From: rarachel@prism.poly.edu (Arsen Ray Arachelian)
Date: Wed, 23 Mar 94 13:50:21 PST
To: a-ophirr@microsoft.com (Ophir Ronen)
Subject: Re: A possibility...
In-Reply-To: <9403232005.AA26190@netmail2.microsoft.com>
Message-ID: <9403232137.AA17936@prism.poly.edu>
MIME-Version: 1.0
Content-Type: text


> I have an idea that may or may not be feasible so I am offering it for
> scrutiny.

Congrats, you've tripped on an idea of mine, which I've already implemented.
:-)  Actually, I've done a lot more than just encrypt the data with noise
and mixed the two, as that would be insecure once you know the algorithm...

If you guys are interested, I'll post up a "alpha-development" version of the
paper I wrote describing this thing.  This btw is my project for the crypt-
analysis class I'm taking...  On the surface it seems pretty secure... There
was a hole I plugged up, but that's another issue...

I did have a working version of the software, however a rather uncool program
decided to eat my hard drive for breakfast, and unfortunatly I lost it.  THe
last two backups are of a previous (unplugged) version and of a half-way
inbetween non-working version...  Now I've been busy working on a SecureDrive
type program for the Mac so I haven't payed much attention to this, however
it has some very interesting possibilities including the ability to be used
in stego in a verry cool way...
 
> What I am thinking is to have one person encrypt using a RNG (noise). He
> encrypts his message using the noise on one channel and then copies the noise
> utilized on another channel. He then mixes the two channels to 
> create....noise.

Ditto, only I used several channels, not just two...

> On the other side the person must be expecting the noise and must know how to
> seperate the signal. It seems that this leaves things open by having the
> encryption key along with the message, but the message itself is just noise.
> (i.e. the interloper must know of the signal, and know how to split it into
> it's two components.)

That's how mine works... it works by taking in the noise and a passkey. The
passkey determines which bits are noise and which are data, and more importantly
how to decrypt the data...

If you take the random noise data you feed in, and replace it with the low bits
of a picture, you've got a pretty good stego program...