nrf9160, AGPS works longer than GPS

[Stas Jis]

We use nRF9160 on our own board.
The firmware "Serial lte modem" is installed on the chip with a connection to an external MCU (#define MQTT_MESSAGE_BUFFER_LEN 8192)
We tested the environment with periodic switching from AGPS to GPS.
And in tests, we constantly see that the operating time with AGPS is twice as long as with just using GPS.

Average waiting time when using AGPS: 1 minute + 10-15 seconds;
Average waiting time when using GPS: 30-40 seconds.

What could be the reason for such behavior? It was expected that the operating time with AGPS would be significantly faster than with GPS.

The command sequence for AGPS:
AT+CEREG=5
AT+CFUN=1
AT#XNRFCLOUD=1
AT#XAGPS=1,1

The command sequence for GPS:
AT+CEREG=5
AT+CFUN=31
AT#XGPS=1,1

The tests were carried out statically, without movement.

[Håkon Alseth]

Hi,

 

You are counting from the wrong timestamp. You must compare the two options when the GPS is actually running.

When using AGPS, you will first download GPS data from the internet, then you'll have to wait until the GPS can run (ie. LTE Link is idle) before you will get valid pvt/nmea frames.

 

If your conclusion is that you do not need AGPS, that is an application specific decision, but we do offer this to avoid a cold start scenario for the GPS. 

 

Kind regards,

Håkon

[Roman1982]

This is not related to conclusion,

i will explain it again , our target is to achieve as fastest lock as we can .

i am asking to provide a sequence how to achieve it!

i thing the better way to handle this issue that you will provide sach sequence and we will test it on our side. 

as i mentioned in SDK 1.7.1 we had a great performances.

[Håkon Alseth]

You need to explain what you're doing and how you're doing it.

My former responses explains why fetching AGPS data from the internet will block the GNSS from running, thus gives a seemingly larger time to first fix in some scenarios.

Kind regards,

Håkon

[Roman1982]

Our application is operates like a tracker and have a two operation modes: 

1. send location every 15 min 

2. stream continues location one at 1 min

the application is LOW poser and this is why the lock duration is importend .

the sequence that this case begins is a sequence we are currently using.

please advice 

[Roman1982]

one more thing 

we testing TELIT GPS performance and the time for a LOCK with assistant GPS is standing on 20 sec.

(of cause they also have a wikipwdia about the idle modes etc.. but on the field we observe match better performances)

this lead me to understanding that we are doing some thing wrong and we need your assist to define the shortest way to have a GPS lock and sending this information to the cloud. 

Thanks in advance ,

waiting for your respond