[E-voting] [a.oostveen@rathenau.nl: BREAKING NEWS]

Justin Mason jm at jmason.org
Thu Nov 2 12:22:07 GMT 2006


Colm MacCarthaigh writes:
> On Thu, Nov 02, 2006 at 11:17:29AM +0000, Michael McMahon wrote:
> > Colm MacCarthaigh wrote:
> > >On Wed, Nov 01, 2006 at 06:36:53PM -0500, Aengus Lawlor wrote:
> > >  
> > >>  The Irish Times reports that electronic voting machines bought by
> > >>  the Irish government are not in danger of being bugged;
> > >
> > >http://www.youtube.com/watch?v=B05wPomCjEY
> > >  
> > Very good demo. I haven't got the Irish Times to hand. Do they offer
> > any scepticism about this "claim" that our machines are not in danger?
> 
> Nope, they just accept what the Dutch testing institute are saying. It
> is a bit bizarre what they are saying, since it's so demonstrably
> possible. The disparity seems to arise over differing tolerances for the
> signal levels. But as the demo shows, it works through walls and up to
> 25 metres away.
> 
> > Like has anyone tested this? Afaik, neither the CEV, nor the govt.s test
> > did.
> 
> TNO did some testing of the machines succeptability to Electro-magnetic
> interference. testing carried out by TCD on behalf of the CEV also
> exposed to machine to an EMF of 7 Tesla. But afaik, noone did anything
> like what we're looking at now.

Those two tests both sound like tests of how the device reacts to external
EMI.  TEMPEST testing is about the opposite -- how much EMI/RFI the device
*emits*.

There is another type of common EMI/RFI testing, typically performed on
consumer electronic equipment for a CE mark or the like, that looks for
levels of EMI leakage *from* the device -- that leakage can cause
interference with radios, tvs, etc.  However even if that testing took
place, I don't think it involves looking for TEMPEST-style security issues
(e.g. patterns in EMI matching display patterns etc.)

http://www.maxim-ic.com/appnotes.cfm/appnote_number/1167/ has some more
background on non-TEMPEST EMI/RFI testing.

By the way, from what I know about TEMPEST and EMI, it sounds like it
might be possible to guess where TEMPEST problems are likely to crop up,
based on the location of unshielded cables and the data that travels on
those cables.  For example, in the Dutch case, it's the wires leading from
the main board to the display LCD that emits the signals that can be
snooped, iirc.  But I guess actively attacking the Irish system a la the
Dutch researchers would be quite a big project, overall, really ;)

--j.



More information about the E-voting mailing list