Efficient and Privacy-Preserving Skyline Queries Over Encrypted Data Under a Blockchain-Based Audit Architecture

Skyline queries is an advanced data mining algorithm suitable for multi-criteria decision-making scenarios (i.e., medical pre-diagnosis). Privacy-preserving skyline queries schemes are usually constructed by certain methods of cryptography such as additive homomorphic cryptosystem, secret sharing technology, etc. Interestingly, these secure skyline queries schemes require that skyline computations do not reveal any message details, including encrypted inter-tuple domination relations, among which privacy schemes based on homomorphic cryptosystems are the most popular due to their strong security. However, existing secure skyline queries schemes not only suffer from low computational efficiency, but also do not have sufficient security for privacy-key management in the system. To address the above issues, this paper designs an efficient and privacy-preserving skyline queries over encrypted data under a blockchain-based audit architecture. Firstly, we propose a blockchain-based audit architecture that not only provides error auditing functionality but also makes our scheme suitable for (distributed) multi-user scenarios while providing secure key management in the system. Secondly, we implement a series of secure sub-protocols using the CRT-Based Paillier encryption algorithm and construct a privacy sparse matrix elimination protocol to reduce the size of the dataset, leading to a significant reduction in computational cost without compromising privacy. Finally, we put forward our secure skyline queries protocol and prove its security. The performance evaluation shows that our proposed method our proposed method is significantly more efficient (at least 7.4 times faster) compared to current methods.

I. Introduction

With the development of cloud computing technology, cloud service providers can provide data outsourcing computing services with ultra-high computing power and large storage capacity, which has attracted widespread attention from academia and industry. However, in some special application scenarios (such as medical treatment, finance, biological experiments, etc.), the data to be calculated is often highly sensitive. Therefore, there is an urgent need to embed privacy protection schemes in these outsourced services to ensure that computations can be performed without data disclosure. A major number of literature [1], [2], [3], [4], [5] have proposed work on protecting the confidentiality of outsourced computing.