[time-nuts] GPS 18 behavior

Jim Lux jimlux at earthlink.net
Mon Jul 22 09:44:59 EDT 2013


On 7/22/13 6:30 AM, Tom Van Baak wrote:
>> For what it's worth, the application is a radar that detects
>> buried victims in disaster rubble, so the data we are collecting is
>> basically heartbeats and breathing.  the "when was the data taken"
>> is a "where were we when the data was collected" need.  The "sync"
>> requirement comes from being able to find the same heartbeat in
>> multiple data streams.
>
> Jim,
>
> That's a fascinating application. Ok, one last comment then. As much
> as GPS is an obvious solution, did you consider the use of multiple
> homebrew timing pulse pseudolites instead? If one placed a couple of
> them around the vicinity of the disaster area you could triangulate
> for your ranging and timing information. Since the transmitters would
> then be less than 1 km away instead of more than 20,000 km away, you
> avoid all the limitations of GPS. Moreover, the solution would work
> if GPS were functional or not, something perhaps important for a
> disaster situation. Using some clever waveform (like what Loran-C
> did) you could better penetrate into all sorts of rubble and overcome
> multi-path at the same time.

Great minds think along similar paths.. Yes, in the long term plan, 
that's essentially what we'll do: multilateration among the multiple 
units.  But for the mean time, we need to mark the locations with GPS 
derived coordinates.  The search and rescue folks currently use Garmin 
handhelds, so they're aware of where it works and where it doesn't.




>
> I'm pretty sure there are a number of papers on local ranging and
> navigation based on these technique (e.g., UAV, robotics). The
> advantage is that it's a local solution, it can work indoors,
> underground, underwater, even on the moon; in general, it doesn't
> rely on a functioning GPS constellation with its inherently weak or
> jammable signals.

Interestingly, there's a whole lot of "laboratory demos" of various and 
sundry techniques, but the GPS-denied time/frequency/position problem is 
very non-trivial.  For instance propagation through rubble is a fairly 
harsh multipath/scattering environment.  The delay spread is about 10% 
of the distance traveled, and knowing what the average "epsilon" is, so 
you can turn propagation time into distance, is not easy to determine.

Anyone who claims that they can do centimeter level position 
determination at a range of 10 meters in rubble (or can do microwave 
imaging/SAR/etc, which implies the same thing) is, I think, blowing 
smoke, unless they have a VERY sophisticated measurement setup and a 
huge amount of computational horsepower to solve the "inversion 
problem". (essentially, look at the first returns in time, use that to 
solve for the EM parameters of the first layer; then move the analysis 
front forward a step, do it again, etc., etc.,etc.)

Fortunately, unlike seismic data processing, for the most part the 
medium is linear, albeit somewhat anisotropic.


>
> Of course all this from the guy who once tried to use seismic sensors
> and GPSDO to triangulate moles in his backyard...

A valiant effort I'm sure.. And, I'll bet you learned all about the 
propagation anisotropy of soil<grin>


But yes, in general there are general strategies to do multilateration 
(and multiangulation using direction of arrival as well) among multiple 
nodes that are cooperating.  Taking a general strategy to something that 
actually works in the field, preferably using off the shelf hardware 
modules of some sort, is a bit of a chore.



More information about the time-nuts mailing list