Efficient privacy-preserving frequent itemset query over semantically secure encrypted cloud database

Abstract

In recent years, more users tend to use data mining as a service (DMaaS) provided by cloud service providers. However, while enjoying the convenient pay-per-use mode and powerful capacity of cloud computing, users are also threatened by the potential risk of privacy leakage. In this paper, we aim to efficiently perform privacy-preserving DMaaS, and focus on frequent itemset mining over encrypted database in outsourced cloud environment. Existing work apply different encryption methods to design various privacy-preserving mining solutions. Nevertheless, these approaches either cannot provide sufficient security requirements, or introduce heavy computation costs. Some of them also need users staying on-line to execute computations, which are not practical in real-world applications. In this paper, we propose a novel efficient privacy-preserving frequent itemset query (PPFIQ) scheme using two homomorphic encryptions and ciphertext packing technique. The proposed scheme protects transaction database with semantic security, preserves mining privacy and resists frequency analysis attacks. Meanwhile, efficiency is guaranteed by inherent parallel computations for packed plaintexts and users could stay off-line during the mining process. We provide formal security analysis and evaluate the performance of our scheme with extensive experiments. The experiment results demonstrate that the proposed scheme can be efficiently implemented on large databases.