[time-nuts] PN sequence generation using GPS

[jimlux]

On 2/26/11 5:23 AM, ehydra wrote:
>
> If one looks in the spectrum in a very fine granular structure the data
> transferred by SS will be seen on every single spread-code bit!! All
> needed is a high-enough S/N and a lot of computing power. On the analog
> side the receiver must be very strong signal capable so other
> 'intruders' will not be mixed-in in the wanted signal.
>

For the types of system time-nuts might have, aside from the "total 
power radiometer" (aka "crystal video") approach (which requires good 
knowledge of the expected noise power, so you can see the increase)..

100 percent probability of intercept receivers (of which an FFT is one) 
have a dynamic range challenge. either you have a narrow band, with good 
dynamic range (slow ADC, lots of bits), and miss a lot of the energy, or 
you have a wideband system, with less dynamic range (so the noise floor 
comes up from quantization noise).

As Henry points out, if there's any other signals in the band you're 
looking at, you need to run your ADC loading so that they aren't 
clipped, making the guy you're looking for that much closer to the grass.

It's all about Time-Bandwidth product and detection statistics.  You 
drag out your copy of the Radar Handbook, etc.

If the transmitter happens to have some obvious vulnerabilities (a new 
hop starts exactly as the previous hop ends, and the hop rate is 
constant) you can leverage that.

But this is getting pretty far afield from time nuts..

I was intrigued to think about the ubiquity of time/frequency 
distribution these days, as a way to more easily get sync. That was 
always the challenge back in the 80s.  Today, getting a 10 Mchip/s 
system synced would be no problem with even a fairly grungy 50ns 
uncertainty 1pps from GPS to get close, and FFTs to do parallel acquisition.

The folks doing this for real must be doing something far more exotic, 
then.. one can only wonder.