Mitigation of Odometer Fraud for In-Vehicle Security Using the Discrete Hartley Transform

Abstract

Odometer fraud is a serious offense in the automotive sector and indicates the disconnection, resetting, or alteration of a vehicle’s odometer and the related sensor with the intent to change the number of miles/Kms indicated or recorded to report false information. This paper focuses specifically on the threat scenario where the odometer sensor (i.e., Hall Sensor) is manipulated or replaced to implement an odometer fraud. This paper proposes a technique to mitigate odometer fraud by performing a physical layer authentication of the Hall Sensor, which takes in consideration the limitation of the in-vehicle networks and microprocessors. In particular, the Discrete Hartley Transform (DHT) in combination with machine learning algorithms is used to perform the authentication on an experimental data set of 12 Hall Sensors, which has been collected by the authors. The results shows that features extracted with DHT have more discriminating power than the original time domain and the frequency domain representations based on the Fast Fourier Transform (FFT) especially in presence of noise.