Re: portable atomic clocks

From: B Magnus B{ckstr|m (b@eta.chalmers.se)
Date: Fri Jan 19 2001 - 01:42:01 PST

  • Next message: Rod Sladen: "(no subject)"

    On Thu, 18 Jan 2001, suhas wrote:
    
    > In this method of synchronization
    > 1. what is the time error of communication?
    
    The time error of any network communication varies unpredictably and
    depends on the types of network between you and the peer, and on current 
    circumstances such as network congestion.  NTP is designed to measure
    such delays and take them into account.  If the dispersion of the
    round-trip delays is large, your synchronisation will be poor.  This can
    be compensated by gathering a larger set of samples and talking to several
    reference clocks on different parts of the network.
    
    Myself, I have xntpd running in the background on my freebsd box.  It
    talks to two clocks -- one at the local university and another in
    the next city.  And of course it can only talk to them while my modem
    link is up.  I have a fairly ordinary 33 kbit modem.  So while I'm surfing
    and reading news and e-mail (creating unpredictable network delays, that
    is :-)) my xntpd periodically exchanges a few packets with its friends on
    the network.  Last I checked, there was no distinguishable offset between
    my computer's clock and the telephone time service.  That puts it within
    some 50ms, which I'm happy with.
    
    I have no doubt that it is possible to achieve synchronism to millisecond
    accuracy if one so wishes.
    
    The NTP standard is interesting reading:
    http://www.uni-duesseldorf.de/ftp/pub/doc/rfc/RFC1300-1399/rfc1305.pdf
    
    > 2. does this depend upon where we communicate for synchronization?
    
    Yes.  Don't choose clocks that are at the opposite end of the internet.
    
    Magnus
    
    -----------------------------------------------------------------
    Unsubscribe from SeeSat-L by sending a message with 'unsubscribe'
    in the SUBJECT to SeeSat-L-request@lists.satellite.eu.org
    http://www2.satellite.eu.org/seesat/seesatindex.html
    



    This archive was generated by hypermail 2b29 : Fri Jan 19 2001 - 01:42:35 PST