Packet-level labeling method for fine-grained multi-webpage browsing behavior recognition

To address the challenge of fine-grained multi-webpage browsing behavior recognition in mixed encrypted traffic under concurrent multi-webpage access, a packet-level labeling method based on inter-packet temporal correlation features, called PLL, was proposed. This method combined one-dimensional co...

Full description

Saved in:
Bibliographic Details
Main Authors: GU Yue, CHEN Li, LI Dan, GAO Kaihui
Format: Article
Language:zho
Published: Editorial Department of Journal on Communications 2025-07-01
Series:Tongxin xuebao
Subjects:
Online Access:http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2025122/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:To address the challenge of fine-grained multi-webpage browsing behavior recognition in mixed encrypted traffic under concurrent multi-webpage access, a packet-level labeling method based on inter-packet temporal correlation features, called PLL, was proposed. This method combined one-dimensional convolutional neural networks with multi-head attention mechanisms to learn both local and global temporal correlation features between different packets within the same webpage. Additionally, one-dimensional transposed convolution was employed to restore features to the original packet sequence length, thereby establishing a precise correspondence between each packet and its contextual features. This enabled accurate packet-level labeling and further supported the inference of user’s browsing time information for multiple webpage, such as the start time, end time, and duration of each webpage visit. Experimental results show that PLL achieves 98% and 97% accuracy in identifying visit start time and duration, effectively solving the critical issue of multi-webpage browsing behavior recognition in mixed encrypted traffic.
ISSN:1000-436X