Re:My Experience With the Great Firewall of China
by aexaey on Friday January 15, 2016 @05:23PM
(#10912693)
There is an article[1] about finding out what location at google maps user is looking at, based on observing TLS packets and nothing else.
If task at hand is to match encrypted connection to a smaller number of alternative (say identify accesses to popular place, like gmail/facebook login page, etc), same technique of observing packet size distribution over time should probably work quite reliably, and cgiproxy will do nothing to hide it.
[1] http://www.ioactive.com/pdfs/SSLTrafficAnalysisOnGoogleMaps....
Re:My Experience With the Great Firewall of China
by mike-cardwell on Friday January 15, 2016 @05:39PM
(#10912814)
Interesting. CGIProxy could be easily modified to add random padding to the HTML or headers of responses. Probably a couple of lines of Perl.
Re:My Experience With the Great Firewall of China
by gtank on Friday January 15, 2016 @06:28PM
(#10913092)
There's a lot more to defeating traffic analysis than random padding: http://freehaven.net/anonbib/cache/oakland2012-peekaboo.pdf
Re:My Experience With the Great Firewall of China
by JoshTriplett on Friday January 15, 2016 @06:15PM
(#10913009)
You don't necessarily want random padding; you really want every single packet exactly the same size, potentially with dummy data transmitted to mask data rate as well.
Install that on your own server behind Basic Auth. Then it just looks like normal HTTPS traffic to any sniffer. Because that's exactly what it is.