Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:10 ADT
Ryan Lackey (rdl@MIT.EDU)
07 Jul 1998 12:03:11 -0400
You don't need to shield against radiation, neutrons, etc. You just need to
detect high ambient flux -- neutrons are the main problem for shielding, yes,
and you can buy a NAG glass lithium-free glass scintillator or whatever to
detect a high neutron flux.
Then you either turn off the source of random data to the application and
warn the user or touch off the thermonuclear self destruct device or
whatever else you wish to use.
The timing-differential-between-intervals is the preferred method from
everything I've read (there *are* papers on this subject, but I forget if
they were in crypto journals or nuclear engineering journals -- they're
all fun to read after finals week). Events per bit is an important metric,
but remember, with nukes you can just increase the event rate, either by using
a more energetic source, or by using sources in parallel.
Diodes seem to be used in most actual production systems on the open market,
though. I'm a fan of FIPS 140-1 protected RNGs...the IBM 4758 contains a
nice diode, and is well sealed.
You need to destroy or discontinue your RNG in the event of physical compromise
anyway, so you might as well include sensors for out-of-range environmental
situations, like rad flux, temperature, physical tamepring with the enclosure,
etc.
Such fun. It seems that cryptographers, upon learning that they were
trafficking in munitions, decided to get involved in other kinds of
dual-use technology and munitions as well. Such is the fate of overbroad
regulation.
When cryptography is outlawed, only cryptographers will have energetic
nuclear sources.
-- Ryan Lackey rdl@mit.edu http://sof.mit.edu/rdl/
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:10 ADT