On Generic Groups and Related Bilinear Problems

Groups with pairing are now considered as standard building blocks for cryptographic primitives. The security of schemes based on such groups relies on hypotheses related to the discrete logarithm problem. As these hypotheses are not proved, one would like to have some positive security argument for them. It is usual to assess their security in the so called generic group model introduced by Nechaev and Shoup. Over the time, this model has been extended in different directions to cover new features. The relevance of this model is nevertheless subject to criticisms: in particular, the fact that the answer to any fresh query is a random bit string is not what one expects from a usual group law. In this paper, we develop a generic group model with pairing which generalizes all the models seen so far in the literature. We provide a general framework in order to prove difficulty assumptions in this setting. In order to improve the realism of this model, we introduce the notion of pseudo-random families of groups. We show how to reduce the security of a problem in such a family to the security of the same problem in the generic group model and to the security of an underlying strong pseudo-random family of permutations.