Relations among notions of complete non-malleability: indistinguishability characterisation and efficient construction without random oracles

Abstract

We study relations among various notions of complete non-malleability, where an adversary can tamper with both ciphertexts and public-keys, and ciphertext indistinguishability. We follow the pattern of relations previously established for standard non-malleability. To this end, we propose a more convenient and conceptually simpler indistinguishability-based security model to analyse completely non-malleable schemes. Our model is based on strong decryption oracles, which provide decryptions under arbitrarily chosen public keys. We give the first precise definition of a strong decryption oracle, pointing out the subtleties in different approaches that can be taken. We construct the first efficient scheme, which is fully secure against strong chosen-ciphertext attacks, and therefore completely non-malleable, without random oracles.