AGPS fix times

[Tron]

Hi

I'm on a custom board with NCS 1.2.0 and modem firmware 1.2.0. 

We have implemented agps, and testing with 2 units with agps colocated with 1 unit without agps. All 3 are running on firmware 1.2.0. They are all on a 2 hour reporting interval. All GPS'es time out after 150 secs. 

The units with agps are in the same platic casing, but the casing for the agps units are not welded. 

The non-agps unit (randomly picked) outperforms the agps units with a factor of 5, ie the fix times are 5 times larger on the agps units. This baffles us, and we want to know the reason. . 

Yours, 

Tron

[Håkon Alseth]

Hi Tron,

 

The non-agps unit (randomly picked) outperforms the agps units with a factor of 5, ie the fix times are 5 times larger on the agps units. This baffles us, and we want to know the reason. . 

Q1: Can you share the fix times you're seeing on both agps and non-agps devices? Do they all start from cold-start?

Q2: Do you run the AGPS sequence every time a fix is to be reported? If yes; this would add some seconds to the full process, as the connection to the SUPL server can take several seconds to fully complete.

Q3: Do you have any logs with timestamps to show the time of each individual function/task?

 

Kind regards,

Håkon

[Tron]

A short explanation is required: 

• Unit with firmware 20 (2008962) is the non-agps unit
• Seq number 0 indicates a restart, unless the previous seqnumber is 63
• For the non-agps unit, the Dispersjonstid column have to be multiplied by 2 to get the gps fix time
• For the agps units the Dispersjonstid is the time from when the GPS is turned on after getting the agps data from the supl server (google) in seconds
• AFAICS, the unit requests AGPS data every time it tries to get the osition, as long as the interval is bigger than 1h. 

No logs are provided, this is because the units are attached to Vilde, out test sheep. 

Tron

[Håkon Alseth]

Hi Tron,

 

If you start the GPS right after the AGPS session is done, it will take some time before PSM is entered successfully. There's a period, hard-coded in each network, where you are in RRC idle, before PSM is entered. This time has been observed to be 60 seconds in the Telenor Network.

 

Is it possible to start the SUPL/AGPS session some minutes before you should sample the GPS?

 

Kind regards,

Håkon

[Tron]

Hi Håkon

I'm not sure I understand. AFAIK, I don't stop or reconnect the modem in any way after the AGPS data is received. I only use the GPS stop socket command to stop the gps, and it is restarted in the same manner, with a socket command. This is done according to the recommendation of one of your your colleagues (see https://devzone.nordicsemi.com/f/nordic-q-a/61342/agps-and-psm)

There it is also stated that the AGPS and PSM doesn't interfere, and i therefore saw no reason to disable PSM bfore the supl session, and also no need to enable it after the supl session. It is already enabled. 

Tron

[Håkon Alseth]

Hi Tron,

 

When you exit PSM (by opening the connection to the SUPL server), you automatically enter PSM after the transmission is finished. However; there is a period between re-requesting PSM and entering PSM, to ensure that a UE is reachable in the network. This period of listening is unchangeable from the UE side, and in telenor; its 60 seconds of listening before entering PSM again.

 

Kind regards,

Håkon