关键词: 生物识别, 云计算, 外包计算, 数据安全, 隐私保护

Abstract: Biometric Identification（BI） is to identify an unknown individual by matching his biometric trait against a pre-established biometric database. In the era of E-service, BI has recently been widely used. As cloud computing is beginning more and more popular, BI can be aided with cloud service, which helps reduce the computation costs of the database owner. However, the connection of BI and cloud computing brings some new privacy concerns, such as leakage and unauthorized use of individuals’biometrics. This paper studis the problem of cloud based biometric identification. It firstly presents the security analysis of a recent scheme on BI outsourcing, and discloses its privacy flaws, then combines the techniques of data splitting and matrix transformation, and constructs an improved solution called EBIO. Theoretical analysis is presented to evaluate the proposal in terms of correctness and security. Lastly, the paper conducts extensive experiments on a prototype of the scheme. The experimental results show that the proposal is efficient and appropriate for practical use.

Key words: biometric identification, cloud computing, outsourcing computation, data security, privacy preserving