[Cryptography] 'The intelligence coup of the century'

[Henry Baker]

At 10:23 AM 2/14/2020, John Young wrote:
>What's the current protection for compromising radio signals?

I don't know about TEMPEST-like stuff, but most digital
devices today must pass an FCC test to make sure that they
don't interfere with intentional radio communications signals.

One of the main offenders of modern clocked logic is the
clock frequency itself (and its harmonics).  One way to
reduce the maximum power in these frequencies is to utilize
"spread spectrum clocking" which mixes in a pseudo-random
code to "spread" the power around the spectrum instead of
having high peaks on these harmonics.

These spread-spectrum clocking pseudo-random codes aren't
very long, and aren't designed with cryptography in mind,
but you could easily build your own pseudo-random code
clock circuitry which could be programmed on-the-fly just
like chips today control their own clock frequencies in an
extremely crude way to control power consumption.

There are standing FCC rules which forbid the use of
extremely long & crypto-grade codes for spread spectrum
communications, but if these codes were to be used for
clocking an otherwise normal CPU, I'm having trouble
understanding how anyone would know -- unless perhaps
the FCC/NSA has a comprehensive catalog of all existing
spread spectrum clock codes, and yours pops up as not
being in that catalog.

Oh, and yes, these modulated clocks could be used to
exfiltrate data quite long distances, since long
spread spectrum codes can be used to hide the data
many dB's below the noise floor.  E.g. battery-powered
spread spectrum radios which can run for *10 years*
and which live in water meters -- i.e., below ground --
and can communicate a number of miles, were built in
the 1990's, and I presume are even better today.

(Hint: NASA JPL uses long spread spectrum codes like
these -- although not crypto-grade -- to talk to
probes that are beyond the orbit of Pluto.)