Expressive Search on Encrypted Data

• Main Authors: Tseng, Fu-Kuo, 曾輔國
• Other Authors: Chen, Rong-Jaye
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/10297396247490569512

博士 === 國立交通大學 === 資訊科學與工程研究所 === 104 === Nowadays, online storage services are available to host and manage user data in a more economical way than ever before. Studies have suggested the use of encryption schemes while posing a demand for searching directly on encrypted data. Recently, searchable encryption schemes are proposed to transform keywords into ciphertexts and search predicates into search tokens. In addition, the embedded testing function in these schemes can check whether the ciphertext satisfies the search token or not. Therefore, users can delegate search tokens to service providers to retrieve the interested encrypted data. However, the existing schemes deal with keyword-based predicates without supporting the predicates of partial keywords or specific patterns. Therefore, to enrich the expressiveness of searchable encryption schemes becomes a major research direction. This dissertation proposes three encryption schemes which are all based on elliptic curves and bilinear pairing. The contribution of this dissertation is mainly divided into three parts: (1) We make use of the existing keyword-based searchable encryption scheme to devise statistical procedures for a collection of ciphertexts such as variance or linear regression. (2) We propose a position-aware symbol-based construction for the predicates of partial keywords. This construction provides predicates such as ``the leading pattern is `ABC'" or ``the third position is `D'". (3) Finally, we devise a more advanced symbol-based searchable encryption scheme for the predicates described by regular languages. This construction supports more flexible symbol-based predicates such as ``with a symbol `B' before a symbol `A'" or ``with a substring `ABC'". With the help of these schemes, users can enjoy more efficient and flexible online storage services, while preserving keyword and predicate privacy.