On codes used in biometrical cryptosystems

Problems of using error-correcting codes in biometric cryptosystems are studied. Several constructions of codes with parameters better than parameters of the code from the original biometric cryptosystem of F. Hao, R. Anderson, and J. Daugman (2006) are proposed. A new upper bound for the size of a binary code based on its possibility to correct not more than $t$ errors with probability $1$ and $t+1$ errors with a probability $p$ is proposed. For the cases $t=0,1,2$ we study, it is possible to reach this bound.