Locating Encrypted Data Precisely without Leaking Their Distribution

Abstract

Data encryption is a popular solution to ensure the privacy of the data in outsourced databases. A typical strategy is to store sensitive data encrypted and map those original values into bucket tags for querying on encrypted data. To achieve computations over encrypted data, the homomorphic encryption (HE) methods are proposed. However, performing those computations needs locating data precisely. Existing test-over-encrypted-data methods cannot prevent a curious service provider doing in the same way and causing the leaks of original data distribution. In this paper, we propose a method, named Splitting-Duplicating, to support encrypted data locating precisely by introducing an auxiliary value tag. To protect the privacy of original data distribution, we limit the frequencies of different tag values in a given range. We use an entropy based metric to measure the degree of privacy protected. We have conducted some experiments to validate our proposed method.