secure fax (Re: email Crypto- third party)

[Pete Bentley]

At Tue, 03 Aug 1999 14:39:30 BST, Tom.A.Parker@icl.com writes:
>I had occasion to try using transparencies to demonstrate one time pad
>principles in a crypto lecture I gave some time ago. The aim was to use
>overlays to show that choosing two different keys caused two different
>apparent messages. However you can't do a straight overlay because it
>doesn't give an XOR. 
>
>Black + Black = Black 
>Black + White = Black as well.
>
>To solve this problem I had a rather complicated design of concentric black
>and white circles. It worked, sort of, but I would be interested in more
>details of the fax approach, which on the face of it doesn't seem to work.

Of course, if you're using fax modems connected to computers rather
than paper fax machines, then the fax is represented as a bitmap in
memory and you *can* perform XOR operations on it.  Or indeed any
other encryption operation that yields a bitmap of the same size. 

This is, of course, harder to demonstrate in a lecture but doable
nonetheless ("See bitmap A, this is the encrypted fax.  See bitmap B,
my overlay. See bitmap C which you will have to take on trust is the
result of A XOR B and which is the legible fax").

Presumably though, faxes encrypted in this manner will look like
random bit patterns.  In turn, those bit patterns will not compress
well using the Huffman coding scheme that fax machines use, which will
make them stand out to anyone performing large scale telecomms
analysis (such as our good friends at ECHELON)

Pete.