Mul_STK: Efficient and privacy-preserving query with spatio-temporal-keyword multiple attributes in cloud computing

With the explosive growth of spatio-temporal-keyword data and the popularity of cloud computing, data owners often encrypt and outsource massive data to cloud servers to provide secure query services. To improve query efficiency, cloud servers typically optimize the organization of massive spatio-temporal data for efficient keyword-based query. However, for the multi-attribute query, the existing works lack an integrated coding theory, which cannot realize a parallelized and efficient query. Moreover, the existing serialized query for each attribute is inefficient and leads to users’ privacy leakage. To address these issues, we propose a privacy-preserving and efficient multi-attribute query scheme in cloud computing for massive data scenarios (Mul_STK), which can realize the following two guarantees for outsourced computing. Firstly, to realize the parallelized and efficient query with multiple attributes, we design a multi-attribute unified encoding technique to encode multiple attributes into unified vectors and construct an STK-BH tree structure. We further design an efficient filtration-verification query algorithm based on the STK-BH tree to fully utilize the characteristics of multi-dimensional attributes and realize parallelized dynamic pruning query. Secondly, to realize a secure multi-attribute query, three secure atomic predicate encryption protocols are constructed based on techniques of improved symmetric homomorphic encryption (iSHE), advanced encryption standard (AES), and lightweight matrix encryption. In addition, we combine these secure protocols with the efficient filtration-verification algorithm to propose Mul_STK, which guarantees the balance between efficiency and privacy-preservation in cloud computing environments. Security analysis and experiments show that Mul_STK achieves high query efficiency in cloud computing while ensuring data privacy, query privacy, and access pattern privacy.