[FOSS-GPS] RE: Post-processing RINEX to simulate RTK

Antonio Pestana afsm.pestana at gmail.com
Wed Oct 26 07:29:29 EDT 2011

Dear Thomas:

> Who is the manufacturer of your GPS equipment that you used to collect the
> Rinex data?

AS10 antennas, GMX 902 GNSS  receptors and software GNSS Spider, everything
from Leica. I want do the tests in real-world situations, and I already have
one that set that spans for two months at 20 MHz, almost without

> Please define ‘very demanding’ in terms of horizontal and vertical
> accuracy requirements

We want to test GNSS near real-time positioning techniques in the monitoring
of large structures (mainly bridges and tall buildings). Better than 1 cm is
mandatory; better than 0.5 cm is what I would like to get. Our baselines are
not long: usually less than 2 km. We want near real-time (real-time with a
constant, and small, time delay) because we want to plot the behavior of the
structures under dynamic loads.

> RTK by it’s very nature does not utilize a intermediate PC.

The GMX 902 receptors are just dummy receptors: they don't have processing
capacity apart from the one that is needed to get the code and phase
observations. So we need PC's to do the configuration of the receptors
(elevation mask, rate of observations, etc.), defining the data to be logged
if any (and the formats to use) end do the positioning computations. In my
setup I used a PC for each receptor. I could not do the displacement
analysis in real-time using Spider because I could not connect both rover
and reference to the same PC running Spider. That's why I decided to store
the observations (rover and reference) in RINEX files. No I want do process
these files as if they were being collected just now.

> The rover applies a correcting factor that was developed by the base and
> broadcast in some fashion.

I am aware of these techniques mostly used in large-area RTK network for GIS
and Surveying. The  observations made by the reference stations belonging to
the network are used to build model of the errors for all the area of the
network. Then there are two main "ways of doing the things": a) the rover
sends is approximate position to the network and the network returns to the
rover the corrections computed for the position of the rover; b) the network
sends to the rover the parameters of the error model and the rover computes
the actual values to his position. 

Given that my baselines are small I think that in my case the best approach
is to use plain-old double-differences. As you stated the use of
ionospheric-free combination of observations may prove to be disadvantageous
(theoretically these linear combinations have the potential to increase the
noise): clearly this is one aspect that needs to be tested. But I need to
get one position for epoch, that is to say, I want to get positions at a 20
MHz rate.

> To answer your question, the processing software that come with your GPS
> units should be able to do the work depending on the features that were
> licensed with it

Yes, Spider could to the job. But as I have already written it needs to be
connected to the two receivers at the same time. This connection proved to
be impossible in site.



View this message in context: http://open-source-gps-related-discussion-and-support.1099874.n2.nabble.com/Post-processing-RINEX-to-simulate-RTK-tp6926010p6932334.html
Sent from the Open Source GPS-related discussion and support mailing list archive at Nabble.com.

More information about the FOSS-GPS mailing list