cloud-to-cloud A-GPS validity / caching

We are looking into setting up a cloud-to-cloud system to obtain GPS assistance data for the nRF91 in our product.

It seems that for A-GPS, the intended use is for our server to act as a proxy and redirect all queries to the nRF Cloud.

This is obviously necessary if the request include LTE cell information, as that will be different for each device.

However, is it possible to cache the A-GPS data on our server, when cell location is not used?

According to the documentation, the A-GPS data is valid for 4 hours. It should be possible to have our server poll the nRF Cloud every hour or so, and to provide that data to our devices. This would avoid the round-trip to nRF Cloud upon each connection.

Will the A-GPS data still work when it is an hour old? I'm concerned about the system clock data (type 7), and maybe the UTC parameters (type 1).

Finally, what is the difference between A-GPS and P-GPS with predictionCount=1 ? I suspect that would give similar performance and data sizes, is that correct?

Parents
  • Hi Stephan,

    You can cache the A-GPS data on your server. 4 hours is a little long, though. The longer you store it, the less accurate it becomes, increasing TTFF. I suggest an hour at most. Even shorter than that if using a mask. 

    A-GPS and P-GPS are similar, especially with predictionCount=1. A-GPS has the potential to be less data because you can request only specific data structures (P-GPS delivers everything by default). 

    Cole

Reply
  • Hi Stephan,

    You can cache the A-GPS data on your server. 4 hours is a little long, though. The longer you store it, the less accurate it becomes, increasing TTFF. I suggest an hour at most. Even shorter than that if using a mask. 

    A-GPS and P-GPS are similar, especially with predictionCount=1. A-GPS has the potential to be less data because you can request only specific data structures (P-GPS delivers everything by default). 

    Cole

Children
No Data
Related