Achieving efficient conjunctive keyword searches over encrypted data

Abstract

We present two provably secure and efficient schemes for performing conjunctive keyword searches over symmetrically encrypted data. Our first scheme is based on Shamir Secret Sharing and provides the most efficient search technique in this context to date. Although the size of its trapdoors is linear in the number of documents being searched, we empirically show that this overhead remains reasonable in practice. Nonetheless, to address this limitation we provide an alternative based on bilinear pairings that yields constant size trapdoors. This latter construction is not only asymptotically more efficient than previous secure conjunctive keyword search schemes in the symmetric setting, but incurs significantly less storage overhead. Additionally, unlike most previous work, our constructions are proven secure in the standard model. © Springer-Verlag Berlin Heidelberg 2005.