Countering statistical attacks in cloud-based searchable encryption

• Main Authors: Ahsan, M. A. Manazir, Ali, Ihsan, Idris, Mohd Yamani Idna, Imran, Muhammad, Shoaib, Muhammad
• Format: Article
• Published: Springer/Plenum Publishers 2020
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://eprints.um.edu.my/36627/

Searchable encryption (SE) is appearing as a prominent solution in the intersection of privacy protection and efficient retrieval of data outsourced to cloud computing storage. While it preserves privacy by encrypting data, yet supports search operation without data leakage. Due to its applicability, many research communities have proposed different SE schemes under various security definitions with numerous customary features (i.e. multi keyword search, ranked search). However, by reason of multi-keyword ranked search, SE discloses encrypted document list corresponding to multiple (secure) query keywords (or trapdoor). Such disclosure of statistical information helps an attacker to analyze and deduce the content of the data. To counter statistical information leakage in SE, we propose a scheme referred to as Countering Statistical Attack in Cloud based Searchable Encryption (CSA-CSE) that resorts to randomness in all components of an SE. CSA-CSE adopts inverted index that is built with a hash digest of a pair of keywords. Unlike existing schemes, ranking factors (i.e. relevance scores) rank the documents and then they no longer exist in the secure index (neither in order preserving encrypted form). Query keywords are also garbled with randomness in order to hide actual query/result statistics. Our security analysis and experiment on request for comments database ensure the security and efficiency of CSA-CSE.