Scientists from University of Michigan and the University of South Carolina have found a way to attack accelerometers by using sound waves. This technique has allowed them to spoof accelerometer signals and feed devices and applications with fake data.
This new attack poses a serious security threat because most devices inherently trust all the data coming from accelerometers. And most devices don’t feature protection systems to detect spoofed signals coming from local sensors.
Sound waves induce fake motion into accelerometers
The principle behind this hack is simple. Accelerometers have a sensing mass that sits on a series of springs. By sending finely-tuned sound waves to the accelerometer, the sound waves induce controlled movement into sensing mass.
Furthermore, this action is recorded by the springs as a motion, even if the accelerometer and the rest of the device (where the accelerometer has been embedded in) stands still.
By using previously prepared sound waves, researchers made the car move without interacting with the smartphone.
Similarly, they also used sound waves aimed at a Fitbit device to record 3,000 steps on the device, generating fake data and earning financial rewards.
A speaker can make tones that fool a sensor and cause a microprocessor to accept the sensor readings. Credit Joseph Xu/University of Michigan
Attack has widespread repercussions
The attacks were trivial and had no immediate effect on the security of the device’s human operator.
Nevertheless, their research shows a new attack vector that can be truly devastating if used for nefarious purposes. And also against devices in industrial systems, where accelerometers deploy oftenly.
Scientists say the sound waves used in these attacks could be embedded in songs played over a user’s speakers. However, no need of high-end speakers, and regular $5 speakers are enough to replay the malicious sound waves.
For further technical details of attack, refer this research paper.