Improving User Privacy in Identity-Based Encryption Environments

Abstract

The promise of identity-based systems is that they maintain the functionality of public key cryptography while eliminating the need for public key certificates. The first efficient identity-based encryption (IBE) scheme was proposed by Boneh and Franklin in 2001; variations have been proposed by many researchers since then. However, a common drawback is the requirement for a private key generator (PKG) that uses its own master private key to compute private keys for end users. Thus, the PKG can potentially decrypt all ciphertext in the environment (regardless of who the intended recipient is), which can have undesirable privacy implications. This has led to limited adoption and deployment of IBE technology. There have been numerous proposals to address this situation (which are often characterized as methods to reduce trust in the PKG). These typically involve threshold mechanisms or separation-of-duty architectures, but unfortunately often rely on non-collusion assumptions that cannot be guaranteed in real-world settings. This paper proposes a separation architecture that instantiates several intermediate CAs (ICAs), rather than one (as in previous work). We employ digital credentials (containing a specially-designed attribute based on bilinear maps) as the blind tokens issued by the ICAs, which allows a user to easily obtain multiple layers of pseudonymization prior to interacting with the PKG. As a result, our proposed architecture does not rely on unrealistic non-collusion assumptions and allows a user to reduce the probability of a privacy breach to an arbitrarily small value.