Biometric identification provides a promising method in access control to authenticate users by their biometric traits, i.e. physiological traits (e.g. fingerprints, iris, face) and behavioral traits (e.g. voice, gait, typing rhythms). Biometric traits can not be lost, stolen, or forgotten like passwords, since they are naturally bound up with individuals [1], [2]. Just as Schneier has said: “You are your key” [3]. Due to such a strong bond, biometric identification is a much more reliable and convenient approach than passwords, the most traditional authentication method. With the boom of smartphones, the integration of biometric sensors into mobile phones has boosted the adoption of biometric technologies. For instance, more online banks seek to incorporate biometric identification into their systems [4]. In addition to the online payment, there are many other access control scenarios, which will lead to a growing demand for biometric technologies, such as industrial IoT deployment [5], telecare medical information system [6], smart city [7] and other applications [8]–[9][10]. According to Research and Markets Ltd., the largest market research store, the global biometric market will reach $42.4 billion by 2021 [11].

Despite the positive prospect and trend in biometric identification, there are still many challenges concerned with the privac