Three-Patterns-Protected Searchable Encryption Supporting Disjunctive Keyword Search

Searchable encryption (SE) enables the client to execute keyword searches in encrypted data stored on the untrusted server and has been widely studied in cloud storage. To achieve higher efficiency and more functionalities, most SE schemes allowed the client to leak some information to the server. These leaked information are commonly referred to as leakage patterns. There are three important leakage patterns: search pattern, access pattern and volume pattern. Recent research has exploited at least one of these three patterns to attack SE schemes, resulting in the compromise of the confidentiality of encrypted data and queried keywords. Although existing SE schemes support conjunctive keyword search and protect these three patterns, these schemes do not support disjunctive keyword search and have a higher computational cost. In this paper, we use a private set union protocol based on additively symmetric homomorphic encryption to construct an SE scheme, which not only protects three patterns but also supports disjunctive keyword search. Specifically, we design an efficient token generation algorithm to protect the search pattern and a non-naive padding method to protect the volume pattern. Furthermore, we prove the correctness of our scheme through theoretical analysis and strictly prove the security under the leakage function. Finally, performance evaluation demonstrates that our scheme supports disjunctive keyword search while achieving a favorable trade-off between leakage protection and efficiency. Moreover, for components that exhibit relatively higher overhead during evaluation, we introduce optimization strategies that effectively enhance search efficiency and scalability.