Re: TEMPEST Question

>> Cpu signal is the easiest to detect, but not the only
	>> one.  Other possible sources of emissions are all
	>> kinds of communications cables.  Unshielded RS-232 to
	>> the modem, possibly even the connections to the disk
	>> drives.

Synching up to monitor signals, I can understand.

But as a non-technical person, what I'm struggling to understand is how
a surveillance team could monitor the emmisions from such cables and
have any clue as to what they are. Let's say they zeroed in on my
poorly shielded modem cable and were able to tune into a stream of 0's
and 1's. How could they then resolve that digital data into anything
meaningful? It could be one of any number of documents created by one
of any number of programs on one of any number of platforms. How do the
spies know they're dealing with a Frame page layout document created on
a Sun workstation versus a spreadsheet created on a Mac?

Even if it's just a plain text file how could the surveillance team
read it? Does each member of the ASCII character set have specific and
identifiable radiation signatures? For example, does the letter "k" as
it passes through my modem cable have a characteristic EMR that is the
same for all machines?

Sorry if this query is too basic, but I would appreciate any enlightenment...

Doug

>>>>> Doug.Brightwell@corp.sun.com (Doug Brightwell) writes:

	Doug> But as a non-technical person, what I'm struggling to
	Doug> understand is how a surveillance team could monitor the
	Doug> emmisions from such cables and have any clue as to what
	Doug> they are. Let's say they zeroed in on my poorly shielded
	Doug> modem cable and were able to tune into a stream of 0's
	Doug> and 1's. How could they then resolve that digital data
	Doug> into anything meaningful? ...

My understanding, being only a moderate RF weenie, is that UARTs, the
devices which, more often than not, are driving your serial ports,
generate an emission very much like good old frequency shift keying.
It makes sense - having listened to computers on radios before. Think
in terms of sound first, as it's easier. A given sound represents a
"1", and another tone is "0". You waver back and forth between the
tones to describe the data you're transmitting. 

The idea is that some of the normal emission characteristics from
components in computers (like UARTs) correspond to this sort of
modulation. 

With regard to machine types, if it's a serial line, it's fairly easy
to map the communications into a set of probably protocol suites. If
you know *anything* about the use of the machine, it becomes a lot
easier. For instance, given an intercept of TCP/IP traffic over a SLIP
line, I could probably reconstruct a TELNET session log, but it's
nowhere as easy as just reading a terminal session on a link to some
BBS.

	Doug> Even if it's just a plain text file how could the
	Doug> surveillance team read it? Does each member of the ASCII
	Doug> character set have specific and identifiable radiation
	Doug> signatures? For example, does the letter "k" as it
	Doug> passes through my modem cable have a characteristic EMR
	Doug> that is the same for all machines?

You might also think in terms of things other than screens and modem
cables. For instance, many keyboards emit RF as they are used. It's
not irrational to suspect that these emissions might be tied in some
way to the use of the keyboard - I have an old FCC Class A Wyse
terminal that sounds different depending on which keys I hit. Since we
know that every piece of equipment has a uniquely identifiable (even
compared to other units of the same type) emission signature, it's not
too outlandish to expect that different key encodings in a keyboard
might generate different emission patterns.

	Doug> Sorry if this query is too basic, but I would appreciate
	Doug> any enlightenment...

It's weird stuff, and some people would rather not have the world
learning how to do this. In fact, Van Eck's original article is known
to have some deliberate misinformation in it, as the author didn't
want to make it "too easy" to learn how to do this kind of ELINT.


--Strat