Abstract
Distributed and open environments require flexible, scalable and extendible trust verification mechanisms to access resources. To address this, the use of digital credentials as a means for making access decisions has been promoted. The resource owner needs to verify if the requester's credentials satisfy the security policy of the owner. However, such verification becomes a challenging problem when either the requester does not wish to disclose her credentials before the verification is complete, or the owner wishes to keep its security policy confidential from the requester, or both. In addition, the requester may associate a score to each of her credentials based on her perceived level of privacy. Earlier proposals to address this problem limit the owners policy to be a set of credentials. However, real world policies are more complex than a simple set. In this paper, we present three alternative privacy preserving trust verification solutions that protect both the owner's policy and requester's credentials, while at the same time allowing more expressive owner's policies that can be specified as a tree structure. We analyze their computational complexity, communication cost and the amount of disclosure.