Privacy-Preserving kNN Spatial Query Using Voronoi Diagram

Abstract

AbstractWith the increased demand for outsourcing databases, there is a demand to enable secure and efficient communications. The concern regarding outsourcing data is mainly providing confidentiality and integrity to the data. This paper proposes a novel solution to answering kNN queries at the cloud server over encrypted data. Data owners transform their data from a native domain to a new domain to assist in nearest neighbors’ classification. The transformation is achieved by the Voronoi diagram, which transforms the data space into numerous small regions, simplifying the nearest neighbor search. However, because the regions that make up a Voronoi diagram are irregularly shaped, the search through the network becomes hard to accomplish.Thus, the solution includes a Grid-based indexing approach for the Voronoi diagram to expedite the kNN search. Additionally, a strong encryption algorithm, like AES, is used to encrypt the data objects being sent from the data owner to the cloud. The cloud service provider utilizes the proposed indexing scheme to identify a superset of the nearest neighboring objects to be sent back to the user. An authorized user wants to send encrypted kNN queries to the cloud where the query is processed over encrypted data. However, the user should not have a copy of the encryption key. Therefore, Trusted Party or a Trusted Machine (Query Proxy) was added that handles the encryption and decryption of the communication between the user and the service provider.KeywordskNN queryDatabase outsourcingSpatial dataVoronoi partitioning