353network traffic classificationdeep learningfeature selectionmachine learning |
spellingShingle | Tao Yang Rui Jiang Hongli Deng Qinru Li Ziyu Liu AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement Applied Sciences network traffic classification deep learning feature selection machine learning |
title | AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement |
title_full | AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement |
title_fullStr | AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement |
title_full_unstemmed | AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement |
title_short | AE-DTI: An Efficient Darknet Traffic Identification Method Based on Autoencoder Improvement |
title_sort | ae dti an efficient darknet traffic identification method based on autoencoder improvement |
topic | network traffic classification deep learning feature selection machine learning |
url | https://www.mdpi.com/2076-3417/13/16/9353 |
work_keys_str_mv | AT taoyang aedtianefficientdarknettrafficidentificationmethodbasedonautoencoderimprovement AT ruijiang aedtianefficientdarknettrafficidentificationmethodbasedonautoencoderimprovement AT honglideng aedtianefficientdarknettrafficidentificationmethodbasedonautoencoderimprovement AT qinruli aedtianefficientdarknettrafficidentificationmethodbasedonautoencoderimprovement AT ziyuliu aedtianefficientdarknettrafficidentificationmethodbasedonautoencoderimprovement |