-
After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.
All Things GPS
- Thread starter SamsungVibrant
- Start date
-
- Tags
- samsung epic 4g
jayhawksean
Newbie
hmmm, i am in Grand Rapids, MI. we have 4G. i tried having the GPS get me home from work downtown. it took about 5 minutes to find the satellites and start navigating using Google's Navigate. i liked being able to speak my address and that worked just swell. however, the phone was telling me i was on roads i'd never heard of and then it would put me accurately on the road i was on, but then immediately switch to another road and explain how to get back to the road i was already on. i gave it another 5 minutes and then gave up. very disappointing.
Yep, sadly, you are mistaken.
Some current cell phones might claim to get you to 10ft of accuracy but without WAAS its not routinely happening on any of them as anything more than luck.
WAAS is simply an error correction mechanism. It yields benefits to both altitudinal and lateral references, but the lateral reference is what is was created for (this is an FAA thing ... measuring altitude accurately has never needed GPS!).
A good, dedicated, GPS receiver with a proper antenna, WAAS support and decent conditions will get you to 10 feet. Really good receivers (which is much about clock error as anything) in optimal conditions, manage 3 feet.
If your cell phone is routinely achieving (not reporting ) 33 feet then that is pretty damned good.
It's all in the specs
No, not sadly, but yes perhaps mistaken. Ill have to go read up on it.
Not sure how to respond to the 10-33 statement exactly because I dont wanna be part of the mine does your doesnt crowd. I can say that using maps does always show well better that 33' accuracy when in open unobstructed areas and tracks very well, I'm sure alot of others with various devices have the same experience. I havent made the comparison with My Tracks though users here claim the accuracy is much better than the 33' as well.
Some claim that their device will correctly show their location indoors (in what room they are in) that I would agree with as luck.
Not sure how to respond to the 10-33 statement exactly because I dont wanna be part of the mine does your doesnt crowd. I can say that using maps does always show well better that 33' accuracy when in open unobstructed areas and tracks very well, I'm sure alot of others with various devices have the same experience. I havent made the comparison with My Tracks though users here claim the accuracy is much better than the 33' as well.
Some claim that their device will correctly show their location indoors (in what room they are in) that I would agree with as luck.
It's not about mine does/yours doesn't, it is, perhaps, about failing to understand what the accuracy ratings tell you.
That 33 feet accuracy is a margin of error. So you can be anywhere inside that region, including right on the very spot the GPS shows you at. That's not the same as having assurance that that is ALWAYS what the GPS is going to show you.
GPS works, fundamentally, using time references, very accurate clocks, relative distances and some clever math (that even takes into account things relativistic effects like time dilation). Given a good radio, GPS accuracy comes down to how accurate your clock is relative to the clock data encoded in the GPS (and WAAS) signals you receive.
Try the same test when the clocks are slightly off, which will happen cyclically, (bear in mind 1 second net error would put you out of position by ~300,000 km!) and you'll find you're at the limits of that margin of error.
On average, over a number of controlled tests (not just looking at a dot on a map, we're talking directly comparing several significant digits of long/lat) you'll find that the net accuracy without WAAS is, at best, 33 feet.
That quote from Garmin says the same thing I am. It's rounded differently, but the numbers are fundmentally the same.
10 feet with WAAS, not without.
Guess which GPS implementation cellphones use ...
Read the GPS and WAAS specs, that's all about math and physics in ideal implementations, not a GPS vending corporation's easy-to-digest blurbs.
Edit: I'll leave this alone at this point, it's getting off the topic of the Epics GPS performance (which seems fine for mine when it will get a lock which is the actual issue I'm having ... after some period of time I have to reboot it to get it to lock again ... even though it can see 8-10 birds).
Hey all, new Android user here. I just thought I'd chime in with my experiences.
Using GPS Test (with "Use wireless networks" disabled), I've made the following observations:
In general terms, I'm able to see (and lock onto) multiple (4+) satellites relatively quickly (30 seconds to several minutes). However, like others here, my accuracy is locked at 98.4 feet. SnR for most satellites is in the 20's - 30's. The compass is really wonky, too.
I did a comparison between my Epic and a friends HTC Incredible (again, using GPS Test) at the same location (standing outside in Webster, NY). This is what I found:
So far, I like the Epic but I may end up returning it if Samsung doesn't address this issue. Google Maps is usable (but not perfect) and Google Tracks is problematic. I also know some golfer friends who wouldn't use the Epic since the accuracy is crap. Given that this is considered a "premium" phone, this behavior (poor accuracy and wonky compass) simply isn't acceptable - especially when "lesser" phones do a better job.
Using GPS Test (with "Use wireless networks" disabled), I've made the following observations:
In general terms, I'm able to see (and lock onto) multiple (4+) satellites relatively quickly (30 seconds to several minutes). However, like others here, my accuracy is locked at 98.4 feet. SnR for most satellites is in the 20's - 30's. The compass is really wonky, too.
I did a comparison between my Epic and a friends HTC Incredible (again, using GPS Test) at the same location (standing outside in Webster, NY). This is what I found:
- The Incredible saw more satellites. (8 - 12 vs 5 - 10)
- The Incredible typically had a higher SnR reading per satellite (more consistently in the low to mid 30's, whereas the Epic would average in the mid 20's).
- The Incredible had an accuracy of about 10 feet, but occasionally jumped to 30 feet now and again. The Epic had an accuracy of exactly 98.4 feet and this value never changed.
- The compass on the Incredible is much more consistent.
So far, I like the Epic but I may end up returning it if Samsung doesn't address this issue. Google Maps is usable (but not perfect) and Google Tracks is problematic. I also know some golfer friends who wouldn't use the Epic since the accuracy is crap. Given that this is considered a "premium" phone, this behavior (poor accuracy and wonky compass) simply isn't acceptable - especially when "lesser" phones do a better job.
You're disregarding DGPS. Now, I don't know if cell phones use DGPS or not and if not then the point is moot. But if they do use DGPS, then 3-5 meters is expected.That quote from Garmin says the same thing I am. It's rounded differently, but the numbers are fundmentally the same.
10 feet with WAAS, not without.
Guess which GPS implementation cellphones use ...
And if luck has anything to do with routinely getting accuracy better than 10 meters, other phones seem to get really lucky vs the Galaxy S line.
Yep, my waas statement was pretty far off, shoulda known better since my garmin unit has it lol.
I understand the margin of error, and I would never claim that it is always accurate, under all conditions, at every specific reference point. However under practical use, especially when in motion 2-3 mph I can say with experience that yes it is well more accurate than 33'and is very well within 10'. Thats not to say that it doesnt ocassional jump off target by more than 10'. Reason says in use you can disregard those jumps whereas figures for accuracy/margin of error would be including them.
I appreciate your detailed explanation, it is very informative indeed, but I think on the scale that we use our devices its not as applicable as it would seem. Im confident that you could take any of your handheld gps devices and get, ~10' and better accuracy WHEN IN MOTION on any decent map.
I understand the margin of error, and I would never claim that it is always accurate, under all conditions, at every specific reference point. However under practical use, especially when in motion 2-3 mph I can say with experience that yes it is well more accurate than 33'and is very well within 10'. Thats not to say that it doesnt ocassional jump off target by more than 10'. Reason says in use you can disregard those jumps whereas figures for accuracy/margin of error would be including them.
I appreciate your detailed explanation, it is very informative indeed, but I think on the scale that we use our devices its not as applicable as it would seem. Im confident that you could take any of your handheld gps devices and get, ~10' and better accuracy WHEN IN MOTION on any decent map.
I'm wondering about that 30 meter accuracy myself. If it was off by 98', why I'm I crossing the cross streets at exactly the same time Goggle maps shows me crossing them? Seems dead on to me.
The same thing was reported in sprints community forum Samsung Epic:Anyone else having a problem with GPS...
It seems pretty clear that the reported accuracy is either hard-coded or GPS Test is unable to read it correctly. I'm guessing the former.
Every other GPS-capable device I have reports varying (if not always believable) accuracy.
Much appreciated, thanks!
boomerbubba
Android Enthusiast
I agree with all of that. This 30-meter (98.4 feet) bug is new and unique to the Epic, I believe. None of the other Galaxy S variants have exhibited this strange behavior, although they all have shown different weirdness in their satellite-locking details. (Different symptoms, all bad, on different platforms and firmwares.)
Since the imputed accuracy being reported on the Epic is obviously wrong, I wonder how much we can rely on the other details: number of satellites used, SNR levels, etc.
Meanwhile, it may turn out that those reported internal details really do not affect the actual accuracy of the lat/lon coordinates. This is just my speculation, but it is consistent with the results of test cases such as Adrift's. The coordinates seem pretty accurate, even though the utilities show the imputed accuracy (always 30 meters or 98 feet) is not.
SamsungVibrant
Android Expert
Yep, sadly, you are mistaken.
Some current cell phones might claim to get you to 10ft of accuracy but without WAAS its not routinely happening on any of them as anything more than luck.
A good, dedicated, GPS receiver with a proper antenna, WAAS support and decent conditions will get you to 10 feet. Really good receivers (which is much about clock error as anything) in optimal conditions, manage 3 feet.
PLEASE stop saying smartpones with built in GPS are incapable of getting below 10ft accuracy. Every smartphone I've had, besides the vibrant, GETS BELOW 10 FT ACCURACY. It is not a fake claim or figment of my imagination, it is REALITY.
This was taken, litterly minutes ago, around 10:15pm on a different HTC smartphone I own. IT SHOWS 6.6 FT ACCURACY.
It may be possible that since the Epic came out later than the other 2 carrier's phones, Samsung may have tried something to "correct" the GPS problem, but of which may have left some software bugs in their rush, such as erroneously reporting 30m/98.4' in GPS Test. It may be a lot more accurate, but because of incompatibilities with some 3rd party apps, we just do not know how accurate it is.
I would like to see when the T-Mo version comes out if they will be having a similar problem like the first 2, or a problem like our Epic, or if they will have all the kinks worked out by then. We'll know on Sept 16 when the Fascinate comes up and see if there has been any improvement.
I would like to see when the T-Mo version comes out if they will be having a similar problem like the first 2, or a problem like our Epic, or if they will have all the kinks worked out by then. We'll know on Sept 16 when the Fascinate comes up and see if there has been any improvement.
SamsungVibrant
Android Expert
You can also use the built in native GPS test.
Go to your phones dialer, as if you are making a phone call, and dial in *#*#1472365#*#* Then click on "get position test."
This test will show you your longitude and latitude, you can then enter this into Google maps to see how accurately it located you. It gives you accuracy in meters, convert to feet by multiplying the number given in meters by 3.28, you will now have your accuracy in ft.
The native built in test will show you have many satellites it has in view, the satellites that it actually locks onto will have an asterisk next to it. Please report both these numbers.
Remember you need 4 locked on satellites for true precise 3D positioning.
boomerbubba
Android Enthusiast
Unfortunately, that secret code to open the LbsTest menu has been disabled on the Epic. I think all such codes are disabled, but I have not tried them all.
SamsungVibrant
Android Expert
boomerbubba
Android Enthusiast
Thanks!
That seems to access a different and simpler hidden menu than all the other Galaxy S phones have. My bet is that this is the user interface to a hidden system app called GpsSetup.apk, which is new on the Epic.
When I access that menu on my Epic, and go to Operating Mode, I see that the default setting is not using pure satellite GPS, which is the default in the other Galaxy S phones. (That mode is called Standalone mode. It is available as an option.)
Rather, the default mode on the Epic is called MS Based, which is a reasonable operation setting for the phone. It uses network data to accelerate the initialization of the GPS by downloading almanac and ephemeris data from the network instead of from a satellite as would be the case in Standalone mode. This gets the fix faster, by telling the GPS receiver which satellites to expect where and on what frequency. But under MS Based mode the final fix is still obtained from the satellite.
MS Based should fall back to operate like Standalone when no network data is available. If not, then Standalone mode can be selected manually.
Note that when GPS problems first surfaced on the other Galaxy S models, the first "fixes" circulated on the forums -- just configuration changes, really, they did not fix all the underlying problem with GPS reception -- were to access a similar hidden menu and enable MS Based mode to accerate the fix. So we see that is being set by default on the Epic. I see no problem operating in this mode, since it is a cellphone after all, and the final fix still comes from the birds.
The last option, MS Assisted, is a different matter entirely. It uses the only the network data for triangulation, and is not as accurate. This mode is a last resort, generally for use indoors and on phones without a full GPS receiver.
Google aGPS or A-GPS for a fuller explanation.
this is untrue, most smartphoens do better than dedicated (retail $400, street $200) GPS at this point.
Seriously I don't think you have been around for some of the phones that have had problems and the stringent testing methods we have used on various forums to objectively quantify and to diagnose cause.
I can tell you right off the bat that the SNR on the Epic is the worse of any phone I have tested and I have tested half a dozen comparatively in the same sport at the same time. That points to a real problem.
The first problem is that you are leaving out a key fact, that accuracy is variable depends on which satellites are locked, which ones are used, how many there are, and how often a good lock on three, a marginal lock on two and poor reception of four or five might cause variable inaccuracies as the device moves, encounters foliage, blockage from landscape or buildings, or loss from cloud cover.
You see the problem. I have my Epic sitting next to an evo, a touch pro 2 and a treo pro on my back deck right now, they have been sitting there with GPS enabled for 15 minutes. I am getting a range of 48 to 35 SNR for the top four birds on all devices except the epic which is getting a range of 25 to 12. The problem is that in less than optimal conditions birds will start dropping out of usage.
The epic is also the only device shifting its gps fix in that stable positon
Not exactly. MS based does not necessarily only pull Ephemeris and almanac for network instead of the birds. MS based can also use other information from the towers/network provider. MS based can include other data, such as network time -- and also can include hybrids of say two or three satellites and some positioning information from towers.
The real difference between ms based (ms-b) and Ms-A (MS assisted) is (not the "assisted in "aGPS") it is solely about is where the final calculation is done. Precisely: ms-b = the device calculating its location based on information obtained by the device and by the network (gps or overthrew) or ms-b = the network (sprint) calculating the position of the deivce and then telling the deivce where it is.
Ie MSB vs MSA is a non issue in smartphones as MSA is not really used , it is really for dumbphones that don't have reception equipment and or processing power to calculate their positions or certain emergency situations. MSA data is generally NOT providered to LBAs (programs on yoru phone that use GPS) and often is not updated more than once per hour.
So forget the MS-A v MS-B issue, here is what the Epic is probably using for LBAs (programs that use position):
Standalone GPS
aGPS
Hybrid
Tower info alone (unusually AFLT)
Standalone GPS uses no network resources at all, just you and the one-way downlink from the satellites. Benefits: works away from your carrier. Exampls of benefical sutuations: a) works on the 95% of the earth's surface where there is no sprint signal. b) benefical to people who don't have data incldued and/or pay per kb (ie most of Samsungs market for Galaxy before making US versions)
aGPS: Uses the sats and uses sat data to calculate positon on the deivce, but the network can send the prliminary nformation about where the sats are (alamanc and emphemiers) to the smartphone faster than it can download them from the sats. aGPS can also be simply other network resources like the exact time. often aGPS no longer needs a network connection after getting its first fix of the day. Benefits: faster GPS fixes (TTFF). Drawbacks: requires some network conneciton. Accuracy: Identical to Standalone GPS
Hybrid (GPS plus AFLT): this can use one two or three sats, plus some network resources. This is what you get indoors if you are near window with line of site to sky or top floor light structure roof.. Accurtacy of this kind of fix is usually 30 meters . Benefits. works indoors in some cases and more accurate than (a)FLT
cell Sector/FLT/AFLT: Essentially tower only triangualtion. This can be accurate from as bad as 2000m to 10 meters in perfect certain locations. But it is less accurate and lags severely while moving.
so what do we know the Epic is doing?
1) We can see it doing land based tower triangulation. if you do it indoors totally shielded from sats it does get a fix
2) It is presumably capable of standalone/autonomous GPS, but we won't know for certain until someone tests it after hard rest out of CDMA coverage.
3) aGPS this seems to be working albeit with lower accuracy than devices in its class. We can see EPIC has worse sNR numbers on objective tests than other similar smartphones, meaning its hardware is probably deficient.
4) Hybrid. This is working. I am getting non AFLT fixes indoors with one sats.
I think one strong possibility is that the poor attenuation of the actual satellite GPS reception by the Epic is causing a falback to hybrid which is less accurte, and typcially 100' accuracy, sooner than other devices.
What everyone is reporting is perfectly consistent with poor satellite reception by Galaxies, in the case of Epic compensated for by Samsung with a higher reliance on hybrid.
If you are using hybrid you will see multiple birds in reporting software. The api though maybe going to hybrid at a certain SNR point.
If I put the the Epic on my air conditioner mount in my car with the Evo next to it, the EVO is getting more sats and stronger SNR. The EVO is lagging less. In deep canyons in the cirty the Epic drops out of GPS where the Evo does not.
SamsungVibrant
Android Expert
does that code work similar to the other galaxy s phones? does it give you access to the lbstestmode? It does on my vibrant, just wondering if it does on epic also.
PLEASE stop saying smartpones with built in GPS are incapable of getting below 10ft accuracy. Every smartphone I've had, besides the vibrant, GETS BELOW 10 FT ACCURACY. It is not a fake claim or figment of my imagination, it is REALITY.
This was taken, litterly minutes ago, around 10:15pm on a different HTC smartphone I own. IT SHOWS 6.6 FT ACCURACY.
You realize that's an estimate, yes?
Do you understand how that estimate is arrived at?
Implement a GPS setup from the hardware on up, then come back and talk to me about the realities of the system. I don't care what a $200 phone is telling you, the PHYSICS and limits of CLOCK PRECISION mean the margin for error is ~10 feet with WAAS and higher without.
In motion, you can use various algorithms to average out the error, but those don't work statically.
Hell, aviation grade GPS systems costing the same as a small car, and equipped with motion adjusting algorithms are about the only devices to even claim <10 feet accuracy (and they don't guarantee that in static situations).
Can it get within 10 feet? Sure! Is it because it's routinely that accurate rather than just the various variables happen to be favorable? No.
Even the damn designers of GPS wouldn't tell you otherwise.
Err ... not sure what you're on about here ... my Epic has GPS problems too so I don't need convincing that it's borked.
Once locked mine has acceptable accuracy; my VERY first post on this matter specifically said it wasn't as good as any of my other devices.
A bigger issue for me is the damn things outright failure to acquire a lock most of the time, and it needing to be rebooted even for that!
Yes, the SNR is lousy. It'll see 10 birds and not be able to lock on any of them. Yet an Evo, an iPhone 3GS and iPhone 4, an ancient eTrex GPS, hell an old LowRance 5 channel model, let alone one of my modern receivers are up, locked and solid with easily double the SNR for most birds in seconds.
I built a GPS board using a PCB trace as an antenna recently and IT gets better SNR than the Epic.