[ntp:questions] Using jitter as an indicator of the expecting time

B bertil84 at gmail.com
Wed Nov 4 09:45:42 UTC 2009


Hi, This is a really interesting question for me.

I know about the error bounds, but never heard of using the jitter as
an indicator of the expecting time relative offset. While there aren't
any certain distribution within the error bound, an indicator of
expecting time is important. The jitter isn't introduced before NTPv4
and we are using NTPv3(RFC-1305).

Here is a good thread were David L. Mills wrote about using an
indicator for the expecting time(NTPv4):
http://groups.google.com/group/comp.protocols.time.ntp/browse_thread/thread/e56b7b499b6defc8/870223aa124b5117?lnk=gst&q=by+design+jitter+indicator#870223aa124b5117

There are two statistics involved: jitter, which represents a random
error average estimate and dispersion, which represents a worst case
error bound.

Is there any correspondence to jitter in v3 as in v4? If there isn't
any way to calculate a estimate or distribution within the error
bound.
My idea, peer.dispersion represents the maximum error in offset and
maximum error of half the roundtrip delay. Couldn't that be used
instead of jitter, due to dispersion > jitter ?

Bertil




More information about the questions mailing list