Instead of padding all packets to be the same size, two solutions come to mind:

It would be relatively easy to pad them with shorter random length strings, so that if the average vbr bit rate was n,
and the fully padded constant bitrate would be m, we could achieve a bit rate of about (n+m)/2,
which is at least better than m..

another option could be to fragment the packets and adjoin them so that they are all about or equal in size..
intertween some creative packets esp. during periods of silence, and that should throw anyone off..

Side channel attacks will always exist, and though scary at first, can usually be engineered around, rather than giving up on the technology..

-lc

On Wed, Jun 11, 2008 at 1:59 PM, jeff sacksteder <jsacksteder@gmail.com> wrote:
I don't know if I'd say that, exactly. This is a similar problem to
anonymous remailers. To prevent traffic analysis, you need to
continuously emit equal sized messages of encrypted noise so that you
deny the bad guys information about the time and length of your
messages hidden within.

You could use VBR codecs but you would have to pad the data stream out
to a constant size, nullifing the point of using VBR. You could use
them if compatibility was a concern, for instance. It would not be of
any bandwidth advantage, however.

> This is very interesting.  That rules out ALL VBR codecs when using
> SRTP for security reasons.

_______________________________________________
Freeswitch-users mailing list
Freeswitch-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org