On Selected Issues of Behavioural Patterns in Computer Security

This dissertation deals with behavioural patterns and their impact on computer security. The work is divided into two main parts covering different areas: anonymous communication and biometric authentication.
In the part dealing with anonymous communication, we want to extend the knowledge of security limits of anonymity systems with information based on real-world traffic data. We perform a set of analyses targeting the behavioural patterns of users and their impact on such systems. The first part of the analyses are focused on the properties of email communication relevant to the designers of anonymous systems. The three parameters we were interested in were inter-arrival times, the number of messages sent by an individual user and the size of messages. The importance of our analyses' results for the design and implementation of anonymity systems is discussed.
The second part of the analyses are focused on social networks. Information about user profiling, properties of identifiable social networks and their development are presented in the context of the security of anonymous systems.
In the part dealing with biometric authentication, we propose an analysis of challenge-response approach usability together with an appropriate candidate for such an approach - visual evoked potentials (VEP). Their physiological properties and recognition capabilities are presented. We design and implement methods covering all phases necessary for processing VEP as a biometric characteristic and perform them on a test database. The efficiency of our approach is presented in an appropriate way. We confront our results with other biometrics.