Proceedings of the 23rd Annual IFIP WG 11.3 Working Conference on Data and Applications Security XXIII
| Hi-index | 0.00 |
The random oracle model is an idealized theoretical model that has been successfully used for designing many cryptographic algorithms and protocols. Unfortunately, a series of results has shown that proofs of security in the idealized random oracle model do not translate into security in the standard model (basically synonymous with "real systems"), so the reasoning that protocols designed using random oracles are secure on real systems is heuristic at best, and fundamentally flawed at worst. In this paper, we consider how architectural changes taking place in real systems today, specifically the introduction of the trusted platform module, affect the realizability of random oracles. In particular, we show how a TPM that is only trivially enhanced from real, standard TPMs can leverage one of its most powerful capabilities -- the capability of keeping secrets from the host in which it resides -- in order to provide functionality that is indistinguishable from a true random oracle to any polynomial time adversary. In addition to a careful description of how this works, we provide security proofs based on assumptions of TPM security, and provide concrete performance estimates through benchmarks using a current TPM. To prove the security of our TPM-based scheme, we formally define and prove properties about a cryptographic primitive which we call a "hybrid pseudo-random function" that may be of independent interest.