A Secure Voice Signature Based Lightweight Authentication Approach for Remote Access Control

Crypto-authentication schemes become unreliable whenever the private key is compromised, making them unfit for any system or network that requires high level of confidentiality. Key compromise is inevitable due to the wider space of operability of key in most of the cryptography based authentication schemes. To improve the performance of any authentication system, there is need to narrow its key operability to the key owner, that is eliminating the influence of other parties in the key generation and operability. However, this is quite difficult especially for any network that is characterised with low computation and energy, and relied on the third party for key management. In this paper, we propose the adoption of Mel-Frequency Cepstral Coefficients (MFCC) based voice signature based authentication scheme with a taint of cryptography operation. The scheme extracts the user’s voice signature as the hash of the MFCC parameters of the voice and unique passcode which is used for the authentication. We used exclusive or to filter off remnant of noise from MFCC values without incurring extra hardware cost. The performance evaluation results show authentication accuracy of 87.8% at low computation cost and communication overhead.