Blind Adversarial Network Perturbations

TL;DR

This paper demonstrates that adversarial perturbations can effectively deceive deep neural network-based traffic analysis methods, highlighting a new vulnerability in these models.

Contribution

It introduces the concept of applying adversarial perturbations to live network traffic to defeat DNN-based traffic analysis techniques, a novel approach in this domain.

Findings

Adversarial perturbations can mislead traffic analysis DNNs

Live traffic can be manipulated with small perturbations

Vulnerability of DNN traffic analysis to adversarial attacks

Abstract

Deep Neural Networks (DNNs) are commonly used for various traffic analysis problems, such as website fingerprinting and flow correlation, as they outperform traditional (e.g., statistical) techniques by large margins. However, deep neural networks are known to be vulnerable to adversarial examples: adversarial inputs to the model that get labeled incorrectly by the model due to small adversarial perturbations. In this paper, for the first time, we show that an adversary can defeat DNN-based traffic analysis techniques by applying \emph{adversarial perturbations} on the patterns of \emph{live} network traffic.