Statistical Privacy for Streaming Traffic

Abstract

Machine learning empowers traffic-analysis attacks that breach users' privacy from their encrypted traffic. Recent advances in deep learning drastically escalate such threats. One prominent example demonstrated recently is a traffic-analysis attack against video streaming by using convolutional neural networks. In this paper, we explore the adaption of techniques previously used in the domains of adversarial machine learning and differential privacy to mitigate the machine-learning-powered analysis of streaming traffic. Our findings are twofold. First, constructing adversarial samples effectively confounds an adversary with a predetermined classifier but is less effective when the adversary can adapt to the defense by using alternative classifiers or training the classifier with adversarial samples. Second, differential-privacy guarantees are very effective against such statistical-inference-based traffic analysis, while remaining agnostic to the machine learning classifiers used by the adversary. We propose two mechanisms for enforcing differential privacy for encrypted streaming traffic, and evaluate their security and utility. Our empirical implementation and evaluation suggest that the proposed statistical privacy approaches are promising solutions in the underlying scenarios.

GuardSpark++: Fine-Grained Purpose-Aware Access Control for Secure Data Sharing and Analysis in Spark

“…GuardSpark++ is an access control mechanism, not a silver bullet that addresses all attacks. Conventional attacks, such as DoS [55], collusion attacks [77], or traffic analysis [75], will still work even if GuardSpark++ is in place. In particular, if the Spark platform is untrusted or compromised, GuardSpark++ becomes ineffective.…”

MemGuard

“…Such unique constraints require substantially 5 https://bit.ly/2IDchsx different methods to find adversarial confidence score vectors. Other studies have leveraged adversarial examples to defend against traffic analysis [71] and author identification [38,54]. However, these studies did not consider formal utility-loss guarantees.…”

Efficient yet Robust Privacy Preservation for MPEG-DASH-Based Video Streaming

“…Video streaming is a computationally expensive process, and if a request pattern is obfuscated too much, it will cause video lag or video buffering because not enough data is being sent or, conversely, because excess data is being downloaded. To the best of our knowledge at the time of this writing, only one defense strategy has been proposed, by Zhang et al [7]. It leverages differential privacy and works by setting a proxy between the client and server in the form of a browser extension.…”

“…Differential privacy adds noise to data; in the case of video streaming, this noise changes the time intervals of the requests from the client and the amount of data requested by the client. e defense scheme proposed by Zhang et al [7] leverages two differential privacy methods for obfuscation, d * − privacy and FPA k .…”

“…To stop the bit-rate streaming pattern from being able to be identified by an adversary, the streaming pattern must be obfuscated. To the best of our knowledge, the only defense algorithm was proposed by Zhang et al [7]. is paper focused on defending against the CNN attack model mentioned above [2].…”