Encrypted data inner product KNN secure query based on BALL-PB tree

With the increased data volume, data query service outsourcing to cloud servers is widely used. However, enabling authorized users to access confidential data is critical when conducting queries on untrusted cloud servers. To this end, an inner-product k-nearest neighbor (KNN) query scheme with access control (IPKNN_AC) is proposed under privacy protection. Firstly, this scheme utilizes the designed Ball-PB tree structure to partition the dataset into multiple subsets. Considering the efficiency and confidentiality of inner-product queries on the tree, the scheme represents the internal nodes and leaf nodes accordingly and defines a secure inner-product calculation protocol, EncInp. Secondly, relying on EncInp, a query algorithm is employed to perform similarity inner-product queries on the encrypted tree representation. Finally, the scheme is shown to be secure through security proofs of homomorphic encryption. Experimental evaluation results on medical datasets demonstrate the effectiveness of the scheme.