Location and the iPad

Location is not Place, but the two concepts do intervolve. Or perhaps (when I’m feeling well-disposed to the world) they intertwingle.

So I’m interested in location, and for this reason will be buying the wifi+3G iPad, which has a comprehensive suite of location-awareness technologies, rather than the wifi iPad, which is also location-aware but less comprehensively so.

I keep seeing absurd fallacies being promulgated about the iPad and Assisted GPS. I think “promulgated” is a word that is now entirely reserved for absurd fallacies. Do you think anyone is out there promulgating enlightenment? If they are, they’re not posting to the Wired Gadget Lab weblog comment threads, anyway.

So here for your edification is the truth about A-GPS vs. GPS vs. wi-fi triangulation.

Note: this is dull, don’t bother reading it. I just had to get this rant down to stop me boring people with it in person.

GPS uses satellites to find your location. It’s quite accurate but it takes a long time to get a satellite fix, especially if you are in a new location. So A-GPS was invented – “a system which can improve the startup performance of a GPS satellite-based positioning system“. A-GPS is GPS, plus extra functions to make it faster. It uses a database of cell-tower locations to get a very rough fix, which makes finding the real satellite fix much faster. It saves battery life, speeds up your GPS fix, and does not damage accuracy at all. It’s not a special iPhone thing, just about any cellphone with GPS now uses A-GPS, because it’s better.

A-GPS does not require data service, just cell tower locations and GPS satellites. Google Maps requires data to download its maps, but if you only have wifi you can cache them and they’ll still be available; or you can use another app that keeps its maps on the device.

[UPDATE: as Aram points out below, it does require a data connection every few days to update its database of satellite locations. However just as with caching the Google Maps tiles, connecting to a wifi network every so often takes care of that so you can, as I say, get away without buying data service.]

The original iPhone did not have GPS or A-GPS. It uses skyhook, which triangulates from a database of known wifi base station locations. This is very accurate if you’re in a densely populated city with lots of wifi base stations around. It’s so accurate because when it can, it will send pings out to three base stations, time how long it takes to get a response from each of them, and triangulate your position. It’s how GPS works, but instead of satellites it uses wifi base stations. It’s pure genius. In the city it’s actually better than GPS because it can be hard to get line-of-sight to three satellites when you’re surrounded by skyscrapers or underground. However in less densely populated areas where wifi is sparse it becomes inaccurate. This skyhook service is what the wifi-only iPad is using. Fantastic in the city, useless in the country.

The iPhone 3G, 3GS and wifi+3G iPad all use A-GPS. As I mentioned earlier, this is GPS. It uses GPS satellites. But because it’s Assisted, it’s faster than unassisted GPS. This means that they can get an accurate position fix anywhere in the world, except a few locations where the US government intentionally blurs GPS (mostly war zones). I expect Apple also has skyhook running on these, so even underground in cities where GPS can’t reach they can still get a location fix through wifi triangulation.

Now, as a bonus, a look into the future: location services will get even better. The Smart Internet Technology CRC Australia developed a system that will triangulate location from known fixed bluetooth signals. This means in super-dense environments like shopping malls and convention centres, you could have centimetre-accurate location. Also, if you happen to be in Japan rejoice: a small constellation of Japan-specific satellites is going up, to bring a super-accurate GPS-augmenting local system to the home islands.

So, please stop writing that “A-GPS” is “fake”. It’s not; it’s the real deal. A-GPS plus skyhook is absolutely the best (fastest, most accurate, most comprehensive) location service available at the moment. The wifi+3G iPad will be, like the iPhone 3G and 3GS, an unparalleled device for location-based services – for the brief time before the rest of the industry catches up.

6 Responses to “Location and the iPad”

  1. Aram says:

    You almost got it right. A-GPS does require a data service. Read the Wikipedia article you linked to more thoroughly.

    For me, the more bothersome absurd fallacy that’s being promulgated is the notion that GPS-enabled mobile phones are an acceptable substitute for a dedicated Navman/Garmin GPS device for serious use. Case in point: a $99 TomTom app in the App Store that may lead you to believe the iPhone’s GPS accuracy/reliability can compare to a dedicated device.

    • viveka says:

      Hi Aram, nice to have you here :)

      And dear readers, some context: this is the next phase of a good old-fashioned Mac-vs-PC flame war that Aram and I are having on facebook. Takes me right back, this does!

      So – actually I’m utterly, completely, 100% right. Probably. As the Wikipedia article says A-GPS uses “data available from a network”, but as I wrote it does not require data *service*. It pings towers, but does not transmit through them. It piggybacks on data infrastructure but doesn’t require a subscription to it.

      At least that’s what I’ve read. Not in Wikipedia; in the technical documentation that I had to plough through when I was running a geospatial technology company full-time for two and a half years. However arguments from authority don’t have real merit; any claim should be able to be tested.

      So to primary data: as evidence I will upload screenshots from my wife’s iPhone 3G, which has no data plan, and with wi-fi turned off, showing that it can nonetheless obtain an accurate GPS position. It’s possible that it’s falling back to unassisted GPS to do this, but I timed the acquisition at under 10 seconds, which suggests to me that it’s A-GPS.

      I will admit that it’s possible that the iPhone falls back to unassisted GPS when it’s not on a data plan; can anyone confirm? If it is just GPS then it’s uncommonly fast. And if it’s just GPS then the worst-case scenario with an iPhone is the same as the best-case on a pure-GPS device.

      Now to your second concern: “the notion that GPS-enabled mobile phones are an acceptable substitute for a dedicated Navman/Garmin GPS device for serious use”. If your idea of serious use is driving directions (as you imply) then there is no cause for concern. I have owned three Garmins, and the iPhone is just as accurate and much faster. GPS is rocket science, but it’s not magic. Garmin was a pioneer, but they own no eldritch formula for making better GPS devices.

      I used a Garmin iQue for driving directions for a couple of years, and it was far, far worse than the iPhone running TomTom that I use now. It had a touch-screen with a row of 16×16 pixel icons, so I had to get out a stylus and squint to interact with it. Using the built-in flip-out antenna it would take minutes to get a GPS fix and then lose signal constantly. I immediately bought a third-party external antenna, which I stuck to the outside of my car. This helped a lot, but I would still have to sit around for a minute or two waiting for a GPS fix, then drive a block to give it a heading, and then wait another minute for the route calculation before setting off. It made the whole process sufficiently off-putting that I would often choose not to subject my passengers to it.

      My Garmin had turn-by-turn navigation, but if you went off-route it would take about a minute to recalculate, so I avoided going off-route whenever possible. Updating the maps for it cost a couple of hundred dollars, and they had to be sideloaded through an excruciating sync-and-authenticate process. I once spent a very long time on hold on an international phone call to Garmin’s Asia-Pacific offices in Singapore because the authentication key they sent me stopped working.

      I’ve also used built-in car GPS devices. Invariably the interfaces are laughably bad.

      Oh, and in the CBD my Garmin was completely useless; skyscraper canyons block out line-of-sight to the satellites. This is where Skyhook on the iPhone comes into its own. It seamlessly takes over, and the position fix never wavers.

      Now my iPhone running TomTom is mounted in the same place on my dash that the Garmin was. I have not bought the $100 TomTom external antenna (includes mount and dock, pretty good deal all up). I’m considering it, but it’s nothing like as urgent as it was for the Garmin. The position fix is fast, solid and reliable. When I go off-route it recalculates in seconds. It’s integrated with Google Maps search. I can choose from something like a dozen voices; it reads street names out to me. It’s up-to-date, it knows about speed cameras and tollways, and unlike the Garmin it doesn’t keep trying to send me down the traffic nightmare of Parramatta Road. When it’s time to pick a freeway exit it shows me a clear lane diagram. It blows the Garmin out of the water.

      Speaking of water, Garmin probably still has the edge in maritime navigation. iPhones are definitely not waterproof. And more recent Garmin devices are surely better than my old iQue. However I have to tell you, I have direct, real, in-the-field experience of serious dedicated devices that sold for more than my iPhone, and the iPhone with TomTom is superior in every way.

      So there 😉


      • Aram says:

        Here’s a very comprehensive article on the subject of A-GPS from Ars Technica for your perusal:


        I’ll sum it up for you: it says you’re 100% wrong.

        The old “MS-assisted” location services that used the control plane are no longer used in modern devices (due to network/operator incompatibilities and general suckage), all of current location assistance is done using the user plane, i.e. data services (via IP). It most certainly does not “piggyback” on existing data infrastructure without subscribing to it, which is a pretty absurd notion.

        The data service is primarily used for fetching an almanac of ephemerides (fancy talk for a database of satellite orbits). With Skyhook, it is obviously also used for location queries (any caching is limited and subject to expiration), as well as data uploads to Skyhook (a form of distributed wardriving, if you will).

        Now this almanac is what brings me to your anecdotal results with your wife’s iPhone. A cached almanac takes a couple of days to expire, and until it does, can be used to assist the GPS to get a TTFF of under 6 seconds. For a more accurate experiment, turn off the data and wi-fi on the iPhone, then turn the phone itself off for a week or so, perhaps even move it to a new location for good measure, then try to see how long it takes to get a fix. If it’s under 10 seconds, then you can print out a copy of this comment and I will eat it.

        Having laid the issue of “does A-GPS use data services” to rest, let’s move on to the accuracy of the GPS fix.

        To better understand my point, consider the analogy with phone cameras and dedicated cameras. Phone users are fully aware that the cameras on their phone are vastly inferior to even the cheapest digital cameras, and you won’t find anyone trying to convince them otherwise. The users know that their phone camera is good enough for a snapshot for Twitpic, or for a reasonable record of a happenstance event, but if they’re going on a sightseeing trip or such, you can be sure that they’ll be bringing their digital camera along.

        In contrast, with GPS on phones, users are not aware of how inferior the functionality on their phones is in comparison with a dedicated device. They might even think that it’s good enough to take on a bushwalk through the Snowy Mountains (the lack of a replaceable battery on the iPhone makes it doubly unsuitable for this purpose). Do you see what I’m getting at?

        Now, you tell of anecdotal data about how you’ve found the iPhone to be acceptable for in-car navigation. My own anecdotal data sharply contradicts yours. I constantly find the iPhone telling me I’m on a parallel street, that I’ve inexplicably made a turn onto a back alley from a highway, or that I’m inside a building that looks like a well guarded military warehouse from where I’m actually standing. Furthermore, the compass is subject to constant interference from sources unknown, and the altimeter is inaccurate to the point of being useless.

        But as you well know, the plural of anecdote is not data, so I present you with a scientific journal article about how much the iPhone GPS sucks:


        Furthermore, in your anecdotal comparisons with dedicated devices, you’re comparing the iPhone with relics from a time when dinosaurs walked the Earth. None of their shortcomings you listed hold true for modern devices.

        I also find it funny that you exalt the features of the TomTom app, when all of those features are also available on dedicated devices manufactured by TomTom that cost only $70 more than the app.

        And don’t kid yourself, this is not a PC vs. Mac debate, all this applies to phones that used A-GPS long before the iPhone zygote implanted itself in the dark womb of Jobs. Though I am always happy to mention that the iPhone is undeniably the worst abomination since the N-Gage that dared call itself a mobile phone.

        • viveka says:

          Good lord man, you’ve made a couple of nice points but don’t you think you’re rather overstating your case?

          You point out that A-GPS needs a data connection to download an ephemeris every so often. Indeed it does. That hardly makes me “100% wrong”. iPhone mapping uses data, which I’d already pointed out; but you can get by with occasional wifi. It does not *require* a *data service* from your telco. And if you read back, you’ll see that’s all I claimed.

          Your suggested test of disconnecting the iPhone from all data networks for a week would indeed prove your point about ephemeris download, which I happily concede. However it is not “a more accurate experiment”, because it’s not how iPhones are actually used. This is a socio-technical system, so case studies (with all their limitations) are in fact our best method of gathering meaningful data. In practice, my wife’s iPhone which has no data service is still getting getting the benefit of A-GPS.

          Early A-GPS systems *did* use the control plane, as you admit. And current systems do something very similar over the user plane; they’re still pinging towers. They use the data network, but as the very nice Ars Technica article you cite points out network operators support it even for devices that do not have *data service*.

          This is piggybacking on existing data infrastructure, which is not absurd at all; it’s also what Skyhook does. The Ars article is roundly in praise of A-GPS, and contains some harsh criticisms of the cold start problems that plague unassisted GPS. I really don’t see how it helps your case here, except that it points out that control-plane systems are no longer in use. But once again, I had already said that I didn’t know what kind of A-GPS the iPhone was using, and I assert that it’s not the point. And again you misquote me: I did not say that A-GPS doesn’t “use data services”, I said that the iPhone’s A-GPS doesn’t “require a data service”.

          So dammit, read back, I was careful in my assertions. You can demolish straw men all you like, but you’re disproving something that I did not claim.

          In any case this has no bearing on my robust, reasonable and wise point that the iPhone and iPad’s A-GPS is not fake and will be excellent for location-based services, as opposed to your unsupported assertion that it “cannot compare” to a dedicated device like a TomTom.

          You then complain that I’m talking about old devices. I specifically acknowledged that newer Garmin devices are undoubtedly better than my old iQue. Nonetheless that was a “serious” device made by a company you cited. I know internet time is fast, but the iQue 3600 launched in 2003, was still being sold and actively promoted by Garmin in 2007 and was discontinued only in June 2008. For you that counts as “relics from a time when dinosaurs walked the Earth”?

          Now, I emphatically agree that the iPhone is the wrong GPS to take on a serious bushwalk. So is a TomTom. It’s not the point you expressed earlier but it’s a good one and I agree with it. I think that iPhone users are only too aware of its battery limitations, but if someone wasn’t and thought they could take it instead of a map and compass or a proper hiking GPS on a three-day hike to the Bluegum Forest, I’d strongly advise them not to be daft. A lack of replaceable internal battery isn’t the problem, as there are multiple external battery solutions available (including solar rechargers), but there’s no way I would consider the device robust enough to stake my life on. I would require something shockproof and waterproof instead, and I’d still take a map and compass.

          I agree the iPhone compass is too often subject to interference, but it hasn’t been a problem for me while driving, and I’m prepared to accept that the altimeter is useless; I’ve never been curious about the altitude of my car. I believe you when you say you’ve had a different experience to me; are you using the TomTom app? If you are, then it might just be something about the configuration of your car. An external antenna is the usual solution, just as it is with any GPS device. Every review I’ve read of the TomTom for iPhone Car Kit indicates that it is every bit as good as a regular TomTom. Of course it is pricier to buy an iPhone plus the TomTom app and the Car Kit, unless of course you *already have the iPhone*. If you really want you can jailbreak and use a cheap bluetooth eternal GPS receiver, but that seems like a lot of bother to save fifty bucks.

          Finally. What you call the “scientific journal article about how much the iPhone GPS sucks” is in fact a balanced review, and you are heinously mis-stating its position. The abstract does say that in their tests the iPhone’s A-GPS was less accurate than a dedicated GPS. That sounds like a big deal until you read the actual article and take a look at their methodology. “A-GPS locations were collected at outdoor sites under ideal conditions, i.e. excellent satellite visibility” (p. 14). What about urban canyons, the common scenario that I mentioned? Well, they confess that they didn’t test those in the section on limitations of their methodology. They did try indoors, where the iPhone switched to other positioning methods, which are less accurate than GPS, but they *didn’t even bother testing the dedicated GPS* because it would have given them nothing. By the way, their “low-cost GPS” is a Garmin GPSMAP 60Cx, $412 from my favourite GPS store. http://www.ja-gps.com.au/Garmin/handheld-gpsmap60cx/ – and a very nice piece of kit. All measurements were taken in Albuquerque, 59th in the US for metro population and around 1/10th the population of Sydney, so hardly the best-case for wifi density; nonetheless “WiFi and cellular positioning were able to obtain a position fix at most of the indoor sites where A-GPS failed”.

          They go on to attribute the worst position errors to the iPhone snapping reported positions to the road network. Rather than attempting to get a position fix along a road where most driving and walking happens, the researchers had chosen random points on a map, in the middle of various fields (and excluding any points without perfect satellite visibility). They cite other studies of positions taken while driving along roads that give much better accuracy figures, and conclude that positional accuracy away from roads is likely to be less accurate.

          So in the end this study weakens your claim that the iPhone is bad for driving directions, but it strengthens your new point – the iPhone isn’t the right GPS for bushwalking.

          Personally, I’m interested in location-based services, and the standard canonical data point for a premises is its point of entry from the street. So for that purpose, locations that are accurate along roads will do nicely.

          The iPhone is the worst smartphone except for all other smartphones so far. It’s made locative media an actual thing, instead of a dream. We can do much, much, better; but no-one has yet.

  2. Jess says:

    Eventhough a lot of manufacturers have already said that cellphones nowadays are GPS ready, still there I don’t buy their commercial. Nothing can still beat the efficiency of having a real GPS device in hand.

    • viveka says:

      I think it depends what you’re using it for. Specialised GPS devices for applications like marine navigation or hiking still make sense. I think it’s sad that the incumbent GPS industry missed the boat on the rising wave of urban location-based services, and the coming tide of the read-write geospatial web. Location-enabled smartphones are taking up the slack, and the location-aware iPad will spawn a whole new category of apps.

      It reminds me of how the established GIS players failed to see the real opportunity of web mapping and let Google Maps create the category instead. They could leapfrog – there’s still so much more that could be done using structured data that Google doesn’t seem interested in – but they’re sticking with the old enterprise model instead; hostage to their existing supply chains and customer base. Maybe it’s time for some spinoff companies to emerge.

Leave a Reply