[email protected] said:
> te a histogram of the values (along with the average and standard
> deviation).   I'm now using the peak histogram bin(s) to determine the
> message offset time.   The histogram technique has the advantage of ignoring
> outlier points that can be caused by the system being tied up / interrupted
> doing other things (like shoving a Windows 10 upgrade up your systems' rear
> I/O port  ;-()

How do you determine the bin size?

ntpd has an interesting filter in that area.  For refclocks where it has many 
samples, it sorts them, then discards roughly 1/3 of them as outliers.  The 
code is simple.  Compute the average then check the first and last samples to 
see which is farther away.  Drop it, iterate.

-- 
These are my opinions.  I hate spam.



_______________________________________________
time-nuts mailing list -- [email protected]
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Reply via email to