Enhancing Privacy-Preserving Multi-Authority Attribute-Based Encryption: Addressing Rogue-Key Attacks Under Adaptive Corruption of Authorities

A practical Multi-authority attribute-based encryption (MA-ABE) scheme with privacy-preserving properties realized by Inner Product Predicate Encryption (IPPE) was introduced by Michalevksy and Joye at ESORICS 2018. It requires secret keys linked to an attribute vector v and ciphertext associated with policy vector x to satisfy a certain predicate P(x, v) for decryption. Notably, the attribute information remains undisclosed during ciphertext inspection, provided no additional information about x and v is available. Despite extensive discussions in the literature concerning its efficiency, decentralization, and access policy structure, the security of certain schemes under the fully adaptive security model—first defined by Datta, Komargodski, and Waters at Eurocrypt 2023, which reflects real-world situations—has not been sufficiently explored. Specifically, their scheme is vulnerable to rogue-key attacks: an adversary can compromise a single attribute authority and then adaptively delay the publication of the compromised authority’s public key, crafting it based on the public keys of all other authorities. This ability to introduce a rogue key by compromising a single authority poses a significant security risk, as the adversary could decrypt any ciphertext without satisfying the policy. Addressing this gap, we propose an enhanced MA-ABE scheme building on Michalevsky and Joye (ESORICS’18), offering proof of resistance against rogue-key attacks. Additionally, we have conducted simulations of both the original and the enhanced schemes to evaluate the defense mechanisms’ resource requirements under consistent parameters and analyzed the cost implications of rogue-key attacks as the number of attribute authorities increases in the scheme.