[ntp:questions] NTP architecture recommendation

jlevine jlevine at boulder.nist.gov
Sun Aug 19 15:59:59 UTC 2007

   The degradations of "Selective Availability" included a number of
different possible effects, but the one that was generally used was a
relatively slow dither of the clock on the satellite. The dither on
each satellite was different. Since GPS receivers determine a position
by measuring the distance (signal transit time) from each satellite in
view to the receiver, the dither produced a corresponding fluctuation
in the position solution. The magnitude of the dither was much smaller
than 1 microsecond, so that it would generally not be observable with
the usual NTP setup. The corresponding errors in position were less
than hundreds of meters, but the exact value varied. As with many
other things, your mileage would vary. It was  easy to see the effect
on the position solutions of any GPS receiver, and the effect on the
time solution could be seen with a reasonably good oscillator -- even
a good quartz oscillator could see the effect over periods of seconds
or minutes.
    Since the clock dithering had a mean of 0 in the long term, the
long-term average solution (either position or time) had no offset.
The degradation was primarily directed towards real-time applications
or those timing applications where the local clock had such poor
stability that it was impossible to average the received signal for
any appreciable time so as to take advantage of the fact that the long-
term mean was 0.
    The whole business was turned off in 2000, and it will probably
never be used again, but you never know.

Best wishes,

Judah Levine
Time and Frequency Division
NIST Boulder

More information about the questions mailing list