All Things GPS

[Aero1]

I don't get a no lock state, in the house i do or in buildings but it can't see the sats, how is it going to lock?

There are unique Epic problems obtaining location using wireless network locations on in non satellite conditions. Most spmartphones never have a problem with this as they will drop down tower based fix after some short period of seeing no satellite, or in better implementations estimate first and give you a tower-based fix withing moments inside a concrete bunker, and then immediately look for satellites. So my question is: when you get no location in your home, what happens if you turn on wireless location, and then restart the phone? On every non epic spartphone I have I get a estimated fix (android "location.coarse"); on Epic this feature is not working for me half the time, Is that the case with you?

Just walked outside my house, rebooted the phone and had USE WIRELESS NETWORKS off (double checked) Activated GPS test and had 8 birds with 7 in use within 15 seconds. Max SNR was 23 and as low as 16.

With the known problem of Epic not refreshing cache, that indicates use within four hours or a reset cache from power cycling the phone. so the example doesn't test the promblem Sprint is trying to get samsung to fix.
The low SNR you report is, while perhaps casual of the Epic software bugs, a tertiary problem, as I don't think the they can do much about that nominal hardware performance, it is in theory adequate, it will just mean once Samsung fixes the bugs, the Epic will have slightly more lag than competing devices and failure to fix under poor conditoons where for example Evo and other dievices will work.

K i just checked my wifes Epic standing outside I only got 3 sats to lock on with 8 in view. On my Evo in the house I use 5 sats with 7 in view. I'd say the GPS on the Epic is a Fail.
Does anyone know if this is a software issue? I will say this at least the Epic only uses the GPS when you use an App that needs it.

This is exactly one of the examples Sprint and we are seeing. The one comprehensive review of Epics GPS, which is on ananetch, shows pictures of exactly this. Failure to use most of the satellites in view could easily be old and bad ephemeris retained in cache. here is the test we would like you to try: When you encounter this specifically, immediately test again with a reboot of your epic and let us know.

 

[robl45]

over and over again, day after day, it locks within 10 seconds, sits all night, locks next day no problem.

There are unique Epic problems obtaining location using wireless network locations on in non satellite conditions. Most spmartphones never have a problem with this as they will drop down tower based fix after some short period of seeing no satellite, or in better implementations estimate first and give you a tower-based fix withing moments inside a concrete bunker, and then immediately look for satellites. So my question is: when you get no location in your home, what happens if you turn on wireless location, and then restart the phone? On every non epic spartphone I have I get a estimated fix (android "location.coarse"); on Epic this feature is not working for me half the time, Is that the case with you?


With the known problem of Epic not refreshing cache, that indicates use within four hours or a reset cache from power cycling the phone. so the example doesn't test the promblem Sprint is trying to get samsung to fix.
The low SNR you report is, while perhaps casual of the Epic software bugs, a tertiary problem, as I don't think the they can do much about that nominal hardware performance, it is in theory adequate, it will just mean once Samsung fixes the bugs, the Epic will have slightly more lag than competing devices and failure to fix under poor conditoons where for example Evo and other dievices will work.
This is exactly one of the examples Sprint and we are seeing. The one comprehensive review of Epics GPS, which is on ananetch, shows pictures of exactly this. Failure to use most of the satellites in view could easily be old and bad ephemeris retained in cache. here is the test we would like you to try: When you encounter this specifically, immediately test again with a reboot of your epic and let us know.

 

[Aero1]

robl: You posted you are getting no lock in house with wireless location on. that is a broken function. I can't even think of a smartphone that cant easily get a location estimate in a concrete bunker as long as it it getting one single bar on its Vonnie channel.

It is a legal requirement and Samsung could not have passed FCC if it cant get a location estimate inside your home. It is the software that is busted in delivering that estimate to your application.

What the Sprint top device spuoort think has happened is that an ad hoc bunch of gps and location driver work-around are causing these multiple problems under various conditions.

So the broken function you are describe is one of the ones Sprint is trying to get samsung to fix.

As far as driving and the general observation that most smartphenes are more accurate than your handheld Garmin, that is true. Smartphones get faster fixes and do better in poor satellite conditions thatn the typcial garmin or tom tom. Epics is one the first one that doesn't.

FYI the build date of our one Epic with somewhat better SNR performance is August 18, the others are Aug 10, the battery build date on the new one is also two weeks newer and its voltages are costantly higher. Can you guys who are also reporting SNR numbers check your respective build dates?

 

[Aero1]

#birds was between 8 and 11 (inside car while moving) and typical use was either one or two less than seen or the same as seen. Tried with USE WIRELESS NETWORKS both on and off and saw absolutely no difference.

yet you also write you are getting terible lock in another thread? three in use is a terrible number

In all cases, 9-11 IN VIEW, 3 IN USE and no matter where I am, a 98.4 foot accuracy...so my guess is that this # is burned into the phone by Samsung

 

[robl45]

i dod get a fix inside the house, but no lock on satellites. do you get a lock inside with gps status showing sattellites locked?

robl: You posted you are getting no lock in house with wireless location on. that is a broken function. I can't even think of a smartphone that cant easily get a location estimate in a concrete bunker as long as it it getting one single bar on its Vonnie channel.

It is a legal requirement and Samsung could not have passed FCC if it cant get a location estimate inside your home. It is the software that is busted in delivering that estimate to your application.

What the Sprint top device spuoort think has happened is that an ad hoc bunch of gps and location driver work-around are causing these multiple problems under various conditions.

So the broken function you are describe is one of the ones Sprint is trying to get samsung to fix.

As far as driving and the general observation that most smartphenes are more accurate than your handheld Garmin, that is true. Smartphones get faster fixes and do better in poor satellite conditions thatn the typcial garmin or tom tom. Epics is one the first one that doesn't.

FYI the build date of our one Epic with somewhat better SNR performance is August 18, the others are Aug 10, the battery build date on the new one is also two weeks newer and its voltages are costantly higher. Can you guys who are also reporting SNR numbers check your respective build dates?

 

[Aero1]

i dod get a fix inside the house, but no lock on satellites. do you get a lock inside with gps status showing sattellites locked?

There is no condition where a functioning smartphone cant discover at least one estimate of its location even in zero GPS conditions

I get a location reported to software (google maps) within 30 seconds inside my house on Evo, Palm Pre, HTC Touch Pro2 Htc Touch pro, Palm Treo Pro, iPhone, and Droid -- but the Epic and only the epic will not.

Do you understand that working smartphones absent a sat signal can and will give you a 20' to 3000 foot fix?

This has to do with faulty drivers or GPS stack on the EPIC. Waht the location finding fo the phone does abset a satellite is part of the same issue probably deriving from same buggy drivers becasue presence or absence of sats.

Here is how it suppsoed to work, on of these two ways:

1. Application calls on smart-phone for location information
2. Device attempts satellite fix first, if it can get a fix that way it does, if not, but it can find one poor satellite signal, attempt hybrid using satellite and and towers, if it can't find satellite at all, it uses any of the four methods that use no satellites.
3) reports best available lcoation data to application (preferring SM tagged fine over coarse)

or

)1. Application calls on device for location information
2. DEvice attempts tower estimate first, and then progresses up to the most accurate fix attempting it with GPS
3) reports best available location data to application (preferring SM tagged fine over coarse)

 

[robl45]

i get a fix in my house with a big red circle, but thats not sat lock. still this thing seems to work fine on my phone. it locks almost instantly all the time. finds me on google maps. just nav is not super accurate. im past the street when its telling me to turn.

There is no condition where a functioning smartphone cant discover at least one estimate of its location even in zero GPS conditions

I get a location reported to software (google maps) within 30 seconds inside my house on Evo, Palm Pre, HTC Touch Pro2 Htc Touch pro, Palm Treo Pro, iPhone, and Droid -- but the Epic and only the epic will not.

Do you understand that working smartphones absent a sat signal can and will give you a 20' to 3000 foot fix?

This has to do with faulty drivers or GPS stack on the EPIC. Waht the location finding fo the phone does abset a satellite is part of the same issue probably deriving from same buggy drivers becasue presence or absence of sats.

Here is how it suppsoed to work, on of these two ways:

1. Application calls on smart-phone for location information
2. Device attempts satellite fix first, if it can get a fix that way it does, if not, but it can find one poor satellite signal, attempt hybrid using satellite and and towers, if it can't find satellite at all, it uses any of the four methods that use no satellites.
3) reports best available lcoation data to application (preferring SM tagged fine over coarse)

or

)1. Application calls on device for location information
2. DEvice attempts tower estimate first, and then progresses up to the most accurate fix attempting it with GPS
3) reports best available location data to application (preferring SM tagged fine over coarse)

 

[SamsungVibrant]

There is no condition where a functioning smartphone cant discover at least one estimate of its location even in zero GPS conditions

what do you mean?

 

[RickJ]

what do you mean?

The rest of the post explains it, so long as you have cell coverage, no matter if your gps is working or not, you will get a location within 3000 feet. If you dont, something is broken.

 

[boomerbubba]

The rest of the post explains it, so long as you have cell coverage, no matter if your gps is working or not, you will get a location within 3000 feet. If you dont, something is broken.

That last-ditch rough location does fall back to pure, dumb tower location. And that requires enabling "use wireless networks." As far as I can tell, that works on the Epic and other Galaxy S phones. It just is not GPS location, and of course is not very accurate.

 

[Aero1]

That last-ditch rough location does fall back to pure, dumb tower location. And that requires enabling "use wireless networks." As far as I can tell, that works on the Epic and other Galaxy S phones. It just is not GPS location, and of course is not very accurate.

Tower only methods can e very inaccurate >5000m or very accurate <30'.

There are about 10 or 15 methods of using towers only. Some methods work on some signal types (gsm vs CDMA), some a carriers implement some but not others, some phone and gps chip makers implement or don't implement some, some are MSa and some MSB, and sometimes a given phone or OS effects which they can use. On top of that there are hybrid methods which use less than four sats plus one or more towers. Those are excellent when you are getting one or two sats indoors but use tower timing etc to help reject multipath.

Any given smartphone should be able to use several tower only methods. With well written software it would deliver the first: where is the tower I am conencted to? And then refine it as able: Where am I in between the three closest towers.

My epic is very often not getting a fix inside with "use wireless" ticked on, and "use GPS" ticked on or off where I am getting very good EVDO and CDMA signals and where I am in a well covered area. I am troubleshooting whether my airave may have an affect. Sometimes it works and sometimes not.

 

[tomw3141]

Sprint Airave can easily mess up any tower based methods. The problem is that it reports its location which your ISP reports - and that is often some major switch facility. In my case, because Verizon provisions my ISP it is about 80 air miles away from my home. I have a Samsung Moment and was about to get an EPIC when I found out that GPS is a problem with the EPIC Samsung product as well. The only way I get correct location at home is to turn off RF (turn off the Phone function) turn on GPS location, go outside, and wait 15 minutes to a half hour for the deaf GPS to lock to four satellites. My Garmin locks indoors in seconds to a half dozen or more satellites. What really grinds my gears is that the Airave has to have a GPS lock to even work! AND it doesn't report where it is! Figures, its built by Samsung. Sorry for the rant folks.. I will shut up now.

 

[boomerbubba]

Tower only methods can e very inaccurate >5000m or very accurate <30'.

There are about 10 or 15 methods of using towers only. Some methods work on some signal types (gsm vs CDMA), some a carriers implement some but not others, some phone and gps chip makers implement or don't implement some, some are MSa and some MSB, and sometimes a given phone or OS effects which they can use. On top of that there are hybrid methods which use less than four sats plus one or more towers. Those are excellent when you are getting one or two sats indoors but use tower timing etc to help reject multipath.

Any given smartphone should be able to use several tower only methods. With well written software it would deliver the first: where is the tower I am conencted to? And then refine it as able: Where am I in between the three closest towers.

I'll defer to your experience about hybrid methods on other state-of-the art phones used indoors. My recollection is that my G1, using two-year-old technology, would use a dumb single-tower circle, just as my Epic does.

FWIW, whether or not I get a fix indoors on my Epic varies completely with the environment. In an apartment setting, close to a window, its ability to get a lock is marginal, just like my G1's always was. Inside my concrete multistory office building, nothing. But sitting in my father's ranch-style home with a tar-and-gravel roof, the Epic would get a good lock on multiple satellites and pinpoint my location accurately. So all of that seems to be a function of attenuated satellite signal strength, not towers. Of course, if I set "Use wireless networks" on, I get a big circle around the closest tower.

 

[Aero1]

I'll defer to your experience about hybrid methods on other state-of-the art phones used indoors. My recollection is that my G1, using two-year-old technology, would use a dumb single-tower circle, just as my Epic does.

FWIW, whether or not I get a fix indoors on my Epic varies completely with the environment. In an apartment setting, close to a window, its ability to get a lock is marginal, just like my G1's always was. Inside my concrete multistory office building, nothing. But sitting in my father's ranch-style home with a tar-and-gravel roof, the Epic would get a good lock on multiple satellites and pinpoint my location accurately. So all of that seems to be a function of attenuated satellite signal strength, not towers. Of course, if I set "Use wireless networks" on, I get a big circle around the closest tower.

On the second point referring to GPS inside, often what you see is less signal strength, plus slighltly multipath, plus the problem of standing still which amplifies multipath problems. Most GPS receivers have more work to do with multipath standing still or moving slowly than moving fast. They do have advanced multipath rejection but the same multipath is harder to deal standing still.

On the first point, tower only use, while there are exceptions, AFLT is the rule on recent Sprint smartphones. There are FCC standards on accuracy and yearly increasing accuracy demands benchmarks for handets sold today are much higher single tower location. Carriers pretty much have to use AFLT or E-OTD which use multple non GPS signal sources and use timing measurments of signals to various towers isntead of simple triangulation. Of course there is not requirement that those better locations rerequired by the FCC E911 benchmarks be reported to the LBA (the applications on your phone useing GPS) or even the MS (your phone), but they are there. For example in my Palm Treo Pro (made actually by HTC), in zero GPS RF you could see it step down (better) in accuracy if you switched it on and got location inside to the point of excellent accuracy with no GPS. Of course thoise excellent accuracies at a standstill with non GPS degrade if moving.

My main point is lots of smartphones today (most?) are doing much better than simple reporting of closest cell tower's postion.

My treo pro and the Evo seem to have AFLT it from the accuracy.

Roughly speaking, absent four birds, there are a dozen or more tower only methods, and half dozen using tower and one bird (hybrid).

Cell site location, reporting to the MS just the cell site itself
Cell site location plus signal strength
CEll site location plus a timing signa
Two cell sites
three site triangulation
E-OTD (timing)
three site AFLT

 

[boomerbubba]

...My main point is lots of smartphones today (most?) are doing much better than simple reporting of closest cell tower's postion....

Thanks for the primer. Googling around, I see several references about AFLT and hybrid technologies.

Heretofore I have focused my own testing on pure GPS and A-GPS accuracy outdoors. And to keep things simple (see the title of this thread) I have usually run my phone with the "Use wireless networks" setting off. But obviously, for testing whatever non-GPS technology is onboard the phone while indoors, I have to turn it on. (One should not expect AFLT, a technology based on cellphone networks, or any hybrid cell/GPS methods to work with this setting off.)

So for testing indoors, I will turn my "Use wireless networks" on. I don't have enough experience with this testing to comment yet on the results.

[EDIT: Well, I think there is definitely some sort of non-GPS network technology beyond single-sell-tower being supported on my Epic. With "Use wireless networks" on, and GPS satellites disabled, while sitting in my apartment I get an estimated location centered very accurately on my actual location, with a circle of uncertainty displayed in Google Maps. In fact, when I reenable the GPS and it has only a very marginal satellite fix, that fix is less much less accurate than the network-based centerpoint estimate. Since this network estimate also works with the GPS receiver turned off, I assume it is AFLT technology. Can't tell if there is any hybrid voodoo being supported with both turned on.]

 

[Wrxdrunkie]

darn man, this thing wont even get a lock on my location if I do not use wireless networks. Its so annoying, when I need directions I need them now, I dont want to be waiting for 3 minutes for my phone to get a gps lock...

 

[Aero1]

Yes I get the same thing, when I get a non GPS fix. the best fix that way usually isnt bad and is clearly either AFTL (one of the most accurate tower only) or another similar advanced method. What I am finding is that often I get no tow fix which makes no sense. It may have to do with the prioritizing. NO phone is perfect in that sense (although my treo pro is darn near perfect using best available for any RF situation). But i think there is some bug. I am trying without my airave and getting what seems soemwhat better results but still often failing to get a fix inside with wireless nets on and gps off anyway.

here are my observatiosn on possible bugs/problems:

1.Sticking Cache: this is affecting GPS mode and can be fixed with a power cycle. Symptoms several stats seen non in use.
2. partial expire/sticking cache. My symptom for this is eight to eleven seen and yet only three in use.
3. Failure to get any location in CDMA range using wireless nets only: This is not occurring most of the time but often enough when it should not.
4. Below Average Signal to noise
5. Bug in api or stack reporting of accuracy. Symptom is 98' or 30 meters confidence range no mater what.

 

[SamsungVibrant]

GPS optimization update will be out soon for vibrant and captivate, will let you guys know how it goes.

 

[jfenton]

It appears that over on Verizon the Fascinate has the issues and IMHO, that doesn't bode well given that it was just released.

 

[jfenton]

Is this a Samsung GPS "optimization" or something a private developer is putting out?

 

[boomerbubba]

It appears that over on Verizon the Fascinate has the issues and IMHO, that doesn't bode well given that it was just released.

I've been reading several of the Fascinate forums, and it is not at all clear what is going on. Some end-users are having GPS success, and others are not. Verizon changed the nomenclature of its GPS configuration settings with no explanation or reasonable default, and most of the problems might be explainable as pilot error.

 

[Aero1]

GPS optimization update will be out soon for vibrant and captivate, will let you guys know how it goes.

Let's not hope (fr teh sake of both Epic or captivate users) that the Epic problems are the result of us beta testing their fixes!

 

[boomerbubba]

GPS optimization update will be out soon for vibrant and captivate, will let you guys know how it goes.

Are you planning to flash the leaked Samsung ROM that reportedly includes GPS fixes (maybe the final version, maybe not) or will you wait for the official OTA for your Vibrant?

 

[Inspekt]

This is exactly one of the examples Sprint and we are seeing. The one comprehensive review of Epics GPS, which is on ananetch, shows pictures of exactly this. Failure to use most of the satellites in view could easily be old and bad ephemeris retained in cache. here is the test we would like you to try: When you encounter this specifically, immediately test again with a reboot of your epic and let us know.


I've been lurking in the forums reading about these problems, which I've also experienced since getting the phone on Sept. 1, and I can't take it any more. I've had the no lock state several times and been unable to get lock on the sats even after rebooting. The wireless networks option is turned off and it will not lock even if I turn it on and reboot.

I've noticed it happen 4 times today between 4-6:30 pm (not sure if the time matters) and I live in Philadelphia. Today, it was kind of rainy and overcast and the phone could see up to 11 sats, with SNR up to 35 on some, and could not lock. I rebooted the phone 4 times and no luck, with about 1 hour passing in between all this. As I write this right now, I have 8 sats in view with SNR between 20 -27 and I cannot lock onto any of them, so I guess that makes 5 times...

I have no idea why the phone is behaving this way. I've used it on other overcast days and when there was slight rain with no problems, but it seems like the issues have become more frequent in the last week or so. I check the GPS obsessively after reading about all these problems and typically it does an OK job of locking onto sats (locks onto 75% of the sats in view typically), sometimes it takes 30-60 seconds, and I've never had a problem getting location for more than 15 or 20 minutes before.

When I do get lock, the accuracy is always reported at 98.4 feet, which several others have mentioned.

I like this phone, but honestly I'm not paying $250 bucks for a POS GPS. Unless there's a fix in the next week or two this sucker is going back to the store, especially after reading about the samsung support horror stories.

Well I gave it one last shot rebooting the phone, but still the GPS isn't doing squat. It's nice to know there's 7satellites out there, 4 with SNR 25-31 (strong enough to get a position every time from this location before), it's just too bad this phone can't talk to one of them. I just tried it with wireless networks on, hoping it would help it to get a fix, but still no luck....

 

[SamsungVibrant]

Let's not hope (fr teh sake of both Epic or captivate users) that the Epic problems are the result of us beta testing their fixes!

Lol i dunno if Samsung used epic users as a beta test...but from the looks of the leaked unofficial vibrant Rom...gps on vibrant doesn't suffer from that large inaccuracy levels you guys are getting on epic.