Open access
Date
2023Type
- Master Thesis
ETH Bibliography
yes
Altmetrics
Abstract
In the realm of network security, machine learning has emerged as a crucial instrument. Unfortunately, the amount of available training data for network traffic-based models is very limited. This study aimed to bridge this gap. A tool was developed to autonomously gather vast amounts of website network traces. It did so by collecting a large amount of website addresses. Upon accessing these sites, the tool captured the resulting encrypted network traffic. This methodology enabled the collection of 900’000 distinct network traces spanning 220’000 unique website addresses. The data collection process was executed across eight geographically dispersed virtual machines, underscoring the comprehensive and diverse nature of the dataset. The effectiveness of the generated dataset was established by evaluating different Deep Fingerprinting attack scenarios. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000632737Publication status
publishedPublisher
ETH ZurichOrganisational unit
09477 - Vanbever, Laurent / Vanbever, Laurent
More
Show all metadata
ETH Bibliography
yes
Altmetrics