Observation-Resistant Multifactor Multimodal Authentication
Aleksander Mallasvik · NORA - Norwegian Open Research Archives · 2010
ENGELSK: This thesis investigates the use of hand gestures as an additional modality in authentication schemes, to thwart the risk of observation (shoulder surfing) attacks. We used the accelerometer already embedded in the iPod Touch to gather accelerometer signals, which were used to conduct experiments on how accurately we could recognize and differentiate different gestures. We restricted ourselves to a pre-defined set of gestures, and achieved an EER of 5% on the controlled wrist movements, and 8% after including two circular motions. The algorithms we used were tailored to fit the limited computational power of the iPod Touch, as we needed a recognition module that could be used in real time for our authentication schemes. After assessing the characteristics of the different hand gestures, we developed two unique authentication schemes that incorporate hand gestures as an additional modality for authentication. We developed suitable attack scenarios, and found that both schemes adds additional entropy to the scheme, as well as a significant amount of shoulder surfing resistance.