Search

The Internet is transforming itself and the daily lives of people at a fascinating pace. It revolutionizes the exchange, retrieval, and publishing of information for companies just as it does for individuals. This transformation manifests itself, for example, in the myriad of services that are being offered over the Internet today. Authentication mechanisms are one aspect that has not kept pace with this development. While being ubiquitous in today's digital society, most of the current service providers still rely on authentication based on a username and password combination.This popular approach to authentication and its realization in practice have several shortcomings. First, users tend to choose weak passwords and to re-use the same username and password for several service providers. This user behaviour gives rise to a multitude of attack vectors threatening the effectiveness of the authentication approach. Second, for accountability reasons, service providers request the release of extensive amounts of personal information at registration time. As only a few attributes comprise authentication capabilities, a user may supply arbitrary attribute values, which results in bad data quality for service providers. Consequently, users and service providers have a mutual interest in looking for innovative approaches to authentication.Techniques to increase the data quality of service providers by exchanging certified attributes, using technologies such as OpenID or SAML, steadily gain interest. However, such methods come at the price of making the personal information hosted at service providers a more attractive target for attackers, thereby raising the cost for protective measures. In addition, they lead to an immense dispersion of personal information of users, effectively leading to all transactions of a person becoming linkable. Data-minimizing authentication does not suffer from such deficiencies. Leveraging cryptographic techniques, this authentication paradigm realizes the seemingly conflicting goals of service providers, who desire good data quality or strong authentication guarantees, and users, who want to only reveal as little (personal) information as possible.The main goal of this work lies in enhancing the practicability of data-minimizing authentication techniques. We structure our contributions into two parts. In the first part, we introduce mechanisms that improve the efficiency of group signature schemes and anonymous credentials, which are systems implementing the concepts of data minimization. Namely, we propose a group signature scheme that provides the shortest signature size and comes with an efficient signature-generation algorithm. In addition, we present an efficient implementation of anonymous credentials on a smart card, an extremely resource-constrained device. Thereby we demonstrate the feasibility of implementing computationally-intensive authentication technology on currently available hardware. Further, we present a modular architecture for the Identity Mixer anonymous credential system and show how it can be integrated into a standards-compliant authentication environment.In the second part, we analyze how people and organizations present themselves in the digital domain and what mechanisms exist to achieve recognition processes as in the offline world. Such an analysis proves useful for anticipating problems that may arise when deploying data-minimizing authentication. At the same time we use the insights gained in the recognition and authentication process to illustrate the main ideas of data minimization, thus contributing to a better understanding of this authentication concept. Further, using our conceptual approach to digital identity, we propose an intuitive mechanism for managing trust in attributes of people. Our approach aims at replacing today's cumbersome bootstrapping of personal trust relations. nrpages: 234 status: published