@article{MFCDE30ED,
title = "Application Traffic Classification through Non-Encrypted Data Learning",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2025",
issn = "1226-4717",
doi = "10.7840/kics.2025.50.2.304",
author = "Ju-Sung Kim, Yoon-Seong Jang, Ui-Jun Baek, Myung-Sup Kim",
keywords = "Application Traffic, Traffic Classification, Deep Learning",
abstract = "With the growth of the internet, various application traffic types have emerged, and the importance of information security is increasingly emphasized. Consequently, the use of encrypted communication has risen, heightening the need for effective network traffic classification in such environments. However, traditional network traffic classification techniques have shown limitations in handling encrypted data and complex patterns, leading to the emergence of new deep learning-based classification methods. Despite this, deep learning-based classification methods also face challenges in environments with encrypted data or high levels of noise, resulting in lower performance. To address these issues, this paper proposes a methodology that classifies traffic data based on TLS header values to distinguish between encrypted and unencrypted data and utilizes only unencrypted data as training input for the deep learning model. The experiment was conducted using the proposed methodology, resulting in an average accuracy that was 9 percentage points higher than that of the general preprocessing method."
}