On the obfuscation of image sensor fingerprints

Marcus Schwarting, University of Louisville
Timothy Burton, University of Louisville
Roman Yampolskiy, University of Louisville

Abstract

To date, there have been several attempts to utilize sensor noise for device identification and authentication. Most of these developments have taken place in mobile image sensors, accelerometers, and microphones. The ramifications of isolating sensor fingerprints is that any third party, utilizing multiple sources, may effectively track the mobile movements of a user without the use of cookies. Sensor fingerprints, especially that of images, could pose quite a challenge to applications and websites that promise online anonymity in picture sharing. This paper describes the threat potential as well as the required steps to remedy said vulnerability. It then suggests the potential for widespread implementation in a more diverse span of fields.