(sp)iPhone: Decoding Vibrations From Nearby Keyboards Using Mobile Phone Accelerometers

This whitepaper demonstrates that an application with access to accelerometer readings on a modern mobile phone can use such information to recover text entered on a nearby keyboard. Note that unlike previous emanation recovery papers, the accelerometers on such devices sample at near the Nyquist rate, making previous techniques unworkable. Their application instead detects and decodes keystrokes by measuring the relative physical position and distance between each vibration. The authors then match abstracted words against candidate dictionaries and record word recovery rates as high as 80%. In so doing, they demonstrate the potential to recover significant information from the vicinity of a mobile device without gaining access to resources generally considered to be the most likely sources of leakage (e.g., microphone, camera).