Schneier on Security
A blog covering security and security technology.
« Security Considerations in Prison Food |
| Mathematicians vs. Cryptographers »
September 27, 2007
NASA Using 1960s Cryptanalysis Techniques
Well, sort of.
This paper from the Goddard Space Center, "NiCd Space Battery Test Data Analysis Project, Phase 2 Quarterly Report, 1 Jan. - 30 Apr. 1967," uses "cryptanalytic techniques" -- some sort of tri-gram frequency analysis, I think -- to ferret out hidden clues about battery failures.
It's hard to imagine non-NSA cryptography in the U.S. from the 1960s. Basically, it was all alphabetic stuff. Even rotor machines were highly classified, and absolutely nothing was being done in binary.
Posted on September 27, 2007 at 6:14 AM
• 9 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
Having had a quick look through the paper I think the refrence to cryptography is a bit thin at best.
Basically they have analysed the batteries for Two-Fault and Three-Fault corelations. In effect they have shown that faults are related to each other in the various manufacturing processes. That is a particular manufacture has defect charecteristics peculier to their manufacturing proces (a not unexpected outcome).
Basically quite simple statistical analysis of a moderatly complex data set. The reference to cryptography is a bit of wishfull thinking at best, possibly to "sex-up" an otherwise dull paper.
"It's hard to imagine non-NSA cryptography from the 1960s...."
There was plenty of commercial and other non-NSA cryptography -- computer-based and otherwise -- from the 1960s. The NSA did make sure it could break any widely published methods. (Some claim that is still the case, despite recent "open" reviews of cryptographic algorithms.)
Non-NSA cryptanalysis from the 1960s was lagging however.
Actually, this is a rather cute idea.
On one hand, it is just a statistical analysis. Breaking most trivial coding reverts to analysis. Instead of starting from scratch and working through each possible source for underlying data, the researchers just pointed this large cannon of cryptography tools at the problem and pulled the trigger.
Given Google's purely statistical translation work, I wonder what problems could be solved by just assuming that language translation is just an encyption operation.
"However, sharing their cryptanalysis techniques with NASA was probably not one of those things. "
Ah have a look at the work of "Clifford Cocks" and "non secret encryption" that predated RSA by some time. It is the most public asspect of the UK USA (BRUSA) sharing agreements. Further info from,
GCHQ (as was) actually passed it over to the NSA who apparently poo pooed it as to complex to be practical...
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.