The paper considers models and methods for solving processing biometric data at the physical and channel levels. A classification of the parameters of such systems is given and those characteristics that play a decisive role in achieving the objective function in limited time intervals are distinguished from them. The work aims to analyze the possibilities of increasing the reliability of biometric devices at the channel level by using error correction codes. The practicability and necessity of error-correcting coding mean using the permutation data decoding algorithms capable of matching biometric identification adequately. Based on the results of statistical modelling of optical communication channels, the necessary and sufficient conditions for the application of permutation decoding tools for binary codes are determined. The problem of minimizing the memory size of the cognitive map of a permutation decoder is solved by allocating the permutation orbits and using the generating combinations of loops as indexes of the reference matrices. An efficient algorithm is proposed for finding a unique orbit number and the corresponding reference matrix when the receiver generates an arbitrary permutation of characters from the set of permissible permutations.