Optional multi-biometric cryptosystem based on fuzzy extractor

Following the wide use of smart devices, biometric cryptosystem is used to protect users' privacy data. However, biometric cryptosystem is rarely used in the scenario of mobile cloud, because the biometric sensors are different on various devices. In this paper, an optional multi-biometric cryptosystem based on fuzzy extractor and secret share technology is proposed. Each of the enrolled biometric modality generates a feature vector, and then the feature vector is put into a fuzzy extractor to get a stable codeword, namely a bit-string. All the codewords are used to bind a random key based on a secret share method, and the key can be used to encrypt users' privacy data. During the verification phase, part of the enrolled biometric modalities are enough to recover the random key. Therefore, the proposed scheme can provide a user the same biometric key on different devices. In addition, experiment on a virtual multi-biometric database shows that the novel concept of optional multi-biometric cryptosystem is better than the corresponding uni-biometric cryptosystem both in matching accuracy and key entropy.