qualcomm gnss official accuracy

[lodrog...@gmail.com]

after many years, finally we have the official accuracy from qualcomm. It is very funny when we hear from 2017 that the new smartphones models will support 30 cm accuracy. Into new latest SoC web pages that support dual band we can read the follow

Accuracy: Sidewalk / lane-level navigation

if i am not wrong, we are talking for 3-5 meters (or more ? )

[Sean Barbeau]

Interesting, thanks for pointing out that Qualcomm is now specifying some level of accuracy. Looks like you're referring to the Snapdragon 888 Plus specs:

https://www.qualcomm.com/products/snapdragon-888-plus-5g-mobile-platform

>Satellite Systems Support: Beidou, Galileo, GLONASS, Dual frequency GNSS, NavIC, NavIC enabled, GNSS, GPS, QZSS, SBAS

>Accuracy: Sidewalk/lane-level navigation
>Dual Frequency Support: Yes (L1/L5)
>Additional Location Features: Qualcomm® Location, Sensor-Assisted Positioning

The next natural question is how wide is the sidewalk or lane? :)

In the USA a standard highway lane is 12ft (3.6m), but can be less on arterial roads (10ft are also common). Sidewalks in the US are typically a minimum of 5 feet (1.5m) wide but can be wider.

So unfortunately I don't think this really specifies additional precision beyond what we've currently come to expect.

What's interesting is that to my knowledge Broadcom was the main chip OEM hyping decimeter accuracy with the BCM4775 series. But they manufacturer the GNSS chip, not the entire phone chipset. Qualcomm on the other hand provides a holistic solution with a full phone chipset. So Qualcomm may be providing more reasonable expectations as to what you should expect from integration with typical mobile phone antennas, etc., while Broadcom is providing a more optimistic view of what it's chip is capable of in many different applications, including ideal antenna configurations, etc. that would never be possible in mobile phones.

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/a716c233-1d67-43a7-902b-f33181caabc0n%40googlegroups.com.

[gpsfan]

Anyway this is going to be pretty academic as there's only so much the chip and the firmware can do, especially in urban environments. Google Maps' "Live View" feature is fantastic as it uses Augmented Reality to place you on the proper side of the street and provide perfectly accurate pedestrian guidance regardless of the multipath problems that no chip can solve be it single or dual frequency. Google was also working on adding a ML based engine via Play Services as explained by the Frank Van Diggelen late last year for "standalone" operation similar to "Live View" but I'm not sure what happened with that. Maybe Sean's seen some developer docs about that ?

[gpsfan]

That's where it was mentioned originally : https://www.gpsworld.com/google-to-improve-urban-gps-accuracy-for-apps/

" Inside the module, the 3D mapping aided corrections algorithms solve the chicken-and-egg problem, which is: if the GPS position is not in the right place, then how do you know which buildings are blocking or reflecting the signals? Having solved this problem, 3D mapping aided corrections provide a set of corrected positions to the FLP. A system API then provides this information to the GPS chip to help the chip improve the accuracy of the next GPS fix."

Rereading it now I'm not sure how that "chicken and egg problem" is solved though ?

[Sean Barbeau]

IIRC Google presented that work at Ion as well as Stanford's PNT Symposium but didn't go into the "special sauce" details that would allow someone to recreate the solution. My impression was they've likely filed for a patent(s) and are waiting on that before going public. The article said that "version 2" should go live for everyone in early 2021, but I haven't heard any additional updates.

Interestingly, looking back now, this quote is in the article (bold and italics are mine):

Charles Abraham, senior director of engineering, Broadcom Inc.:
“Broadcom has integrated Google’s 3D mapping aided corrections into the navigation engine of the BCM47765 dual-frequency GNSS chip. The combination of dual frequency L1 and L5 signals plus 3D mapping aided corrections provides unprecedented accuracy in urban canyons. L5 plus Google’s corrections are a game-changer for GNSS use in cities.”

And apparently the Pixel 6 and 6 Pro are using the BCM 4776 (see the GPSTest Database for the GNSS hardware model name - https://bit.ly/gpstest-device-database). So it's possible these phones will have the "best" implementation because they have hardware that was designed along-side the software solution.

Sean

To view this discussion on the web visit https://groups.google.com/d/msgid/gpstest_android/8f539a66-e041-4098-a1fb-1d29a70f0919n%40googlegroups.com.

[Sean Barbeau]

Oh, and the system "Force full GNSS measurements" Android API for Android 12 was likely created by Google to allow Google Maps to force this mode for their ML solution without relying on users setting it in the Developer Options. So Android 12 and up will see better results across all devices then Android 11 and below.

https://developer.android.com/reference/android/location/GnssMeasurementRequest.Builder#setFullTracking(boolean)

Sean

[lodrog...@gmail.com]

as i have wrote to the first post " Into new latest SoC web pages",  this is plural and not singular. This cover all the series, from 4xx up to 8xx. For example the new 480 plus SoC, has also " Accuracy: Sidewalk / lane-level navigation"
https://www.qualcomm.com/products/snapdragon-480-plus-5g-mobile-platform

the only advantage for the flagship 8xx series, is the support of SBAS. Lower new chip's is missing this feature, according qualcomm web pages.

[Navin Kumar]

The GPS on Dimensity 1200 is very capable of getting on accuracy of upto 1 meter (about 3.28 feet). The Dimensity 1200 has better gps accuracy than Qualcomm chipsets such as snapdragon 888 (1m accuracy vs 3m accuracy). It includes support for practically all GNSS (such as US GPS, Europe's Galileo, Russian Glonass, China's Beidou or Compass, Japan QZSS and India's NavIC), supports SBAS and dual-band GNSS (L1+L5 bands). You can check the review of Xiaomi 11T notebookcheck review (https://www.notebookcheck.net/Xiaomi-11T-smartphone-review-High-End-features-at-a-budget-price.576149.0.html ; section: Communication and GNSS - Fast Wi-Fi, accurate GPS) (disclaimer: not an advertisement or endorsement of product). But, the most accuracy you can get is about 1 meter not cm level accuracy.

But, for Mi 11T, the drawbacks are lack of microsd card, lack of 3.5mm headphone jack, not having flagship level camera performance, MIUI need to be set up like a geek (such as for no intrusive ads, privacy protection and regular synchronization, eg., the battery management is aggressive which leads to lack of synchronization of gmail, facebook, whatsapp notifications in real time).

Re:

Inside the building accuracy Xiaomi 11T (source: notebookcheck)

Outside the building accuracy Xiaomi 11T (source: notebookcheck)

[lodrog...@gmail.com]

there is no true reason to copy - paste tech specs, if you don't know ....

- the GPS on Dimensity 1200
what is the GPS ? it is a navigation system made from one country for world wide coverage. If the device support multi systems the description is GNSS.

- what about 1m accuracy
last 5 years (or more) mediatek from series of Helio SoC support 1m accuracy, according a lot of reviews. Some of them we have found into "notebookchec.com" and some from youtube video

- what about 1m accuracy
in what conditions ? static or movement are those measurements ?

- what about 1m accuracy
have you made a comparison test side by side, of a device with mediatek and qualcomm, to record a track (>1 hour) ? into the city with big buildings or into the forest with the mountains and a lot of trees ?
have check the quality of GNSS receiver measurements ?

- what about 1m accuracy
have you made a comparison test side by side (>1 hour), of a device with mediatek and qualcomm, to  check the "drift" into static measurments ?

of course, there are more questions, but those are a good start ....

[Navin Kumar]

- the GPS on Dimensity 1200

GPS as it is said by the smartphone tech journalists, technically that should be GNSS

- what about 1m accuracy

That's the accuracy what I have got under the wooden roof house and open area over a road , using Poco M4 Pro 5G which has Dimensity 700 which technically has dual-band support but Xiaomi has not enabled L5 band (the Poco smartphone only has support for L1 bands only so drifting is a little bit more I am trying to buy Xiaomi 11T to try the lack of drift or not using L5 band, I had Poco M2 Pro that had L5 band support using snapdragon 720G it has lesser drifting but much less accuracy).

Take this as from a amateur guy who has interest in GNSS in smartphones, I am not a professional expert in GNSS. I would try to figure out accuracy under a lot of trees, recording tracks after I get my hold in Xiaomi 11T.

But, you may be right we ain't getting sub-meter accuracy in smartphones GNSS.

Here's some screenshots: After 2 minutes static recording, aeroplane mode.

[lodrog...@gmail.com]

i am very glad that you understood that you are amateur and not a professional. That's why you are very glad with this device and you stop the tests (measurements) only at static location tests. If you made at 100% all the test as i wrote, you will discover that the devices just like you are using, are very fake. The purpose that they have been made is the navigation with google maps , sygic and tomtom. Not the measurements. The word "accuracy" is simply a joke. Except if you like to compare the 2021 GNSS receiver with one from the 1955. If you are happy with this, please continue to smile ....

[lodrog...@gmail.com]

sorry for the mistake of 1955. The correct year is 1995

[lodrog...@gmail.com]

it's very funny, when you wrote that you have the device into airplane mode .... but with the wifi into ON. This is how the RTK works, from the internet

[Navin Kumar]

Yeah, the accuracy of 1 or 2 meters is obviously fake from what I have tried to achieve comparing to google earth images.

I am glad you understand the meaning of amateur enthusiast. The multi-role of smartphones and miniaturization of GNSS antenna for reception may have made nearly impossible to have a good accuracy of about 2 meters or lower in smartphones (but you can get smartphone like GNSS receiver such as TDC 600 with accuracy of about 2 meters without RTK correction and Garmin GPSMap 65s, a handheld recreational GPS with accuracy of + or -3.65 meters).

In a really dense canopy forest and urban settings, smartphone GNSS starts to act weird, the location changes constantly and hard to have a even a good precision (like 2 to 4 meters range). Though, what I perceive that for good average conditions and open sky, static recording accuracy after 30 seconds to a minute can be obtained in the range of 2-4 meters typically which is okayish comparable to map grade handheld GNSS receiver. But, accuracy of 2 meters or less is hard to come by from what I have seen.

I will update you how the accuracy seems in Mi 11T once I am able to buy it. And, I had my wifi on for seeing my location in google map satellite view. And, I would comment on this thread again with total aeroplane mode with Mi 11T (that time with no wifi only GNSS mode).

[lodrog...@gmail.com]

Yes, you are amateur enthusiast, but with no experience as a customer. You have not realise that you spend your money and searching for technical specs that there are no exists. No matter if you spend 100$ or 1500 for a consumer smartphone device - the accuracy is the same.
Next time you will try to buy a smartphone device for the quality of GNSS receiver, look at manufactured specs. if they advertise only
- antutu score
- the megapixels of camera
- how fast charge the battery
STAY away .....

if you need to travel with speeds over 20km/h, the smartphones devices with 1Hz sampling rate is no good.

my suggestion is to use one external gnss receiver with bluetooth. As an example look this
https://gpswebshop.com/collections/new-arrival/products/top608bt-high-precision-usb-bluetooth-gnss-receiver-centimeter-level-accuracy-multi-band-rtk
connect the usb cable to a power bank and with internal bluetooth, transmit the data to smartphone you have. The specs are
- multi constellation (6)
- multi band (2)
- multi sbas (>4)
- multi Hz (20)
solutions like this, is far more cheap from smartphones like trimble dtc 600, at the 1/4 of the cost and more accurate. Add one smartphone device at 250-400 $ and you will be OK for the next 10 years.

At the end you will be decide. All of them, it was my personal opinion

[lodrog...@gmail.com]

about garmin new 3 handheld models of 65, 65s, 66sr, with support of multibad gnss

- into youtube videos or web pages reviews, i have seen better static accuracy of 1.8 meters, better accuracy inside the house, more accurate track recording compared with more expencive smartphone devices

[Navin Kumar]

Thanks for your detailed reply. I would like to have a device similar to Garmin GPSMap 65s, but accuracy within the range of 30cm to 1 meter and price range of $1000-$2000.

Between, Trimble TDC600 has static horizontal RMS accuracy of less than 1.5 m typically and a similar device Spectra Geospatial SP20 has accuracy of horizontal RMS accuracy of less than 0.5 m typically (with real time SBAS, without subscription based RTK correction as specified by the OEMs), both the device are pretty similar in specifications, but the SP20 has a much larger dedicated antenna compared to the TDC600. From this I assume that the GNSS receiver in the SOCs may be capable of achieving sub-meter accuracy, but the miniature GNSS antenna is to be blamed for not being able to achieve sub-meter accuracy in smartphones.

And can you tell me if GPSMap 65s is capable of 1-2 meter accuracy and can you show me any proof of it? Any further information will be highly appreciated.

[lodrog...@gmail.com]

Trimble and Spectra are the same company. Model tdc600 and sp20 are the same. Both of them are using ublox 8 as gnss receiver. The internal SoC has disabled the gnss receiver.
those 2 devices are made as oem from the company unistrong. Unistrong is making oem products for 7-8 big names into GNSS area (companies)

garmim dual band accuracy
https://www.youtube.com/watch?v=bulWXsKV0WQ
https://www.navigation-professionell.de/en/garmin-gpsmap-66sr-review/
https://www.navigation-professionell.de/en/garmin-gpsmap-65s-review-a-great-outdoor-partner/

garmim dual band accuracy vs L1 only
https://www.youtube.com/watch?v=OMS2I-apI8Q
https://gpsradler.de/video/garmin-gpsmap-multi-band-gps/
https://gpsradler.de/news/garmin-gpsmap-66sr-65s-vorgestellt/

[mmaxi]

Hello,

I had a Garmin 65s and now have a multi-band 66sr, the accuracy is better than on a non-multi-band 66s but it's not that great. There are errors on a recorded track from 2m to 5m, sometimes a little more. Garmin also limits the satellites to the best according to what they say, when on my smartphone I have 25 fixes, on the Garmin 66sr I have about ten.

--

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/6dfb19fb-f0b6-4f16-bbf4-027ca06667dbn%40googlegroups.com.

[lodrog...@gmail.com]

just for the knowledge ...

- garmin 65s and garmin 66sr are both dual band devices
- the static accuracy and the moving accuracy (track recording) has a big gap. This something that they all must know. When we see reviews into the web for track recordings, garmin dual band vs smartphones with dual band (into airplane mode), the winner is the garmin
- who cares if the device No1 can see 8 satellites and the device No2 can see 60 satellites. The user must focus at the best quality of navigation accuracy and then at the static accuracy
- YES, it is well known that the garmin PND has a lot of firmware bugs. But they have proof that they can work better into the real life.

those are my personal thoughts