More about HKUST
Website Fingerprinting Attacks and Defenses
PhD Qualifying Examination
Title: "Website Fingerprinting Attacks and Defenses"
by
Mr. Jiajun GONG
Abstract:
Website fingerprinting (WF) attacks pose a great threat to user privacy on
anonymity networks. It is a kind of traffic analysis attack, where a local
attack passively eavesdrops in network traffic to find out which webpage a
client is visiting. They succeeds in doing so by extracting features from
the collected traffic such as the number of outgoing and incoming packets,
packet timing and the ordering of packets. They can be used by network
surveillants to break the privacy guarantee of those anonymity networks.
What makes WF attacks even threatening is that the local passive
eavesdropper is virtually impossible to detect. A lot of WF attacks based
on various machine learning or deep learning models have been proposed.
They proves to be extremely effective against current anonymity networks.
To mitigate the threat of WF attacks, a lot of defenses have been put
forward. They can be generally categorized into three classes: obfuscation
defense that obfuscates specific features WF attacks rely on, confusion
defense that makes it difficult to determine which of a given set of given
packet sequences is loaded and regularization defense that restricts how
clients send and receive packets to limit the features useful to
attackers. However, they are either too expensive to deploy in reality or
not strong enough to defeat all the attacks. Cheap while effective
defenses are still urgently wanted.
In this survey, we investigate existing WF attacks and defenses in depth.
We analyse their strengths and weaknesses. We believe our survey will shed
light to future work on Website Fingerprinting.
Date: Wednesday, 8 May 2019
Time: 3:00pm - 5:00pm
Venue: Room 4472
Lifts 25/26
Committee Members: Dr. Tao Wang (Supervisor)
Prof. Shing-Chi Cheung (Chairperson)
Prof. Cunsheng Ding
Dr. Raymond Wong
**** ALL are Welcome ****