Google Maps thinks I'm in Seattle WA - why??

It may be the 3G triangulation that is causing it: put it into Airplane mode then try to do the locate.

There have been frequent reports of Seattle - Detroit mix ups.
This is the fault of the Cell Towers, where the iPhone gets its first hint about where you are. The towers are broadcasting the wrong info.

So the solution is to go out side, so the phone can have a clear view of the sky
and give it LOTS of time to get a fix. It takes longer when it has to work out that the hint give by the towers was wrong, but it will eventually get it right.

Also try turning off 3g, and see if the EDGE towers report a correct hint.

If that helps, grouse to your local ATT offices that their towers are wrong.

Airplane mode disables GPS.

What could be incorrect is the A-GPS server's database, which usually looks up the cell id to give back a first location estimate (along with the satellite assistance info).

A GPS receiver requires two sets of data with which to determine location in conjunction with the real-time signals received from the satellites. These are known as almanac and ephemeris data. Normally (a non-assisted GPS receiver), these are obtained from the GPS satellites. Worst case, it can take 12-15 minutes for these files to be downloaded from the satellites.

With Assisted-GPS (as found in the 3G), the device will attempt to get a partial set of almanac/ephemeris data from the A-GPS server, based on where the iPhone thinks it is. This data is transmitted to the iPhone over the cellular data connection (3G or EDGE), or a WiFi connection, whichever is available. If your nearby AT&T cell tower is not correctly identified (bug in the tower CellID info, or an error in the A-GPS server), the iPhone will be given bad ephemeris data. It will then calculate its location based on signals from the GPS satellites and the faulty ephemeris data. That could explain why Detroit becomes Seattle.

Based on what's been reported in the many discussion threads concerning broken GPS, it appears that once the iPhone has obtained (faulty) assist data from the A-GPS server, it doesn't bother to keep listening to the data broadcast from the sky so it never receives correct ephemeris/almanac data.

Like many others, my 3G lost its GPS function after the update to v2.1. I eventually got it working by removing SIM card (so as to disable EDGE and 3G data), turned off WiFi and went outside for 20 minutes. Eventually, the GPS locking in to my location, with the pulsing blue dot. With no A-GPS server to fall back on, I surmise that the GPS receiver had downloaded the almanac/ephemeris data, and was thus able to obtain an accurate GPS location. Then I put the SIM card back in. My Location Services has been working fine since.

@MobileDev - it would be helpful if you could take your "Seattle-based" 3G outside (in clear view of sky) and wait a while to see if it eventually obtains the sky data and realizes that it is really in Detroit. That would confirm if this hypothesis is correct.

@MobileDev - it would be helpful if you could take your "Seattle-based" 3G outside

I think you meant @Goldndoodle 🙂

Based on what's been reported in the many discussion threads concerning broken GPS, it appears that once the iPhone has obtained (faulty) assist data from the A-GPS server, it doesn't bother to keep listening to the data broadcast from the sky so it never receives correct ephemeris/almanac data.

I think that's a great theory. Except perhaps...

With no A-GPS server to fall back on, I surmise that the GPS receiver had downloaded the almanac/ephemeris data, and was thus able to obtain an accurate GPS location. Then I put the SIM card back in. My Location Services has been working fine since.

While the almanac is good for about six months, the ephermis data is only good for about four hours. So, unless your location has been drifting on you, it must've obtained that data again and again... either from an AGPS server or the satellites themselves.

Still, it's easy to visualize a bug in the software, that needs initializing on its own in order to prevent your scenario from occuring. So your theory is not negated.

12.5 minutes to download the full almanac

Slight correction. It takes that long if it only listens to one satellite. With multiple channels, it should be able to more quickly collect the almanac.

Regards.

@MobileDev,
Thanks for the correction re download times for the complete almanac. You are quite right - if the iPhone has access to multiple birds, it can get the full table quicker. I read somewhere (but can't remember where) an opinion that the iPhone does not have a full 12-channel GPS receiver to save cost/space/power. Do you/anyone know how many channels are in the 3G?

Cheers.
Dave

The specs of the PMB2525 chip are on infineon's web site. http://www.infineon.com/cms/en/product/findProductTypeByName.html?query=PMB2525 (in the pdf linked on that page).

There is also other info about this chip embedded in the tech specs of another product which you might find far more interesting:
http://www.lingyitech.com/Preliminary%20DatasheetTrinity_v1.008172007.pdf
That document reveals power consumption and time to acquire (which is sub 1 second if the gps has been running in the background). The PMB2525 keeps it own 4 day almanac if you leave GPS switched on.

It does have 12 channels, but this hardly matters.

Each bird talks on two frequencies using a code division multiple access (CDMA) spread-spectrum technique, and each bird transmits a unique pseudo random sequence known as a Gold code which the receivers can use to distinguish one from another.

At most latitudes you seldom see less than 6 or more than 9 birds at once, and calculations get very time consuming using more than 6 birds, so most chipsets select the best 4 for fast acquisition and only use others for refinement.

So even an 8 channel chipset would be quite sufficient.

I did confirm your theory before the problems were fixed though.

Thanks for running those tests. Saved me a Drive to Detroit.

On another note, I've discovered that when only using Edge (or properly configured 3G towers) you will usually get a faster more accurate fix with WiFI turned off than with it on.

Unless of course you are sitting 4 feet from an AT&T hotspot. Most neighborhood routers are not accurately mapped. (Nor do I feed pre-disposed to submit my router for mapping at skyhook.com)

I've also discovered that it will use wifi routers to get an A-GPS hint even if it is not connected to those routers.

@icebike,
Thanks a lot for the pointers. Great material - essential reading for anybody looking for more detail on what's under the covers of GPS.

@goldndoodle:
Thanks for posting your results. Glad to hear that you've found yourself 🙂. Presumably, AT&T will eventually get all their towers correctly located in the database. It's not clear if the problem was with the towers (base station controllers), or was a database error further up the food chain somewhere.

icebike wrote:

On another note, I've discovered that when only using Edge (or properly configured 3G towers) you will usually get a faster more accurate fix with WiFI turned off than with it on.

I have a theory that if the WiFi is on, the iPhone sniffs the WiFi environment, looking for any access points it can find, and collecting their SSID and MAC addresses. It then sends that information off to Skyhook, along with any other info it can determine about your present location (obtained from cell tower ID and/or GPS). Thus, the SkyHook database would get automatically updated with info about the geographic location of WiFi nodes.

Only a theory, but it would sure be a clever way to keep the Skyhook database updated for free. Millions of iPhones roaming the planet, reporting back to headquarters about where they are, what WiFis are where, what cellular carrier they are connected to, what tune is being played...

I've wondered about that myself.

This would certainly warrant public disclosure if it were the case. I'm not exactly a tinfoil hat type, but having my home router in someone's database without the option to opt-out seems a bit over the line.

SkyHook have an opt-in option on their page where you can submit your own router, but this seems very prone to people poisoning their database with bogus location data.

I'm convinced the iPhone looks up every access point mac address it can see when trying to determine your location, but I'd be a tad miffed if someone else was writing down all those locations.

icebike wrote:
SkyHook have an opt-in option on their page where you can submit your own router, but this seems very prone to people poisoning their database with bogus location data.

When I first got an iPod Touch several months ago, I was astounded that it instantly "knew" where I was, within a couple of houses. The small blue circle was centered pretty much on a neighbor's house, wherein lies one of the few WiFi APs in my neighborhood. I am 100% certain that my neighbor did not submit her router to the Skyhook database. So either someone else did, or, more likely, it was picked up on a drive-by scan made by someone on contract to Skyhook.

So the fact is, your home router is going to get entered in the database without you doing/knowing it, since those pesky radio waves travel outside your property. If it can get captured by a war-driver cruising the neighborhood, it might just as well get captured by someone with an iPhone. If I owned the database in the sky, that's the argument I'd make that it's a legit way to collect the data.