[Openstreetmap] Re: Best time of day for urban canyon performance

[Schuyler Erle]

* On 29-Mar-2005 at  4:27AM PST, SteveC said:
> 
> But, if you need a 60 degree cone up to get the minimum then its
> difficult because modestly built up areas I was wandering around don't
> even have that. So the most simple case of a cone doesn't seem super
> useful.

I suspect that knowing when the satellites are directly overhead still
has *some* utility.

> * You want good *vertical* accuracy. Hmmm.. .would need to turn sat
> position in to some measure of DOP for each dimension.

Yes. I've seen some example Java code lying around to do this, but I
haven't had time to get into it in any depth. Can't seem to find it in
my delicious, though. :-/

> * Instead of looking at a cone, assume you're in the middle of an
> infinite street that runs in a straight line and do the visibility
> analysis for that.

I've thought about this, based on our experiences collecting tracks in
Milano, where the streets are fairly narrow and the buildings high
enough to obstruct a lot of GPS reception. We did indeed find that
certain times of day, it was more advantageous to collect data on
east-west streets and other times on north-south streets. This is a
potentially interesting line of analysis.

SDE