GPS accuracy issue

[Yusuke]

modem fw v1.2.0
ncs v1.3.0
custom board

I'm trying to make a one-shot GPS app without assisted data by modifying gps sample app, and doing field test, but have an accuracy issue. Giving me any advice is helpful.

The app process is simple;
1. initiate GPS socket
2. start GPS, try to fix, and stop GPS (don't tweak "AT+CFUN" setting, just set START/STOP to gnss_ctrl function)
3. sleep for 10 sec
4. repeat step 3. and 4.

Then I walked through a street surrounded by 2 or 3-story buildings which are not so high.

The result doesn't correspond to where I walked.

issue.pdf

I get position data from last_fix.pvt.longitude and last_fix.pvt.latitude.

Why does this occur? I don't think this is hardware issue because it fixes a position at least.

Any tip?

[Yusuke]

The modem apparently stores GPS position data by FIFO buffer. Fixed position data is not removed unless it is read by nrf_recv. Is this correct?

Enabling single-fix mode by nrf_gnss_fix_interval_t fix_interval = 0 fixes the above issue.

[Yusuke]

Hi, may I ask another question?

I would like to get GPS position changing a turn-off period every fix.
I think it is good to use single-fix navigation mode repeatedly, but I'm not sure about the difference between repeatedly using single-fix nav mode and periodic nave mode.
Is there any drawback in using single-fix nav mode instead of periodic nav mode?

[Heidi]

Hi, on the current MFW there is no noticeable difference between the two.

In general if you want to get a fix within 30 minutes (1800 seconds) which is the maximum fix interval time we support, then we recommend using periodic mode.

[Yusuke]

Thank you for your advice. 

>In general if you want to get a fix within 30 minutes (1800 seconds) which is the maximum fix interval time we support, then we recommend using periodic mode.

There is no option to stop fixing in periodic mode though. It seems that I need to use single-fix mode in my application at this moment.