Re: Nuclear Weapons Material

["Ian Farquhar" <ianf@simple.sydney.sgi.com>]

On Aug 25, 11:43pm, Phil Karn wrote:
> Just to bring this back somewhat to cryptography, an interesting topic
> for speculation is the operation of the "permissive action links"
> (PALs) that control these weapons. The complexity of the procedure
> suggests that the precise timing of many events is crucial if a
> high-yield nuclear explosion is to result. This is particularly true
> for the timing of the many HE detonators, the neutron generator and
> the fusion boost injector.  Perhaps these parameters are stored in
> encrypted form in the weapon and can be decrypted for use only with
> the proper externally-provided key? Considering that a brute force key
> search would consume one weapon per trial key, perhaps this technique
> isn't too bad against dictionary attacks? :-)

I heard a rumor (from several independent sources) which indicated that the
firing sequences are essentially encrypted detonator timings that are passed
through the PAL, which decrypts it but makes no value judgement about the
timings themselves. If the timings are wrong, you get a messy squib explosion
which will make a mess for about 100m around the detonation site, and which
will totally destroy the weapon beyond any hope of recovery.

Whether this true is anyone's guess, and there is a lot of quite deliberate
disinformation concerning nuclear weaponary.  Considering that one known
fact is that the original fatboy contained 64 detonators, and that we'd be
talking about timing in hundreds of microseconds, a back of the envelope
calculation indicates that the amount of timing information would be ~900 bits
for a similar device.  If this keyspace is indeed heavily permuted, so that no
intelligent judgements could be made which would reduce the searchable
keyspace, this seems to be rather secure.  Obvious layers of further protection
(eg. adding a counter which will restrict the lifetime of a particular
firing sequence) are also possible.

							Ian.