A Secure Multi-keyword Ranked Search over Encrypted Cloud Data against Memory Leakage Attack

To obtain greater flexibility and cost savings, outsourcing private data to public cloud servers while enabling users to search the data becomes the first choice for more and more users. In view of security, the private data must be encrypted before outsourcing which makes the method of traditional keyword search infeasible. Therefore, searchable encryption is extensively explored in recent years. Taking the practicality into account, multi-keyword ranked search over encrypted data is essential. However, almost all of existing multi-keyword ranked search schemes are suffering the security threats of non-volatile memory leakage attack. To solve this problem, a secure multi-keyword ranked search scheme which resists memory leakage attack (MRSS-ML) is proposed. The proposed scheme utilizes physically unclonable functions (PUFs) to randomize the keywords and document identifiers. Owing to the noisy properties of PUFs, the fuzzy extractor (FE) is used to recover the secret keys. To further enhance the security of the proposed scheme, an order-preserving function is selected to encode the similarity scores. MRSS-ML can resist the memory leakage attack from inner or external attackers. Security analysis and experimental results show that the MRSS-ML scheme is efficient whilst achieve higher security requirements against memory leakage attack.