UC-Secure Cryptographic Reverse Firewall–Guarding Corrupted Systems with the Minimum Trusted Module

Nowadays, mass-surveillance is becoming an increasingly severe threat to the public’s privacy. The PRISM and a series of other events showed that inner attacks such as subversion attacks may exist in the current network extensively. As an important strategy to defend users’ privacy against these attacks, cryptographic reverse firewall (CRF) is designed to be a middle-box, modifying all the messages coming in and out of a computer. However, the current formal definition of CRFs merely considers a single protocol session. If such a CRF applies to multiple entities, the security of every entity could not be deduced directly, which leads to an application limitation. In this work, we re-define the notion of CRF from a new perspective based on UC-emulation. Our new definition expresses all expected properties of a CRF in a more brief way, under the universal composition environment. We present a composition theorem which enables deploying one CRF for a local area of network rather than a single computer, and this can significantly reduce the number of CRFs used in practical applications.