ION Smartphone Decimeter Challenge

[Jim bell]

https://www.kaggle.com/competitions/smartphone-decimeter-2023

I've just seen a reference to this, referred to by Sean Barbeau, Author of GPSTest app.  

    As a ham radio operator, since 1986, N7IJS, I am quite familiar with electronics and radio technology.  I have a suggestion to add.   We all have heard, and presumably understand, that the GNSS antennas are non-ideal, and that this represents some sort of limitation to the accuracy of GNSS readings.   But, I don't know, and you don't know, to what extent such an antenna limitation impacts the actual accuracy as seen in smartphones.

     So, I have a suggestion, that a ham or other electronics expert might do, in parallel with this Smartphone Decimeter Challenge, in order to determine just what the error might be dropped to:  The actual limitation of the antenna, and by extension, the limitation of other factors that might be present in the smartphone itself.  

     My suggestion is to try to make that antenna limitation irrelevant, at least for temporary, experimental purposes.  I will be more specific:  We know that there are more-ideal GNSS antennas.  When powered and mounted appropriately, they receive DC power through the coaxial cable,, and send back an RF signal over that same wire.  Mechanically, and electronically, we can say that they are relatively ideal.

     So, why not expose a smartphone to that idealized, amplified RF signal?  Better, and louder, than any smartphone's antenna could achieve alone.  Perhaps that signal would be 10 db (10x) or even 20 db (100x) 'louder' than the smartphone would ordinarily 'hear'.   Maybe not TOO loud, of course!  But loud enough so that the "bad antenna' is no longer going to be the limitation.

     Why do this?   It might help tell us how good a smartphone could ultimately get.  Remove the antenna limitation, temporarily, and we could then know what the other factors (electronics?) remain.   If, hypothetically, we remove the antenna as a factor, and there is no improvement, it presumably means that the antenna really isn't the problem we thought it was.  Alteernatively, if we remove the antenna limitation, and the accuracy is greatly improved, it means that the antenna is indeed the problem.  

     The electronics behind this test could be exceedingly simple:  Feed a good GNSS antenna with DC power according to its needs, possibly through a coil of wire that is mechanically 'close to' a smartphone.  The signal it sends back will create an RF field that will activate the GNSS antenna within the smartphone itself. 

            Jim Bell

[gpsfan]

Sounds like a good idea! The results should be excellent or do you think there might some other interference inside a smartphone? Quite possible I suppose.

On a related topic, here's an article and solution for the challenge using the free RTKLib app : https://rtklibexplorer.wordpress.com/2023/09/21/2023-google-smartphone-decimeter-challenge

[Sean Barbeau]

Hi all,

I've put some references to existing academic publications here:

https://github.com/barbeau/awesome-gnss#academic-peer-reviewed-publications

IIRC this publication has some results for connecting an external GNSS antenna to a phone:

Yong et al. - "Instantaneous, Dual-Frequency, Multi-GNSS Precise RTK Positioning Using Google Pixel 4 and Samsung Galaxy S20 Smartphones for Zero and Short Baselines", Sensors 2021, 21(24), 8318; https://doi.org/10.3390/s21248318.

Sean

--
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/d61bc1f2-f52d-4547-bea7-469b105a6244n%40googlegroups.com.

[Jim bell]

I have checked out my new OnePlus Nord2 smartphone, which has both dual-frequency GPS, Galileo, Glonass, Beidou, and SBAS (WAAS, at least, maybe more) capability.  I am VERY impressed by its performance.  It seems to be at least as good as 1 meter accuracy.  In fact, the main problem is that the existing GNSS apps aren't really up to the task of measuring (at least) the stability of the signal when placed at a specific location and allowed to receive signals.

GNSS apps typically have one or more of a few limitations:   First, some do not have adequate resolution.  A resolution of 1 meter, approximately, might have seemed okay with old, L1 only, 10-meter-accurate GPS, but times have changed.  I'd like to see DD.ddddddd (seven digits after the decimal point) GPS readings, and so a resolution of 11.1 millimeters.   Or DD MM.mmmmmm, or DD MM SS,ssss.

The lack of averaging is a major problem, too.  A dual-frequency, SBAS-equipped GNSS device which is typically instantaneously off by 1 meter might, if averaged over a few minutes, indicate the location within 0.1 meter.   But I cannot tell, because the software tools aren't up to the task of measuring this error.