The concept of functional encryption (FE) is a generalization of standard encryption, which allows users to delegate to third parties the computation of certain classes of functions of the encrypted data by generating specific secret keys for these functions. Examples of FE schemes include attribute-based encryption (ABE), inner-product predicate encryption (IPPE), and functional encryption for all polynomial-size circuits. In ABE, the encryptor is able to dynamically choose the set of people who will be authorized to decrypt a given ciphertext, by defining attributes and corresponding access policies.
The secret keys can be associated with the attributes and the ciphertext with the policy or the other way around. In IPPE, ciphertexts and secret keys are associated with vectors and the decryption procedure only succeeds when the inner-product between these two vectors is equal to zero. Among the above-mentioned examples of FE schemes, the latter one in, which relies on the powerful notion of indistinguishability obfuscation, is the primitive that provides the highest functionality as it implies all of the previous ones. Unfortunately, it is also impractical, making it interesting only from a theoretical point of view. Hence, one of the main goals of WP4 is to find a reasonable trade-off between efficiency and expressiveness. More precisely, our goal is to design schemes that cover reasonably expressive functionalities while still being efficient. In this deliverable, we describe our progress towards this goal.
More specifically, this report describes two contributions made in the context of FENTEC. First, we describe the unbounded attribute-based encryption scheme proposed by Chen et al. with constant-size public parameters under static assumptions in bilinear groups. This scheme defines the current state of the art in the area of unbounded attribute-based encryption. Second, we describe the inner-product predicate encryption schemes introduced by Chen et al. The proposed schemes achieve adaptive security and full attribute-hiding in the prime-order bilinear group setting and improve current results by Okamoto et al.