Integrated Accumulated Delta Range (ADR) in WLS

[Chavisa Sornsakul]

Hi folks,

I'm currently working on integrating ADR into WLS for improving smartphone relative positioning.

Actually, this approach can be produced by GNSSAnalysis app from google but, unfortunately, the source code doesn't provide any more than WLSPVT from pseudorange (gps-measurement-tools from Google).

I tried working on my android data using GNSSAnalysis app and found that WLS relatve positioning using ADR has high potential to solve my problem. So, I wanna try on this technique and develop from it.

What I found about relative positioning using ADR is this paper: Draganov, A. (2006). GPS accumulated delta range processing for navigation applications (United States Patent US20060195262A1). https://patents.google.com/patent/US20060195262A1/en. 

it's helpful but full of math equation (which is great btw). But, i'm looking for the papers or projects that more practical or it wll be great if code-based provided. So, i'm posting this in case that maybe someone working on these things as well. Please let me know if you have any thoughts.

Thank you for your time, help or comments.

[Sean Barbeau]

I've catalogued some open-source projects here you might want to check out:

https://github.com/barbeau/awesome-gnss

Specifically, RTKLIB will probably be helpful - https://github.com/tomojitakasu/RTKLIB.

Sean

[Jim bell]

The one thing I'd really want you to do for GPSTest is to add averaging to the output. I want to set my dual frequency, SBAS-capable smartphone somewhere, turn on location averaging, and in an hour get a result that is accurate to 10 cm.  Or, at least, STABLE to that value.

Or potentially even better.   

It occurs to me that there might be a market for a product that includes just a gnss antenna and an amplifier, with a cable and an attachment that allows you to insert a smartphone into it, in order to present the amplified (+10dB?) gnss signal to the the smartphone.

The internal RF circuitry of a smartphone, if limited by its antenna, might be quite sufficient if the incoming signal was amplified.

--
You received this message because you are subscribed to the Google Groups "GPSTest" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gpstest_andro...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gpstest_android/655fd005-ad2a-4851-a5a9-14df48c7a45dn%40googlegroups.com.

[Sean Barbeau]

Thanks for the feedback Jim! I have an issue open for adding point averaging at https://github.com/barbeau/gpstest/issues/532, please be sure to upvote it with a "thumbs up" icon.

Also note that GnssLogger does have point averaging via the "Ground truth for residual plot" option, then select "Use Computed location", "Chipset", and the method you want (e.g., Median or Average). The "Residual plot ground truth" will then show up on the map as a black cross with a circle around it.

Sean

[Jim bell]

From reading all that, an old saying comes to mind:

    "never let the perfect be the enemy of the good.".

Yes, you can do this in a very complicated fashion. The fact that it seems to have been an issue so far for three years implies that the perfect is becoming the enemy of the good.

Yes, individually averaging the latitude, the longitude, and the altitude would be an excellent start.  But you could also plot these individual averages over time. This way we could see how long it takes to achieve a stable average.    

A somewhat more complicated method, but I think would still be quite doable, would be a plot of the distance from the average at the the end of the data.  I realize that sounds complicated so I'm going to have to clarify

As this average is created, it should approach ultimately an ideal average value. The hardware and software will not know what that average value eventually is, but it has its own running average. It could create a plot which it goes back updates the value:  how far in three dimensional space were all those individual readings as compared with the eventual average value?

 I hope I can make this clear: in most plots, you would determine what the value was at some specific time, and draw it in the graph. In the system that I am thinking about, the early readings are not static: as the average value of the location changes, the initial plotted values are corrected against the then final value.

So what's the purpose of this?   I think that you want to ask the following question:

"How far is this current reading from the eventual average that this averaging process determines?"

Another question might be: how far from the eventual averaged value is what would be the current average value, if you were to stop the process at this point?  

So I would say this: choose a process and implement it so it can be used.  It looks like for the last nearly 3 years, nothing has happened.

[Chavisa Sornsakul]

Thanks a lot for sharing!