[ntp:questions] Re: Fingerprinting hosts by clock skew

John Ackermann N8UR jra at febo.com
Sat Mar 5 12:06:30 UTC 2005

Mxsmanic wrote:

>David L. Mills writes:
>>I should add to that proposal using NTP frequency to 
>>reveal outdoor temperature if the IT staff would put the computer 
>>outside the building in a doghouse on the roof.
>That's like saying that you can measure accelerations in a fighter jet
>by leaving the flight computers unattached and measuring the error rates
>they experience as they bang around the fuselage.  It's much easier to
>put in accelerometers that the computers can read.  And in this case it
>might be a lot more practical to just put a thermometer outside and keep
>the computer indoors, even though it would take NTP out of the
>weather-prediction loop.

I'm guessing Dave was joking in his comments, but the frequency vs. temp 
curve actually can provide a very accurate thermometer.  HP manufactured 
one along those lines, with a crystal and oscillator contained in the 
sensor and a frequency meter for the display, that had something like 
0.001 degree C resolution.  And, those of us who play with frequency 
standards often see temperature related fluctuations even in units that 
have ovens or double ovens.

Another example is an antenna that I use to receive Loran-C 
transmissions for frequency monitoring.  The antenna is up on my roof 
and contains an FET preamplifier.  I've seen an 80 nanosecond diurnal 
phase shift as the antenna warms up and cools down each day.


More information about the questions mailing list