[time-nuts] Re: 20210423: Introduction and Request for Spirent GSS4200 User Manual / Help

[Lux, Jim]

On 4/24/21 10:31 AM, Andrew Kalman wrote:
> Hi Paul.
>
> Yes, I've been on this same journey. After I learned (somewhat unrelated)
> that one is supposed to have an FCC license to rebroadcast GNSS signals
> (e.g. via a repeater inside a lab, makes eminent sense), I started thinking
> more about GNSS simulators and how they might be added to my company's
> workflow. So I bid on a couple of units, got them for pennies on the
> dollar, and started messing with them in the hope of ending up with an
> ATE/rack-type setup that I can build into a nearly automatic test &
> validation suite.
>
> Let's say I was much more successful with the Spectracom/Orolia GSG-5 than
> with the Spirent GSS4200 ... In the case of the GSG-5, it's really just a
> question of how many options you can afford -- the rest is all there, you
> don't need a support contract, it's all easily accessible in the unit
> itself, and as long as the Internet exists the GSG-5 will probably keep
> working (it gets time, ephemeris and almanac data from servers -- it can
> simulate stuff NOW (wth the right options), not just in the past and
> future). The GSS4200 is about 10-15 years older, and it shows (in terms of
> ease-of-use), along with how Spirent chose to monetize their users /
> subscribers. Also, the GSG-5 adds things like interference to the signals
> (all for a price, of course). IOW, the newer units (at least, from
> Spectracom was XL Microwave is now Orolia) are a whole lot easier to use
> ... but they come at a price. It's an interesting business.
>
> I will say that the build quality of the Spirent is very good. I have not
> opened up the GSG-5, just did a calibration and it was very close.
>
> I'm a little bit surprised that there is not an open-source, SDR-based GNSS
> simulator (at least, one I could find).



Not much demand, I suspect.  I seem to recall a GNSS generator that was 
open source about 5-10 years ago, but I can't find it now.

The record/playback boxes are actually pretty simple - just a single bit 
in many cases. After all, a lot of the receivers use a single bit input, 
because the signal of interest is below the thermal noise floor.

The real challenge isn't the SDR part (a USRP would work just fine as 
long as you get a daughter card that supports L-band) - it's the 
"scenario building" which requires simulating the orbits of the GNSS 
satellites, simulating the track of the receiver, calculating the time 
delays (including iono and tropo effects), and generating the PN codes 
appropriately.

Each of those isn't too tough, but putting it all together is quite 
challenging, and, apparently, it's not "dissertation topic" suitable 
(which is where a lot of niche SDR stuff comes from).

A *real* challenge is that to do it right, you need very good orbit 
propagators - if you're looking to simulate nanosecond scale 
phenomenology, you need to be able to generate orbit behavior on a few 
cm or better sort of uncertainty.  For some applications (differential 
GPS, RTK surveying) you could probably get away with something that's 
not perfect, but doesn't have problems for YOUR specific application.  
But it wouldn't be a generalized box.

One strategy we've used at JPL is to have the fancy expensive box 
generate the signals for a scenario, and record them with a much cheaper 
record/playback box, then use the playback for testing.

Right now, my project (SunRISE mission) is working on how to generate 
realistic test signals for a space interferometer - Where we need to 
generate signals that can be received, and the output of the receiver 
fed into GIPSY-X for post processed precision orbit determination.