18 March 1998


To: cypherpunks@toad.com
Subject: Re: EMI, Van Eck, etc. 
Date: Wed, 18 Mar 1998 13:09:10 +0000
From: Markus Kuhn <Markus.Kuhn@cl.cam.ac.uk>


The Spectre wrote on 1998-03-18 05:11 UTC:
> In almost every writing I've come across regarding Van Eck, I notice the
> phrase "...simply a modified television" or something along those lines.
> 
> Does anyone have a document for actually modifying a television set to do
> this sort of thing?  It doesn't have to be extremely long ranged, and could
> in fact be very short range.. I am interested in performing my own
> experiments into defeating this sort of eavesdropping.

Ingredients for a minimum cost quick & dirty TEMPEST experiment:

1 RF tuner of a VCR
1 antenna amplifier
1 antenna
1 multisync PC monitor
1 PC with a video card (or a pair of tuneable sync oscillators)

Connect the PC with the video card to the SYNC inputs of the multisync
monitor. Program the video card to a video mode with the same deflection
frequencies as that used by the target system. Connect the baseband output
of your tuner to the VIDEO-IN pins of your monitor. Connect the antenna
and amplifier to the RF input of your tuner. Switch on. Fill the
screen of the target device with a big symbol consisting of dithered and
non-dithered areas for best results in the first trials. Now tune
through the VHF bands starting with the dot clock frequency of the
target.

That's it basically. Such a primitive TEMPEST monitor is of course
unsuitable for evaluating the threat from much more sophisticated wide-band
DSP eavesdropping receivers that directly attempt OCR-style algorithms
on the signal with matched filters. But it is fun to play around with, it
is useful for getting a feeling for the effect, and it is suitable for
demonstrating most of the Soft Tempest tricks that I described in
<http://www.cl.cam.ac.uk/~mgk25/ih98-tempest.pdf>.

> Also, would it be possible to scramble the signal into an unusable level by
> simply putting another device emanating RF at the snooping frequencies
> nearby the machine that you want to protect?  Something generating white
> noise at that frequency, but with a purposely built antenna, say a high
> gain type turned outward from the monitor, with a significantly higher
> power output than the monitor?

The FCC and your radiologist advise against this. Shielding is much
more elegant than jamming. Remember that CRT content is a periodic
signal, thus you can suppress uncorrelated noise by periodic averaging
rather easily. Good jamming must produce a correlated output signal.
See United States Patents 5165098 and 5297201 for descriptions of
correlated jammers. I don't think, these are widely used though, as the
TEMPEST standards seem to mandate shielding and not jamming, which I
think is very sensible.

Markus

-- 
Markus G. Kuhn, Security Group, Computer Lab, Cambridge University, UK
email: mkuhn at acm.org,  home page: <http://www.cl.cam.ac.uk/~mgk25/>


JYA Note: TEMPEST information: http://www.eskimo.com/~joelm/tempest.html