An Input-Agnostic Hierarchical Deep Learning Framework for Traffic Fingerprinting

Jian  Qu; Xiaobo Ma; Jianfeng Li; Xiapu Luo; Lei Xue; Junjie Zhang; Zhenhua Li; Feng Li; Xiaohong  Guan

An Input-Agnostic Hierarchical Deep Learning Framework for Traffic Fingerprinting

Jian Qu, Xiaobo Ma, Jianfeng Li, Xiapu Luo, Lei Xue, Junjie Zhang, Zhenhua Li, Feng Li, Xiaohong Guan

Department of Computing

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

Abstract

Deep learning has proven to be promising for traffic fingerprinting that explores features of packet timing and sizes. Although well-known for automatic feature extraction, it is faced with a gap between the heterogeneousness of the traffic (i.e., raw packet timing and sizes) and the homogeneousness of the required input (i.e., input-specific). To address this gap, we design an input-agnostic hierarchical deep learning framework for traffic fingerprinting that can hierarchically abstract comprehensive heterogeneous traffic features into homogeneous vectors seamlessly digestible by existing neural networks for further classification. The extensive evaluation demonstrates that our framework, with just one paradigm, not only supports heterogeneous traffic input but also achieves better or comparable performance compared to state-of-the-art methods black across a wide range of traffic fingerprinting tasks.

Original language	English
Title of host publication	32nd USENIX Security Symposium 2023
Publication status	Published - Aug 2023

Cite this

@inproceedings{cbaff7581e184605ad2eb6b260dd6f44,

title = "An Input-Agnostic Hierarchical Deep Learning Framework for Traffic Fingerprinting",

abstract = "Deep learning has proven to be promising for traffic fingerprinting that explores features of packet timing and sizes. Although well-known for automatic feature extraction, it is faced with a gap between the heterogeneousness of the traffic (i.e., raw packet timing and sizes) and the homogeneousness of the required input (i.e., input-specific). To address this gap, we design an input-agnostic hierarchical deep learning framework for traffic fingerprinting that can hierarchically abstract comprehensive heterogeneous traffic features into homogeneous vectors seamlessly digestible by existing neural networks for further classification. The extensive evaluation demonstrates that our framework, with just one paradigm, not only supports heterogeneous traffic input but also achieves better or comparable performance compared to state-of-the-art methods black across a wide range of traffic fingerprinting tasks.",

author = "Jian Qu and Xiaobo Ma and Jianfeng Li and Xiapu Luo and Lei Xue and Junjie Zhang and Zhenhua Li and Feng Li and Xiaohong Guan",

year = "2023",

month = aug,

language = "English",

booktitle = "32nd USENIX Security Symposium 2023",

}

TY - GEN

T1 - An Input-Agnostic Hierarchical Deep Learning Framework for Traffic Fingerprinting

AU - Qu, Jian

AU - Ma, Xiaobo

AU - Li, Jianfeng

AU - Luo, Xiapu

AU - Xue, Lei

AU - Zhang, Junjie

AU - Li, Zhenhua

AU - Li, Feng

AU - Guan, Xiaohong

PY - 2023/8

Y1 - 2023/8

N2 - Deep learning has proven to be promising for traffic fingerprinting that explores features of packet timing and sizes. Although well-known for automatic feature extraction, it is faced with a gap between the heterogeneousness of the traffic (i.e., raw packet timing and sizes) and the homogeneousness of the required input (i.e., input-specific). To address this gap, we design an input-agnostic hierarchical deep learning framework for traffic fingerprinting that can hierarchically abstract comprehensive heterogeneous traffic features into homogeneous vectors seamlessly digestible by existing neural networks for further classification. The extensive evaluation demonstrates that our framework, with just one paradigm, not only supports heterogeneous traffic input but also achieves better or comparable performance compared to state-of-the-art methods black across a wide range of traffic fingerprinting tasks.

AB - Deep learning has proven to be promising for traffic fingerprinting that explores features of packet timing and sizes. Although well-known for automatic feature extraction, it is faced with a gap between the heterogeneousness of the traffic (i.e., raw packet timing and sizes) and the homogeneousness of the required input (i.e., input-specific). To address this gap, we design an input-agnostic hierarchical deep learning framework for traffic fingerprinting that can hierarchically abstract comprehensive heterogeneous traffic features into homogeneous vectors seamlessly digestible by existing neural networks for further classification. The extensive evaluation demonstrates that our framework, with just one paradigm, not only supports heterogeneous traffic input but also achieves better or comparable performance compared to state-of-the-art methods black across a wide range of traffic fingerprinting tasks.

M3 - Conference article published in proceeding or book

BT - 32nd USENIX Security Symposium 2023

ER -