Privacy-Preserving and Efficient Multi-Keyword Search over Encrypted Data on Blockchain

Abstract

Recent research has demonstrated searchable blockchains that not only provide reliable search over encrypted distributed storage systems but ensure privacy is preserved. Yet, current solutions focus on single-keyword search over encrypted data on the blockchain. To extend such approaches to multi-keyword scenarios, they essentially perform a single-keyword search for multiple times and take the intersection of the results. However, such extensions suffer from privacy and efficiency issues. In particular, the service peers, which process the search requests, will be aware of the intermediate results, which include the data associated with each of the encrypted keywords. Moreover, these multiple traversals incur long delays in performing the search requests one after another with an extra cost in calculating the intersection of multiple sets. Finally, the service peers will charge the data owner a lot for writing the vast intermediate results to the smart contract. In this paper, we propose a bloom filter-enabled multi-keyword search protocol with enhanced efficiency as well as privacy preservation. In the protocol, a low-frequency keyword selected by a bloom filter will be used to filter the database when performing a multi-keyword search operation. Because the keyword is of low frequency, the majority of the data will be excluded from the result, which reduces the computational cost significantly. Moreover, we propose to use pseudorandom tags to facilitate completing each search operation in only one round. In this way, no intermediate results are generated, and the privacy is preserved. Finally, we implement the protocol in a local simulated blockchain network and conduct extensive experiments. The results indicate that our multi-keyword search protocol outperforms the traditional method with an average of 14.67% less time delay and 59.96% less financial cost.