Cross-Domain Access Control Encryption: Arbitrary-policy, Constant-size, Efficient

Abstract

Access control is a fundamental keystone in security. Damgard, Haagh, and Orlandi (TCC 2016) introduced access˚ control encryption (ACE) that enforces no-read and no-write rules without revealing the senders, receivers, or the content of the encrypted traffic. Existing designs of ACE for arbitrary policy (covering all possibilities of read/write relationship) rely on indistinguishability obfuscation or lattice-based assumptions, with either exponential-size ciphertexts or circuit realization of policy. Also, their designs mandate a private sanitizer key to remain perpetually online for sanitization. The only existing scheme that can afford a public sanitizer key supports only simple policies. To summarize, state-of-the-art ACE schemes only feature at most two of the following desirable properties: arbitrarypolicy, constant-size (ciphertext), and efficient (sanitization). This paper introduces an ACE scheme for arbitrary policy without sanitizer key, which solves the open question posed by Kim and Wu (Asiacrypt 2017). We also put forth the notion of cross-domain ACE, separating the key generator into the sender-authority and receiver-authority. Our scheme requires structure-preserving signatures, non-interactive zero-knowledge proof, and sanitizable identity-based broadcast encryption as the building blocks. It can be instantiated directly from pairing-based assumptions and features constant ciphertext size. We also prototyped our scheme and demonstrated its practical efficiency.