[3710] in linux-net channel archive
Re: Encryption
daemon@ATHENA.MIT.EDU (Alan Cox)
Tue Jul 16 00:04:55 1996
From: Alan Cox <alan@cymru.net>
To: kmb203@psu.edu (Kevin M Bealer)
Date: Mon, 15 Jul 1996 09:32:11 +0100 (BST)
Cc: linux-net@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.93.960714002211.10453C-100000@brando> from "Kevin M Bealer" at Jul 14, 96 00:35:04 am
> transmit encrypted data in the "magic number" of a TCP packet? This is
> supposed to be a random number in all regards, but I was thinking if you had
> completely random-looking encrypted data you could slip it through
> undetected. Since these must be present in every packet, and must be
> preserved to detect network loops (am I right about this?), not only could
> you not tell what the information was, it would be impossible to detect
> _whether_ it was being done.
A tcp packet has no magic number. You can however pass data across the top
of TCP (and almost any data carrying interface) in concealed fashion. A
whole variety of these problems exist at both host (eg messages by varying
disk usage) and networking (for example perturbing the retransmit patterns
of a connection). Both are very unreliable but an error correcting code can
be placed over the top (eg Viterbi or Golay codes).
Alan
