• After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.

All Things GPS

Not looking good for the reboot fix.

Here are two runs (exactly the same track)...one with wireless ON (the southernmost route) and the other with it OFF (the northern route). Again, these were exactly the same route, but without wireless assist the GPS was totally crap. I even checked before the run to make sure I had a lock on 4 or 5 birds. In the open I locked onto 8 or 9.

Track 7 - Google Maps

Here is the exact same run, without wireless, with a reboot 2 minutes prior.

Track 9 - Google Maps

It's funny....the last part of the route (near the bottom) there is the best GPS performance I've seen out of the Epic without wireless. But it skipped out on all that data and the other data it did collect is in the wrong location.
 
SamsungVibrant said:
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.
Click to expand...

What? There are some well-documented GPS bugs on the Epic having to do with locking behavior, and another bug having to do with the bogus estimated accuracy level of 30.0 meters. (Likely the easiest bug to fix.) But there really is not a history of actual innacuracies for Epic phones that do lock well. To the contrary, actual accuracy is pretty good.

As for the reports from testers using the leaked Vibrant ROM, I am waiting to see if it exhibits the failure to lock after a period of GPS inactivity without a reboot. The early gushing reports from the field have not yet tested for that. This bug does appear on the Epic, and apparently on the Fascinate -- both phones that presumably incorporated Samsung's purported GPS fixes before they shipped. So it will be interesting to see if this new bug also propagates to Samsung's remedial fix for the Vibrant, Captivate and I9000.
 
4 walking test done today, accurate to 3 feet using google maps with satellite layers on, gps on and wireless networks off. GPS Test app, showed 8 visable and 7 using with an accuracy of 98.4ft.

There is a 30 second delay in walking test I noticed but was always with-in 3 ft, I used the sidewalk squares and steps in my apartment complex with layers to test, walking in all directions and 50 feet from end to end.

What I have noticed is that when I use both the gps and network, 3x it did the same thing, pulled me off course to the west by 15ft.

Using latitude, it shows my wife at work (epic 4g also), but about 30ft off from where she is, unkown if she really turned on her gps as I requested her to do.

Used google maps 5x for directions, accurate again.

Using just wireless networks, useless.

I'm very pleased with the gps.

Update: Did another walking test.
Wireless network only = 20 mile blue bubble with accuracy of about 1/4 mile.
Wireless Network + GPS (6 using) = no blue bubble and about 15-20 feet accuracy
GPS Only = Awesomesness of 3 ft accuracy.

Fact for me is that turning off wireless network and leaving gps on and not moving, 30 seconds later, 3 ft accuracy.

My Epic 4G GPS receiver works flawlessly, sprints triangulate not so much.
 
I tried my phone again this afternoon around 5:50 est and it couldn't lock the satellites, I had it turned on all day. I just rebooted and it did lock them this time, now around 6:30. Funny for me it seems to never work between 4:30 and 6-6:30 pm. I'm not sure if it is because the phone has been on all day, or because it has something to do with my location (philadelphia) and finding the sats around that time.
 
boomerbubba said:
As for the reports from testers using the leaked Vibrant ROM, I am waiting to see if it exhibits the failure to lock after a period of GPS inactivity without a reboot. The early gushing reports from the field have not yet tested for that. This bug does appear on the Epic, and apparently on the Fascinate -- both phones that presumably incorporated Samsung's purported GPS fixes before they shipped. So it will be interesting to see if this new bug also propagates to Samsung's remedial fix for the Vibrant, Captivate and I9000.
Click to expand...
You got it Brother. That is what I kept having to repeat with the first couple of tries of Sprint support, that was maddening!

Me: I have a found a serious problem with GPS than is only repaired by reset.
Sprint support: please reset your phone and see if you still have the problem.
Me: uhm, no. I am not calling you for that. I understand why with your varied userbase, you request that of people, I am cool with why you are asking me to do that. But please understand I am calling to report a problem that is fixed temporarily and only temporarily with a reset. Can you check known issues list? because it isn't there I need to get to the actual Sprint device team for Epic.
Sprint support: please reset your phone and see if you still have the problem.
Me: Thank you for your time.

Epilogue: A few days later I got to the device team though corporate
 
Aero1 said:
You got it Brother. That is what I kept having to repeat with the first couple of tries of Sprint support, that was maddening!

Me: I have a found a serious problem with GPS than is only repaired by reset.
Sprint support: please reset your phone and see if you still have the problem.
Me: uhm, no. I am not calling you for that. I understand why with your varied userbase, you request that of people, I am cool with why you are asking me to do that. But please understand I am calling to report a problem that is fixed temporarily and only temporarily with a reset. Can you check known issues list? because it isn't there I need to get to the actual Sprint device team for Epic.
Sprint support: please reset your phone and see if you still have the problem.
Me: Thank you for your time.


Epilogue: A few days later I got to the device team though corporate
Click to expand...

whats going on with your gps?
 
Aero1 said:
For one, the same problem every Epic has, which is that often it won't clear xtra cache thereby causing GPS often not to acquire without a reset.
Click to expand...

well you guys should start raising hell. The vibrant/captivate gps optimization update looks hopefull according to some people who have installed an unofficial leaked version.

epic is not set to get the optimization, so you guys should raise hell like captivate/vibrant owners did, that way your gps can be optimized also.
 
Out of curiosity, I did a walking accuracy test of the Epic 4G's GPS. Test conditions were walking along a known line (coordinates obtained through Google Earth), using GPS navigation with the "use wireless networks" option off. For what it's worth, this was relatively out in the open, with a few nearby buildings, with heavy overcast to light drizzling rain during the test.

For each position recorded, I took the longitude and calculated the expected latitude (the track was very nearly east-west). I then calculated the expected latitude for that longitude, and took the absolute value of the difference. For the 409 data points recorded, the mean difference between expected and recorded latitude was about 3.008 meters, with a standard deviation of about 2.003 meters. The single worst point was about 8.579 meters off.

Assuming that the longitude measurement is probably roughly as far off as the latitude measurement, multiplying the 3 meter error by sqrt(2) should provide a good estimate for the average total 2D positional error. This gives me an average error of about 4.254 meters, with a worst-case measurement of 12.133 meters.

The average altitude reported was -26.607 meters, with a standard deviation of 3.557 meters, so the altitude information is clearly unreliable.

Reported accuracy was exactly 30.0 meters throughout the test -- I've never seen it report anything other than that when it has a lock. It's definitely hardcoded.

Overall, it works (at least, mine does.) It looks like 2D position information is quite reliable to within 10 meters, and generally within about 3m or so. A nonlinear walking test around the parking lot seems to confirm this level of accuracy, even when walking along an arbitrary track.

Google's Driving Navigation also works quite well, although it definitely does anticipate "correct" movement -- I.E. showing the car starting to take a cloverleaf ramp as requested, when no turn was actually made. It always recovers and reroutes within a few carlengths, though.

I'm working on a more detailed analysis of the data, including pictures -- I'll have them up shortly on The Paleotechnologist .
 
i have not received or manually updated the latest firmware yet, but did have problem today using GPS. initially, i turned GPS on for google navigation and it picked up no problem. i scan the direction than exited navigation until i was closer to destination. a few miles into it, i fired it up again. well, i couldn't get a lock, actually, google navigation locked itself so i gave up and put the phone down. (luckily, my wife had her samsung instinct sprint nav on.) a few minutes later i heard the reboot sound. yep, the epic rebooted itself.

after the self reboot, i fired up the gps and google navigation and it picked up my directions within seconds.

anyway, i'm still not getting the latest OTA, so was curious to see what was going on at the sprint community site. lo and behold, they updated the list of fixes to include GPS.

how much of a fix, don't know. time will tell.

Samsung Epic:9/16 - Samsung Epic Maintenance Release...
Fixes:
4G de-registration when activating Mobile Hotspot
Won't maintain Exchange credentials (emails stuck in outbox and disappearing email accounts)
GPS performance enhancement
HotSpot 3G Idle Battery Issues
32GB SD Card Fix - previously when formatting the card, it wouldn't be recognized
 
SamsungVibrant said:
This cache problem has me worked...you guys think all these gps issues may truely be hardware related?
Click to expand...

I worry less about a hardware root cause than I used to. I tend to think this is a firmware/driver issue. But it may be a new bug, or a new variant of the old bug in Galaxy S phones, in low-level drivers written for Samsung by Broadcom for its GPS chip.

Note that the locking symptoms, which seem to be associated with the almanac and/or ephemeris cache, are easily replicable on the Epic and easy to work around by resetting the cache. These locking symptoms are also different from the locking problems on the stock Vibrant, which in turn were different from the locking problems on the international I9000. They were all broken, but in subtly different ways. Different firmwares, different symptoms. Mostly identical hardware, except the Epic has a CDMA radio and a different form factor.

I do think the worst of the Galaxy S GPS problems, including not just locking behavior but also accuracy issues, that plague the stock Vibrant, Captivate and I9000 seem to be fixed on the Epic. I think that is because the "optimized" GPS fixes from Samsung and Broadcom -- which Vibrants, Captivates and I9000s still await remedially -- got baked into the cake of the Epic and Fascinate when they shipped. But there are new and discrete bugs on the Epic. The picture on the Fascinate is less clear, but there are anecdotal reports that it has cache symptoms similar to the Epic.
 
You know, may captivate and vibrant owners, had working GPS at first, only to stop after a few days. this could be a trend that is not fixable with any firmware.
 
SamsungVibrant said:
You know, may captivate and vibrant owners, had working GPS at first, only to stop after a few days. this could be a trend that is not fixable with any firmware.
Click to expand...

Yes, I absolutely remember that, as well similar patterns in the I9000. There is some time-related bug in all three models, which may well involve the GPS cache for all we know. But their symptoms are different from what we are seeing in the Epic, where the failure to lock shows up fairly soon in easily replicable conditions and is always cured by clearing the cache. Also, the specific locking behavior on the stock Vibrant is different from the behavior on its older siblings. That is what I mean by the observation that the symptoms are similar, but still different, with different firmwares.

To me the most plausible theory is that Samsung, and possibly Broadcom, fixed one or more firmware problems but introduced another. As a previous owner of Vibrants and a current owner of an Epic who has tested both, I would much prefer the replicable bug in the Epic because it can be worked around so consistently. I am hoping that the new "optimized" update for the Vibrant is at least no worse than that, which seems apparent from testing of the leaked ROM. Because I am considering switching back due to unrelated problems on the Epic (poor network performance).
 
boomerbubba said:
Yes, I absolutely remember that, as well similar patterns in the I9000. There is some time-related bug in all three models, which may well involve the GPS cache for all we know. But their symptoms are different from what we are seeing in the Epic, where the failure to lock shows up fairly soon in easily replicable conditions and is always cured by clearing the cache. Also, the specific locking behavior on the stock Vibrant is different from the behavior on its older siblings. That is what I mean by the observation that the symptoms are similar, but still different, with different firmwares.

To me the most plausible theory is that Samsung, and possibly Broadcom, fixed one or more firmware problems but introduced another. As a previous owner of Vibrants and a current owner of an Epic who has tested both, I would much prefer the replicable bug in the Epic because it can be worked around so consistently. I am hoping that the new "optimized" update for the Vibrant is at least no worse than that, which seems apparent from testing of the leaked ROM. Because I am considering switching back due to unrelated problems on the Epic (poor network performance).
Click to expand...

ya, not trying to start a hspa+ vs 4g war, but you really can't beat T-Mobiles hspa+ network, although the vibrant is not a true hspa+ device, it does benefit from hspa+ regardless. T-mobile really has the best network when it comes to download speeds, many people are getting up to 7mbps on vibrant, just imagine the download speeds on the first true hspa+ device, the G2.
 
@SamsungVibrant,

As a p.s. to my last comment, it is possible that the leaked Vibrant ROM being tested by some folks also has symptoms similar to those of the Epic -- the worst of the old Galaxy S GPS symptoms are gone, but some discrete locking problems occur after a day or so. (I can't test this myself; I am just digesting the reports in the forums.)

That would make some intuitive sense, because this Vibrant ROM, like the initial Epic release, presumably contains what Samsung thought was a fix. Samsung said before release that the GPS on the Epic had been "tested and validated," and the Vibrant ROM is believed to contain the "optimized" GPS update.

If so, then Samsung may have a generalized new bug with the same root cause.
 
This is a universal thread to discuss the GPS on the Samsung Epic. If you are having trouble, need help or simply want to discuss the GPS this is the thread for it.

Please keep in mind the rules as well as the zero tolerance policy when posting.

All new GPS threads will be merged into this one.
 
boomerbubba said:
@SamsungVibrant,

As a p.s. to my last comment, it is possible that the leaked Vibrant ROM being tested by some folks also has symptoms similar to those of the Epic -- the worst of the old Galaxy S GPS symptoms are gone, but some discrete locking problems occur after a day or so. (I can't test this myself; I am just digesting the reports in the forums.)

That would make some intuitive sense, because this Vibrant ROM, like the initial Epic release, presumably contains what Samsung thought was a fix. Samsung said before release that the GPS on the Epic had been "tested and validated," and the Vibrant ROM is believed to contain the "optimized" GPS update.

If so, then Samsung may have a generalized new bug with the same root cause.
Click to expand...

well it seems like Vibrant won't be getting the GPS optimization in sept as promised, ugh. Obviously they are having a hard time trying to optimize the GPS. I guess software can only go so far to disguise a hardware problem. They need to just admit the real truth by now, be more transparent. I don't think the Captivate folks had any luck with their update either.
 
once locked my phones gps works pretty well and seems fairly accurate. however, its pain in the arse to get it to lock. shows 11 in view, and takes 3 minutes to lock 1, and and still waiting and waiting. it takes forever to lock on.
 
yes, boomer and I tested that workaround and convention usually has those type of fixes sticked.

In any case the the mode may wish to rename this the location thread, or create a completely separate as "use wireless network" triangulation system thread since there are two unrelated location methods: a) aGPS and b)use wireless networks." and they should not be conflated or confused.

Here are areas were some testing and discussion are needed:

1) aGPS cache issues (usually presented as seeing sats at good SNR, and not locking), confirmed present, workaround seems to fix for most.
2) SNR observations (stat seen at low SNR)
3) the estimated accuracy bug
4) user changes and testing: (examples)
a) changes to SNR minimum settings
b) changes to xtra cache settings
c) other changes or obseratiosn from StatusListener
5) open quesito nof stanaloen/autonomuos GPS capability

6) Issues related to "use wireless netorks. Examples: Are users stickig to cell site location when in typical more accurate forward link pr cell site triagulation conditons?
 
What exactly is the issue with the GPS? Is it with Google Nav or with Sprint Nav or both? I haven't seemed to have any problems other than once in a while it says "Cannot find location" or something of that sort.
 
EpicFan said:
What exactly is the issue with the GPS? Is it with Google Nav or with Sprint Nav or both? I haven't seemed to have any problems other than once in a while it says "Cannot find location" or something of that sort.
Click to expand...
It is not google nav or sprint nav that is the issue, but a Epic specific problem. Sprint has currently recognized a cache problem on Epic, where old "assistance" data used in properly reading the satellites is not removed from the Epic's memory (cache) system consistently. The bug is present on all Epics, but depending on habits you may see it often or rarely. Properly working aGPS phones can see that the data is old and get new data. The Epic sometimes cannot.

The Epic also reports false accuracy numbers. That is documented on all Epics. Some programs are dependent on those nubmers being proper,and Epic is breaking Androids rules by simply supplying a random number for accuracy.

Epic also seems to have constantly somewhat lower Signal to Noise ratio. If it is reporting those numbers correctly and bug free that means the Epic has lower inherent aGPS and GPS capability than competing smartphones. If you live an a nice open area this may affect you infrequently, but if you live and work in an urban area with variable weather this will result in more frequent loss of GPS signal and less accurate results than competing devices.

"Location based applications" (aka LBAs) -- those applications that use your location as reported by your phone -- are one of the fast growing applications sectors. There are literally hundreds of non navigation LBAs, and growing at a phenomenal rate, that need good location data to work.

GPS on cell phones is almost always better and much more capable than most dedicated devices, and that is why they are killing the dedicated GPS market (just like phones are taking a huge chunk out of mp3 player market, and indeed as smartphones are internet connected camcorders and cameras, out of those markets as well). So assessing location performance, and identifying for user or maker fix, any problems that can be fixed is an important activity.
 
After reading 4 large pages of posts, I'm a little hesitant to post this, but here goes.

I've got the opposite problem with my wife's new Epic. In her case, if the GPS is OFF, the location of the phone is incorrect by 24 miles when the phone is at home. When she's out and about, no tracking is done at all. Turn on the GPS (with wireless location on or off), the location is fine.

Anyone seen this behavior?
 
You must log in or register to reply here.
Back
Top Bottom