Android (specifically a Nexus 4) was used to test this location tracking scheme, but it would work equally well on any mobile device that offered network access and battery stats. Tracking location based on battery activity is predicated on the assumption that the farther a device is from the a cell tower, the more power it uses to maintain a connection. The same is true when it’s inside a building or otherwise obscured by structures.The phone you carry around all day has a myriad of different sensors that measure everything from location to barometric pressure. Apps usually have to adhere to the permission control system built into platforms like iOS and Android to get that information, but a team of researchers at Stanford University has devised a way to collect location information without talking to the GPS hardware. All they need to figure out where you’ve been is access to the battery levels.
The researchers call their proof-of-concept application PowerSpy. Before this app can be useful, it must first know what the battery power map of an a route is. In the same way GPS systems use satellites as a point of reference, PowerSpy needs to know what battery performance to expect at different points in a journey and pre-associate that with a GPS location. So as you take the bugged phone around town, the PowerSpy app could watch the amount of juice being drawn and fit that to the map.
So what about all the other things that cause a phone to draw power? Using apps, placing calls, and simply having the screen on will drain a lot of battery. Indeed, this does increase noise in the data. According to the researchers, the algorithm used to track location is not attuned to these short-term fluctuations, but rather to a measurement period of several minutes. This allows the system to filter most of the battery usage that isn’t related to location and the cellular radio.
To test PowerSpy, the team mapped out a number of different routes between two points. The goal was to determine which route the phone (and the human carrying it) took based entirely on the battery drain. With few apps running, PowerSpy could identify the exact route two-thirds of the time and its overall distance error was 150m on average. That’s not much worse than the course location permission on Android. With a larger suite of apps running in the background including Facebook, Twitter, and Waze, the effectiveness dropped to 20% for exact route fits, but the average distance was still only 400m.
It’s far from perfect, but rather impressive for using only battery drain. The study saysPowerSpy could be improved further if it accounted for the battery usage of individual apps and services, as well as wakelock state. This data is all exposed by Android to some degree without root access. An app like PowerSpy could be used with malicious purposes in mind, but the requirement that you have a map of battery usage for various routes currently limits its real world utility. If that data is acquired and associated with GPScoordinates, this technology could be more useful, and also potentially worrisome.
No comments:
Post a Comment