https://news.ycombinator.com/item?id=12089993 … < On WhatsApp not working in Saudi. Calls are filtered by firewall, so the UI just blocks them rather than timeout.
Steganography to fool humans or smart analysis tools costs a lot in BW. OTOH what I described is just a few % BW cost.
-
-
then I don't understand why can't you sample the mp3 and block ones with high entropy in DCT coeff
-
Roughly the same reason gzipping an mp3 doesn't get you much.
-
Of course more advanced analysis is possible, but there are development and runtime costs, hopefully high.
-
I don't claim this is a fully-developed solution, just a possible framework for how it could be done.
End of conversation
New conversation -
-
-
It should be possible to impose much higher cost on attacker's computing resources than on user's BW resources.
-
but he merely samples you, that's by definition cheap
-
Deciding whether a stream is plausible non-encrypted data should be computationally expensive. That's the goal.
-
but if you deliver real time high BW enc data, you, by definition move lot of high entropy data. adv'd measure entropy
-
I don't understand how can you make it more expensive than measuring entropy, without sacrificing BW.
-
Let's try to get on the same page. What entropy metric do you have in mind?
-
example metric finds keys, http://blog.dkbza.org/2007/05/scanning-data-for-entropy-anomalies.html … Far from expert, will be happy to understand what I'm missing.
-
prof Orr Dunkleman referred me to https://people.csail.mit.edu/rivest/Chaffing.txt … with linear AONT http://cacr.uwaterloo.ca/~dstinson/papers/AON.ps …
- 2 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.