TY - GEN
T1 - I Know what you Watch
T2 - 2025 IEEE 10th International Conference on Data Science in Cyberspace, DSC 2025
AU - Liu, Likun
AU - Li, Jingwei
AU - Ge, Mengmeng
AU - Hu, Zhichao
AU - Yu, Xiangzhan
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - Based on the current era of video streams, improving the ability to identify encrypted video streams can effectively suppress the transmission of bad video streams and create a healthy network environment. This paper proposes a method to find video titles in encrypted HTTP adaptive video streaming. By analyzing the content leakage of the video platform built based on the DASH protocol and VBR, the inherent burst timing features of the video stream of this type of platform can be extracted. 10,000 pieces of video streams are collected from the Youtube platform through automated data collection. The LightGBM model is used to train burst time series features, and conduct five-fold crossvalidation. 20 % of the dataset is used as the test set, and the accuracy of the video stream is 98.77 %. Therefore, based on this method, the title of the video stream can be effectively identified.
AB - Based on the current era of video streams, improving the ability to identify encrypted video streams can effectively suppress the transmission of bad video streams and create a healthy network environment. This paper proposes a method to find video titles in encrypted HTTP adaptive video streaming. By analyzing the content leakage of the video platform built based on the DASH protocol and VBR, the inherent burst timing features of the video stream of this type of platform can be extracted. 10,000 pieces of video streams are collected from the Youtube platform through automated data collection. The LightGBM model is used to train burst time series features, and conduct five-fold crossvalidation. 20 % of the dataset is used as the test set, and the accuracy of the video stream is 98.77 %. Therefore, based on this method, the title of the video stream can be effectively identified.
KW - Burst feature
KW - DASH protocol
KW - Encrypted traffic classification
KW - LightGBM model
UR - https://www.scopus.com/pages/publications/105033018541
U2 - 10.1109/DSC67331.2025.00033
DO - 10.1109/DSC67331.2025.00033
M3 - 会议稿件
AN - SCOPUS:105033018541
T3 - Proceedings - 2025 IEEE 10th International Conference on Data Science in Cyberspace, DSC 2025
SP - 191
EP - 198
BT - Proceedings - 2025 IEEE 10th International Conference on Data Science in Cyberspace, DSC 2025
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 15 August 2025 through 17 August 2025
ER -