Efficient sanitizable signatures without random oracles

Abstract

Sanitizable signatures, introduced by Ateniese et al. (ESORICS ’05), allow the signer to delegate the sanitization right of signed messages. The sanitizer can modify the message and update the signature accordingly, so that the sanitized part of the message is kept private. For stronger protection of sensitive information, it is desirable that no one can link sanitized message-signature pairs of the same document. This idea was formalized by Brzuska et al. (PKC ’10) as unlinkability, which was followed up recently by Fleischhacker et al. (PKC ’16). Unfortunately, these generic constructions of sanitizable signatures, unlinkable or not, are based on building blocks with specially crafted features which efficient (standard model) instantiations are absent. Basing on existing primitives or a conceptually simple primitive is more desirable. In this work, we present two such generic constructions, leading to efficient instantiations in the standard model. The first one is based on rerandomizable tagging, a new primitive which may find independent interests. It captures the core accountability mechanism of sanitizable signatures. The second one is based on accountable ring signatures (CARDIS ’04, ESORICS ’15). As an intermediate result, we propose the first accountable ring signature scheme in the standard model.