Scientists at Rutgers University have developed a method to determine which route someone has taken just by measuring the speed of a car. With this they want to show that even with only this data privacy can be at stake.
The method, which the researchers call elastic pathing, can reproduce a motorist’s exact route without using GPS or similar techniques. This only requires a starting position and an OBD connector for the car, with which the speed can be read via the vehicle management system and forwarded via Bluetooth to, for example, a smartphone.
The software can be used to determine where a motorist is driving based on the speed. The algorithm looks at how fast someone is driving and compares that with the roads on a map from the starting position. As soon as the speed changes and the image no longer matches the assumed road, the software searches for nearby roads where it does.
The scientists conducted 240 trips with six drivers in New Jersey with 46 final destinations. In addition, they conducted a similar test with 21 drivers, who drove 691 trips with 240 final destinations in Seattle. These experiments showed that the method correctly identified the destination for 26 percent, with a margin of error of almost half a kilometer.
The method is not completely perfect, the scientists emphasize. Drivers differ from each other in terms of driving styles and the situation on the road at that time must also be taken into account. Despite this, they believe they can demonstrate that the location of a driver can be traced, while no GPS is used, for example.
With this observation, the computer scientists point to American insurance companies, which more often request the speed of cars in order to map the driving behavior of their customers. “The data, once collected, doesn’t go away. Improvements also make it possible to retrieve more personal information,” they say. They therefore believe that alternatives should be considered to better protect privacy. The scientists will present their final findings at the International Joint Conference on Pervasive and Ubiquitous Computing next month.