A novel deep-learning based approach to DNS over HTTPS network traffic detection
Open Access
- 1 December 2023
- journal article
- research article
- Published by Institute of Advanced Engineering and Science in International Journal of Electrical and Computer Engineering (IJECE)
- Vol. 13 (6) , 6691-6700
Abstract
Domain name system (DNS) over hypertext transfer protocol secure (HTTPS) (DoH) is currently a new standard for secure communication between DNS servers and end-users. Secure sockets layer (SSL)/transport layer security (TLS) encryption should guarantee the user a high level of privacy regarding the impossibility of data content decryption and protocol identification. Our team created a DoH data set from captured real network traffic and proposed novel deep-learning-based detection models allowing encrypted DoH traffic identification. Our detection models were trained on the network traffic from the Czech top-level domain maintainer, Czech network interchange center (CZ.NIC), and successfully applied to the identification of the DoH traffic from Cloudflare. The reached detection model accuracy was near 95%, and it is clear that the encryption does not prohibit the DoH protocol identification.Keywords
This publication has 20 references indexed in Scilit:
- Detecting Data Exfiltration over Encrypted DNS2022
- Detecting DNS over HTTPS based data exfiltrationComputer Networks, 2022
- Detection of Tunneling in DNS over HTTPS2021
- Evaluation and Optimization of learning-based DNS over HTTPS Traffic Classification2021
- A comprehensive survey on DNS tunnel detectionComputer Networks, 2021
- Feature Engineering and Machine Learning Model Comparison for Malicious Activity Detection in the DNS-Over-HTTPS ProtocolIEEE Access, 2021
- Encrypt DNS Traffic: Automated Feature Learning Method for Detecting DNS Tunnels2021
- Privacy of DNS-over-HTTPS: Requiem for a Dream?2021
- Privacy Illusion: Beware of Unpadded DoH2020
- Detection of DoH Tunnels using Time-series Classification of Encrypted Traffic2020