Defense Against Traffic Analysis
Implementing an option that makes traffic analysis, done by AI or otherwise, more difficult. Similar to what Mullvad has created with DAITA

-
Waho App commented
NICE
-
Gemini commented
Doesnt stealth mode do that
-
Jerry Gen commented
I am also looking for this.
-
pupilcrow commented
Obscura VPN, Tor, and Nym have also done research on this subject:
https://obscura.net/#obfuscation
https://www.privacyguides.org/en/advanced/tor-overview/#tor-usage-is-not-undetectable
https://nym.com/mixnet
I think Proton could take some inspiration from them! -
Nuesha Brunelli commented
Information link, Karlstad University, about DAITA:
"Researcher Tobias Pulls from the Department of Computer Science at Karlstad University, together with the Gothenburg-based company Mullvad VPN, has developed a new protection named DAITA ("Defense against AI-guided Traffic Analysis").
The protection against the rapid AI development in traffic analysis is available in Mullvad's VPN app on various platforms throughout the year, which has over a million downloads worldwide."
-
Nuesha Brunelli commented
I use DAITA too, I see no drastic drop in speed. It might depend of which server you are using to connect and of how many features you activate in the app or even which DNS is used.
-
Bird commented
Just sharing my experience. I have been testing DAITA Mullvad feature and on my country the drop speed was around 70%.
I am not against implementing it, but think it twice on how implement this feature as it doesn't seem easy and efficient -
sesamee commented
Absolutely, i'd argue this is one of the most crucial and pertinent features that should be added, i've been researching this specific attack and there's a few of papers showing that traffic-analysis based website fingerprinting is actually very practical, there's been one conducted on the tor network recently with AI, and they had about a 95% accuracy rating of being able to identify which websites or services a user was accessing.
Largely this has to do with identifying unique patterns in traffic flow, using packet timings, packet lengths, and packet spikes to generate a unique profile for each website, and then analyzing the traffic of the victim, comparing those profiles to their traffic.
While defenses against this type of attack are not avaliable on protonVPN, one thing you can do to protect yourself against such an attack in the meantime is running internet radio and twitch livestreams in the background.
This will add significant noise to the traffic and make it substantially more difficult for someone to conduct such an attack.
But protonVPN should absolutely add in a defense against this, mullvad implements DAITA essentially by 1. only transmitting packets within a certain packet size range, and 2. adding noise to the traffic, so it should not be too hard for proton to add in such a feature.
-
Proton User commented
Read more about DAITA here:
https://mullvad.net/en/vpn/daita