thoughts on A-GPS offline

[Joerg Reisenweber]

Am Mi  18. März 2009 schrieb Helge Hafting:
> Daniel Willmann wrote:
> > On Tue, 10 Mar 2009 15:18:23 +0100
> > Helge Hafting <helge.hafting at hist.no> wrote:
> [...]
> >> It was just to be safe. The documentation states that you might not
> >> get a fix _at all_ if either position or time is outside the claimed 
> >> accuracy. Now, maybe it works with 3km anyway after the fixes that 
> >> prevents the chip-crashing exception. I happen to live about 6km away 
> >> from where I work, so 9km was a nice safe value. The default is
> >> 300km, and "100km allows a more optimistic startup." Perhaps such
> >> rough estimates is all that is needed, if it is only used to figure
> >> which satellites that can be seen.
> > 
> > If you don't mind testing please try changing pacc to 100km and see if
> > it affects TTFF adversely in your case. If not we could just use that
> > as a default.
> 
> Shouldn't be too hard to test.
> 
> I think I know one side of having accurate pacc:
> 
> The first fix can happen with only two satellites. I have seen this 
> happen several times. It surprised me at first, but it makes sense. With 
> two satellites (and a reasonable clock), you get a big circle of 
> possible positions. But then there is the data from the "approximate 
> position". It puts you at some height above sea level. The big circle 
> intersects the earth surface at some angle, so with height, we now have 
> two possible spots instead of a big circle. Usually, only one spot will 
> be close to the approximate position, so that is where you are.
> 
> That is an "optimistic startup scenario". A too spread out pacc means 
> both possible spots are within pacc range, and the FR will have to wait 
> for a third satellite to break the tie.
> 
> If you travel a long way and still report the old position with a fake 
> precision pacc, then you might be close to the other of the two possible 
> satellite-based positions. You could then get a fake fix on the wrong 
> spot. As you and/or the satellites move, the wrong spot will move around 
> in strange ways at strange speeds. When more satellites show up, the 
> device might get really confused if it keeps trusting the approximate 
> position. Perhaps even rejecting them as "reflected signals" for a 
> while. Of course, only the manufacturer will know the exact details of 
> what might happen.
> 
> Helge Hafting

Wow, I had to read your posting twice to get how brilliant this analysis is.

cheers
jOERG
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/community/attachments/20090318/01fdf223/attachment.pgp