Realistic Driving Trips For Location Privacy

Simulated, false location reports can be an effective way to confuse a privacy attacker. When a mobile user must transmit his or her location to a central server, these location reports can be accompanied by false reports that, ideally, cannot be distinguished from the true one. The realism of the false reports is important, because otherwise an attacker could filter out all but the real data. Using our database of GPS tracks from over 250 volunteer drivers, we developed probabilistic models of driving behavior and applied the models to create realistic driving trips. The simulations model realistic start and end points, slightly non-optimal routes, realistic driving speeds, and spatially varying GPS noise.