Friday, April 23, 2010
[Editor's note, 5/14/10 -- This post contains incorrect information about our Wi-Fi data collection. We have posted a clarification and update about our process on the Official Google Blog.]
Where? It might be one small word but the answer to that question can really prove decisive. ‘Quick! The volcano has erupted again and my flight’s cancelled, where is the nearest train station?’ or ‘Where’s the best place to get a decent coffee round here?’ I’m always happy to help, but nowadays these questions can often be answered by firing up your smartphone. The key to knowing where you should go is knowing where you are starting -- my recommendation for the best local cafe depends on whether you ask me when I’m at my office or across town near my house.
So it’s no surprise that geolocation services are becoming more popular with people than ever before, and that many companies are adding locations functions to respond to this demand and improve the services they offer their users. For example, Twitter allows users to geotag their tweets. Everyday, millions of twitter.com users on a desktop computer or mobile phone choose to do this. Knowing that someone is posting from Bangkok as opposed to Manchester gives followers an important perspective when they read a tweet about ‘red shirts all over the city.’ On smartphones like iPhone, Palm, and Android devices, services such as Google Maps for mobile, Yelp, and Urbanspoon use location to provide relevant local search results, while apps like Foursquare and Google Latitude let you find your friends on a Friday night.
But how does this work? How do smartphones actually know where you are? Lots of people talk about GPS (the satellite-based global positioning system) but a lesser known fact is that GPS can be slow: it can take tens of seconds and sometimes much longer to “fix” your location depending on the specific hardware and your physical location. Consequently, companies have worked out other solutions. These are generally based around the idea of “listening” for nearby cell towers or Wi-Fi hotspots and using this data to quickly approximate a rough position. In fact the very first iPhone didn’t even have a GPS - it relied solely on this kind of triangulation for location information. By treating Wi-Fi access points or cell towers as “beacons,” smartphones are able to fix their general location quickly in a power-efficient way, even while they may be working on a more precise GPS-based location. This can be done without any intrusion into the privacy of a Wi-Fi network. We only use information that is publicly broadcast (for example, that list of Wi-Fi hotspots you see in the anytime you open the “join network” option on your PC). It doesn’t involve accessing the network to send or receive data. [**See 5/14/10 note above]
However, a database of known network locations is required for both cellular and Wi-Fi triangulation. Companies like Skyhook Wireless compile such databases which are used by many in the industry. Intel has promoted the use of Wi-Fi for “location aware computing” through projects such as its Location Technologies SDK. Google has contributed by creating its “My Location” API, which relies on a database of locations collected by Google Maps for mobile as well as by our Street View cars, which catalog the location of Wi-Fi access points (the public “beacon” part described above) at the same time as they are collecting Street View imagery and 3D laser scans of buildings. Google has also worked to help even notebooks and tablets find out their location in Chrome and Firefox through the W3C Geolocation API.
Location is clearly going to be a big part of the web going forward. It helps bring the power of the Internet to improve the real, everyday things we need to do. With location-enhanced searches on smart phones, mobile PCs and tablets, we can just do them faster and smarter.