[159654] in North American Network Operators' Group
Re: GPS attack vector
daemon@ATHENA.MIT.EDU (Lamar Owen)
Thu Jan 17 08:57:30 2013
Date: Thu, 17 Jan 2013 08:57:20 -0500
From: Lamar Owen <lowen@pari.edu>
To: Jay Ashworth <jra@baylink.com>
In-Reply-To: <11061315.2858.1358384776739.JavaMail.root@benjamin.baylink.com>
Cc: NANOG <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
On 01/16/2013 08:06 PM, Jay Ashworth wrote:
> Do you use GPS to provide any mission critical services (like time of day)
> in your network?
>
> Have you already see this? (I hadn't)
>
> http://arstechnica.com/security/2012/12/how-to-bring-down-mission-critical-gps-networks-with-2500/
>
Hi, Jay,
Yes, saw this about a month ago. We have a UNAVCO Plate Boundary
Observatory station (779) on our site, and it uses a Trimble NetRS. We
also use GPS timing locally to generate NTP stratum 1 for our LAN via
Agilient/HP Z3816 disciplined receivers, and individual GPS receivers
for both of our 26 meter radio telescopes for precision local standard
of rest calculations.
But as a frequency standard for 10MHz, we only use the output of the
frequency locked loops in the Z3816s as references for our Efratom
rubidium standard; even cesium clocks have more drift than rubidium
ones, and the rubidium is manually locked, and is the master reference
for anything that needs a frequency reference; the Z3816's can have
significant jitter (well, significant is relative.....). Last I
checked, the rubidium was 8.5uHz (yes, microHertz) off according to the
GPS disciplined 10MHz signal from one of the Z3816s (we use an HP
differential counter with a very long gate time to get that measurement
precision).
It was interesting timing for the release of this paper, as it was
around the time tick and tock were rebooted and went all 'Doc Brown' on us.
Anyone interested in the vagaries of serious time precision, please
reference the 'Time-Nuts' mailing list, and other content, hosted by
febo.com.
