nRF9160 GPS Cold Fix Guidance

[GNotman]

I was wondering if I could get some guidance with regards to the effect of interrupting the nRF9160 GPS during an initial cold fix.

Consider an application which uses the nRF9160 GPS to perform regular single shot GPS fixes. The device is connected via CAT-M1 and is using eDRX. 

The device transmits information (which may include GPS fix data) at a specified interval, let's say nominally every 30s. These transmissions still occur regardless of whether the device was able to achieve a GPS fix.

Due to the scheduling between the LTE modem and the GPS, the GPS is only able to use the antenna for approximately half of the time, with the transmission and subsequent network release blocking the GPS for about 15s (of every 30s), after each transmission.

My question is, how might these regular interruptions affect the ability of the GPS module to achieve a fix, especially a cold fix?

Will these interruptions completely kill the performance, or might a ~50% duty cycle effectively double the average time to first fix?

[Didrik Rokhaug]

Hi.

If the GPS gets interrupted, it should continue where it left off the next time it starts.

So you should eventually be able to get a fix.

Best regards,

Didrik

[GNotman]

Hi,

I guess my question is really, how might these interruptions affect the performance of achieving the first fix?

How much of an impact might these interruptions have on the device's ability to download the Almanac and Ephemeris data?

Due to the interruptions, some parts of that data will be missing. For the Ephemeris that should be too much of a problem as the transmission period is 30s. The Almanac data however, is transmitted with a 750s period.

[Didrik Rokhaug]

I asked our GPS expert for input, and here is what he answered:

Cold start can roughly be divided into satellite search and ephemeris download.

With a good antenna and open sky, the satellite search could very well be done in 15 s. In more challenging conditions, the satellite search may take considerably longer. This is not a big problem, GPS just interrupts its activities when LTE uses the radio, and continues where it left off when it gets the radio back. In these conditions, satellite search just takes 15 seconds longer per interrupt.

Ephemeris download is more tricky. Nominally, there is 18 seconds of continuous reception needed to download the full data. Ideally, GPS should be provided at least 36 seconds to guarantee full download of ephemeris.

If needed, Ephemeris can be split into 3 x 6 seconds of download, whereby the 15 s LTE , 15 s of GPS is enough. However, in this case the LTE could overlap with the time instant when Ephemeris is available from the satellites. If now the cycle is a multiple of 30 seconds, it could take a very long time to actually wait for the timing to change so that Ephemeris availability coincides with GPS having the RF (LTE eDRX is actually a multiple of 10.24 seconds, while Ephemeris is available in an exactly 30 seconds cycle, so there is indeed a timing difference, albeit very small).

So the recommendation is: If possible, use a long enough interval between LTE transmissions, at least 1 minute, but the longer the better. Secondly, if possible, do not use  an eDRX period that is (close) to a multiple of 30 seconds, e.g. 50 or 70 seconds would be better than 60 seconds.

Almanac suffers from same issues as Ephemeris, however Almanac is not crucial for cold fix.  

In other words, you should be fine, though you do risk having to wait for a very long time if you have bad luck with the timing. You could avoid a lot of these problems by using A-GPS. That will let you download the almanac and ephemerides over LTE, and provide it to the modem.