Privacy Enhancing Technologies for Privacy Enhancing Technologies

This work contributes to the field of unlinkability. Unlinkability describes a situation where attackers are unable to correctly cluster items of interest. The focus of this work is on modelling and measuring unlinkability. Additionally a protocol for provision of unlinkable certificates, tailored to the requirements of vehicular communication is introduced. The vehicular scenario is used throughout this work as a motivating example. In this work, a linkability graph is developed and used to illustrate the tight relation between unlinkability and anonymity in certification schemes. The main distinction between anonymity and unlinkability problems is captured by the notion of inner structure, which is developed in this work. Inner structure describes the inherent difference between hypotheses of unlinkability problems. This difference is expressed by metrics that reflect the attackers’ objectives and world-view. Inner structure motivates consistency of an attacker’s world-view as criterion for assessment of attacker quality. Previous entropy-based global unlinkability measures do not reflect the consistency of attacker probability mass assignments. In this work, these shortcomings are analysed, and criteria for unlinkability measures are developed. These criteria are then discussed for a new unlinkability measure introduced herein. This expected distance unlinkability measure is defined using the inner structure of a space of hypotheses. It thus reflects the consistency of an attacker’s assignment and not only the attacker’s certainty as done by entropy-based measures. This work also contributes a certification protocol that provides revocableanonymous protocols that are not linkable to each other or the authenticated requester by the issuing certification authority or others. Revocation of anonymity is possible only by a quorum of independent revocation authorities. The objective is to reduce the trust that has to be put into single authorities to reduce the risk of misuse. The protocol is based on a previously known cut-and-choose blind signature protocol, but made more flexible to suit vehicular communication scenarios. An extensive security analysis of newly introduced protocol-parameters is provided. Concluding, this work provides a contribution to the ongoing discussion on unlinkability, unlinkability measures and analysis. It draws motivation from a communication scenario that is intensively researched at the moment and contributes to the protocol development there.