arrow
Return

MTL-DoHTA: Multi-Task Learning-Based DNS over HTTPS Traffic Analysis for Enhanced Network Security

delete2025-02-07
delete0
delete
OA
AI
J
Jung, Woong Kyo
K
Kwak, Byung Il *
DOI:10.3390/s25040993delete
deleteOriginal
deleteShare
deleteSave
View PDF
Abstract

Abstract

En 中文
The adoption of DNS over HTTPS (DoH) has significantly enhanced user privacy and security by encrypting DNS queries. However, it also presents new challenges for detecting malicious activities, such as DNS tunneling, within encrypted traffic. In this study, we propose MTL-DoHTA, a multi-task learning-based framework designed to analyze DoH traffic and classify it into three tasks: (1) DoH vs. non-DoH traffic, (2) benign vs. malicious DoH traffic, and (3) the identification of DNS tunneling tools (e.g., dns2tcp, dnscat2, iodine). Leveraging statistical features derived from network traffic and a 2D-CNN architecture enhanced with GradNorm and attention mechanisms, MTL-DoHTA achieves a macro-averaging F1-score of 0.9905 on the CIRA-CIC-DoHBrw-2020 dataset. Furthermore, the model effectively handles class imbalance and mitigates overfitting using downsampling techniques while maintaining high classification performance. The proposed framework can serve as a reliable tool for monitoring and securing sensor-based network systems against sophisticated threats, while also demonstrating its potential to enhance multi-tasking capabilities in resource-constrained sensor environments.
Keywords:
DNS over HTTPS
DNS covert channel
multi-task learning
deep learning
AI Summary

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.

Journal

Sensors cover
Sensors
IF:
3.5
Papers:
7.2W
Citations:
20.9W

Organization

H
Hallym University
Scholars:
7.9K
Papers: 7.4K
Citations: 4.0K
Cited Papers

Cited Papers

errShare
errSave
Encrypted and covert DNS queries for botnets: Challenges and countermeasures
err2020-01-01
err34
errOAAI
errPatsakis, Constantinos; Casino, Fran; Katos, Vasilios
errShare
errSave
MASiNet: Network Intrusion Detection for IoT Security Based on Meta-Learning Framework
err2024-07-15
err2
PREAI
errWu, Yiming; Lin, Gaoyun; Liu, Lisong; Hong, Zhen; Wang, Yangyang; Yang, Xing; Jiang, Zoe L.; Ji, Shouling; Wen, Zhenyu
errShare
errSave
Chemical profile and pharmacological effects of the resin and essential oil from Bursera slechtendalii: A medicinal “copal tree” of southern Mexico
err2018-07-01
err0
PREAI
errNemesio Villa-Ruano; Yesenia Pacheco-Hernández; Elvia Becerra-Martínez; José Alejo Zárate-Reyes; Ramiro Cruz-Durán
errShare
errSave
Visualizing Realistic Benchmarked IDS Dataset: CIRA-CIC-DoHBrw-2020
err2022-01-01
err7
errOAAI
errYusof, Mohammad Hafiz Mohd; Almohammedi, Akram A.; Shepelev, Vladimir; Ahmed, Osman
errShare
errSave
researcher View more