The puncturable registered attribute-based keyword search (PR-ABKS) scheme advances secure cloud data sharing by allowing encrypted retrieval and local key restriction in a single framework. Data owners encrypt files with descriptive keywords and ciphertext tags while specifying revocation lists that determine user exclusion during index generation. Users independently generate secret keys verified by a key curator (KC) who aggregates registration information without holding secret keys, reducing traditional key escrow risks. Crucially, users can locally update their keys to exclude access to certain tags—this "puncturing" ensures they lose decryption capability for marked ciphertexts while maintaining access to others. However, safeguarding confidentiality after key updates demands the complete and secure erasure of earlier key states; the scheme does not automatically retroactively update existing indexes if the exclusion list changes. This combination of encrypted search with selective, tag-level key revocation is a distinctive feature not present in prior attribute-based encryption mechanisms.
The system comprises four key entities interacting under clearly delineated responsibilities. The KC initializes system parameters, verifies and aggregates public user registrations, and issues auxiliary tokens necessary for partial decryption—without ever managing secrets directly. Data owners encrypt sensitive data under defined access policies, assign keywords and tags, and upload ciphertexts and searchable indexes to the cloud. The cloud server stores data and indexes, processes encrypted queries by evaluating trapdoors, and returns matched ciphertexts; it performs partial decryptions but cannot access plaintexts. Data users generate their own key pairs, register their public keys, create encrypted queries (trapdoors), and execute local key puncturing based on the owner’s exclusion policies. Importantly, revocation lists in the indexes are snapshots effective at generation time; stored indexes remain unchanged if exclusions later evolve.
PR-ABKS assumes an honest-but-curious cloud server that follows protocols but attempts to glean plaintext information if possible. The key curator complies with registration tasks but is not trusted with user secret keys. Users may be revoked or compromised, and adversaries can adaptively query or corrupt keys under constrained conditions. The security model targets confidentiality against chosen-plaintext attacks even after some key states are exposed, provided users properly erase previous key versions. User exclusion is enforced through revocation lists embedded at index creation; however, the scheme lacks mechanisms for updating stored indexes or ciphertexts retrospectively. Keyword privacy is maintained by the encryption of query trapdoors and indexes. The model excludes malicious users who retain earlier unpunctured keys. Thus, the scheme’s data confidentiality hinges on secure key updating paired with accurate exclusion at index formation.
The scheme employs bilinear pairings over cyclic groups and bases security on the Decisional Bilinear Diffie-Hellman assumption. User revocation is managed with a binary tree where leaf nodes represent individual users, and coverage sets identify minimal nodes encompassing non-revoked users. Each index embeds a revocation snapshot governing which users can decrypt corresponding ciphertexts. Paths from the root to user leaves help in efficient key management and cover calculations. This structured approach supports fine-grained exclusion as owners define ciphertext tags and revocation lists aligning with user subsets.
Investigators assessed PR-ABKS for computational, communication, and storage costs relative to puncturable registered ABE (PR-ABE) and registered revocable ABKS (RRABKS). Under typical configurations, registration and search costs approached those of baseline schemes, implying efficient scalability. However, the combined functionality introduces added overhead for encryption and decryption, reflecting the complexity of integrating encrypted keyword search with puncturable key updates. This cost analysis clarifies the trade-offs practitioners must consider when deploying PR-ABKS. The authors note that index updates do not occur dynamically with changes in exclusion lists, highlighting limitations around post-creation access control. Further evaluation is necessary to optimize these aspects for practical cloud sharing environments.