[time-nuts] GPS Selective Availability. Is it On or Off?

Magnus Danielson cfmd at bredband.net
Mon Mar 13 17:31:49 EST 2006


From: "Tom Clark, K3IO (ex W3IWI)" <K3IO at verizon.net>
Subject: Re: [time-nuts] GPS Selective Availability. Is it On or Off?
Date: Mon, 13 Mar 2006 16:44:51 -0500
Message-ID: <4415E7D3.8000106 at verizon.net>

> Chuck said
> 
> > I got the notion that it was turned off during Desert Storm,
> > by virtue of being involved in the e-warfare effort that lead
> > up to, and followed the event.
> >
> > I haven't been paying much attention since.  I knew that they
> > had intended to turn SA back on after production of the p-code
> > units was up to speed, but I hadn't heard whether or not they
> > did. 
> Yes, it was turned off for a brief period during DS, largely because the 
> DoD had to scurry around to buy mortal commercial units to fill the 
> need. Also during DS (and the present excursion) lots of parents sent 
> COTS GPS widgets to their kids.
> 
> It turned out that one of the most important uses of cheap GPS receiver 
> in DS was by the food trucks. Troops were deployed in the desert all 
> along the Iraq & Kuwait border. The mess tents were behind the lines, 
> and hot meals needed to be delivered to the remote outposts. The 
> delivery trucks found they could navigate across the roadless desert 
> very well by using GPS receiver intended for navigating civilian boats.
> 
> S/A is a dithering of the clock with a pseudorandom phase jitter. The 
> key to disentangling it was to have the same code generator available on 
> the ground. I use the analogy that DoD had a smart mouse in each 
> satellite running around on a phase resolver. To de-jitter it, you need 
> the mouse's clone inside the receiver.
> 
> The dithering of S/A had nothing to do with the encryption of the P code 
> to make the Y code. The P-code is a LONNNNG code (37 weeks until a 
> repeat) at 10.23 Mbits/sec. Each of the satellites uses the same code 
> stream, offset by some integer number of weeks. The Y-code is an 
> additional secret code that uses a shorter code to (pseudo)randomly flip 
> the phase of the P-code. On the ground, the civilian "code crackers" 
> have found out that the convolution code is running at a rate ~500 
> kbits/sec. This means that the Y-code may be the correct P-code for ~20 
> bits, and then it (may|may not) flip phase to become "anti-P" code. 
> AFAIK, Ashtec's patented "Z-code" receivers generate a hardware estimate 
> of this code and (nearly) coherently demodulate the signal. Other brands 
> have similar tricks up their sleeve.

The Y-code is the P-code xored with the A-code (sometimes also referred to as
the W-code). The A-code is indeed ~500 kbis/sec. The first "codeless" receivers
just squared out the A-code from the equation, but then they had a worse
problem to fight regarding ambiguity. Also, it does not form a very good
receiver. The Ashtec solution is to make the L1 handover from C/A-code to
P-code and predict the A-code, delay that a suitable amount to the L2 Y-code
and attempt to lock up to that. The delay is trimmed to match up with the
L1-L2 delay in P(Y)-code. You could say that the Ashtec receivers cracks the
code, but they really don't since they do not disclose the state of the A-code
generator or its architecture. Infact, they don't even get it rigth all the
time, but sufficiently often for a good lock since each success has a good
quality.

It is interesting that what they did to figure things out was hunting GPS
satellites with a big parabol antenna tracking the satellite and getting a
much better S/N than normal semi-omnidirectional antennas. With that they
could make advanced guesses.

Cheers,
Magnus




More information about the time-nuts mailing list