Library

When interacting with other people in virtual environments we are usually interacting with an avatar that we take on trust to represent a particular person. The fact that the avatar may have virtually any appearance possible and that the same avatar may be controlled by different people at different times raises some interesting questions on why we should trust what we see.

With this in mind we have developed a biometric system that authenticates a user using its own heart beat and brain waves. These signals from a living human may provide a route towards improving the sense of interacting with another person.

In parallel we have developed a system that makes use of specific BCI tasks as a means of authentication. Now we are dealing with an active system: the user needs to perform a specific task (input a password in the computer) in order to be authenticated. There are 2 levels of security: the password and the control of the BCI. The biometry involves two aspects: on the one hand, the features, classifiers and electrodes locations are personalized, on the other hand, only the user knows what imaginary movements he/she needs to perform.

In both cases our work is built on our own ENOBIO wearable electrophysiological recording device and both can be implemented in parallel and continually.

Two major problems are related with electrophysiological biometrics: user acceptance and movement artifacts, including EOG artifacts. By using the ENOBIO amplifier, which is wearable, wireless, can work in dry mode and it is automatically calibrated, the former problem is solved. Regarding the artifacts, two solutions have been explored: the use of an authentication protocol, where the subject is asked to not move during the recording time (EEG Equal Error Rate (EER) = 20.8% and ECG EER = 2.1%) and the use of an artifact correction algorithm (EEG EER = 25.6% and ECG EER = 12.3%).

We have demonstrated proof of concept for electrophysiological biometrics, including the use of a BCI. We believe that apart from authentication this technology will become more important over time for two reasons: it opens the way to providing real time information on the affective state of users and as BCIs become more ubiquitous the system needed will already be in place and will be a natural part of virtual interaction.