TSCRNN: A novel classification scheme of encrypted traffic based on flow spatiotemporal features for efficient management of IIoT

Lin, Kunda; Xu, Xiaolong; Gao, Honghao

doi:10.1016/j.comnet.2021.107974

Cited by 92 publications

(52 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experimental results are presented in Table 9. It can be seen that the DL-based supervised model such as [37] and [38] outperforms all the approaches and specifically our model because they are more complex. However, this is not the case with the DT-based approach [36], where our model gives better results.…”

Section: Experiments On the Vpn-nonvpn Datasetmentioning

confidence: 94%

Handling partially labeled network data: A semi-supervised approach using stacked sparse autoencoder

2022

View full text Add to dashboard Cite

Section: Experiments On the Vpn-nonvpn Datasetmentioning

confidence: 94%

Handling partially labeled network data: A semi-supervised approach using stacked sparse autoencoder

2022

View full text Add to dashboard Cite

“…(2) statistical feature methods: AppScanner [36], CUMUL [27], BIND [1] and k-fingerprinting (K-fp) [12]; (3) deep learning methods: Deep Fingerprinting (DF) [35], FS-Net [22], GraphDApp [33], TSCRNN [21], Deeppacket [25]; (4) pre-training method: PERT [13]. The experimental results are shown in Tables 2 and 3.…”

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

“…Encrypted traffic classification using supervised deep learning have become a popular approach that automatically extracts discriminative features rather than relying on manual design. DF [35] uses convolutional neural networks (CNNs) and FS-Net [22] uses recurrent neural networks (RNNs) to automatically extract representations from raw packet size sequences of encrypted traffic, while Deeppacket [25] and TSCRNN [21] are characterizing raw payloads. However, this approach relies on a large amount of supervised data to capture valid features thus learning biased representations in imbalanced data, while our model does not rely on large labeled data.…”

Section: Related Work 21 Encrypted Traffic Classificationmentioning

confidence: 99%

“…These methods highly rely on expert-designed features and have limited generalization ability. Recently, deep learning methods [21,22] automatically learn complicated patterns from the raw traffic (Figure 1(c)), and achieve remarkable performance improvement. However, these methods highly rely on the amount and distribution of labelled training data, which is easy to cause model bias and hard to adapt to newly emerged encryption.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ET-BERT: A Contextualized Datagram Representation with Pre-training Transformers for Encrypted Traffic Classification

Lin,

Xiong,

Gou

et al. 2022

Preprint

View full text Add to dashboard Cite

Encrypted traffic classification requires discriminative and robust traffic representation captured from content-invisible and imbalanced traffic data for accurate classification, which is challenging but indispensable to achieve network security and network management. The major limitation of existing solutions is that they highly rely on the deep features, which are overly dependent on data size and hard to generalize on unseen data. How to leverage the open-domain unlabeled traffic data to learn representation with strong generalization ability remains a key challenge. In this paper, we propose a new traffic representation model called Encrypted Traffic Bidirectional Encoder Representations from Transformer (ET-BERT), which pre-trains deep contextualized datagram-level representation from large-scale unlabeled data. The pre-trained model can be fine-tuned on a small number of task-specific labeled data and achieves state-of-the-art performance across five encrypted traffic classification tasks, remarkably pushing the F1 of ISCX-VPN-Service to 98.9% (5.2%↑), Cross-Platform (Android) to 92.5% (5.4%↑), CSTNET-TLS 1.3 to 97.4% (10.0%↑). Notably, we provide explanation of the empirically powerful pre-training model by analyzing the randomness of ciphers. It gives us insights in understanding the boundary of classification ability over encrypted traffic.

show abstract

“…In preprocessing stage, raw traffic information is treated using sampling, vectorization, and flow segmentation, so on. Lin et al [14] proposed permissioned private blockchain based solutions for securing the image when encrypting. In this system, the cryptographic pixel value of images is kept on the blockchain, ensuring the security and privacy of image data.…”

Section: Introductionmentioning

confidence: 99%

Encryption with Image Steganography Based Data Hiding Technique in IIoT Environment

Ragab¹,

Alshehri²,

Alhadrami³

et al. 2022

Computers, Materials &Amp; Continua

View full text Add to dashboard Cite

Rapid advancements of the Industrial Internet of Things (IIoT) and artificial intelligence (AI) pose serious security issues by revealing secret data. Therefore, security data becomes a crucial issue in IIoT communication where secrecy needs to be guaranteed in real time. Practically, AI techniques can be utilized to design image steganographic techniques in IIoT. In addition, encryption techniques act as an important role to save the actual information generated from the IIoT devices to avoid unauthorized access. In order to accomplish secure data transmission in IIoT environment, this study presents novel encryption with image steganography based data hiding technique (EIS-DHT) for IIoT environment. The proposed EIS-DHT technique involves a new quantum black widow optimization (QBWO) to competently choose the pixel values for hiding secrete data in the cover image. In addition, the multi-level discrete wavelet transform (DWT) based transformation process takes place. Besides, the secret image is divided into three R, G, and B bands which are then individually encrypted using Blowfish, Twofish, and Lorenz Hyperchaotic System. At last, the stego image gets generated by placing the encrypted images into the optimum pixel locations of the cover image. In order to validate the enhanced data hiding performance of the EIS-DHT technique, a set of simulation analyses take place and the results are inspected interms of different measures. The experimental outcomes stated the supremacy of the EIS-DHT technique over the other existing techniques and ensure maximum security.

show abstract

TSCRNN: A novel classification scheme of encrypted traffic based on flow spatiotemporal features for efficient management of IIoT

Cited by 92 publications

References 19 publications

Handling partially labeled network data: A semi-supervised approach using stacked sparse autoencoder

Handling partially labeled network data: A semi-supervised approach using stacked sparse autoencoder

ET-BERT: A Contextualized Datagram Representation with Pre-training Transformers for Encrypted Traffic Classification

Encryption with Image Steganography Based Data Hiding Technique in IIoT Environment

Contact Info

Product

Resources

About