The algorithm for synthesizing the correspondence table correlation of digits and entropy of 256-bit codes of “biometrics-code” neural network converter

A method for accelerated estimation of Shannon entropy for binary sequences, including those generated by the biometrics-code neural network converter, is proposed. The key idea is to establish a functional relationship be-tween the Pearson correlation coefficient of individual bits and the entropy of a 256-bit code, which eliminates the need for a complete enumeration of all bit configurations. At the initial stage, an array of pseudorandom sequences with a length of up to 256 bits is formed, each of which is divided into blocks of fixed length. For each block, the probability of occurrence of all possible combinations and their corresponding values of entropy are calculated, after which, by aggregation, taking into account interblock correlations, the total entropy of the entire sequence is determined. The resulting “correlation coefficient – entropy” pairs form a table in the correlation range from 0 to 1 in increments of 0.1 or 0.01, depending on the selected step parameter before starting the synthesis of the table. Further, the indicated tabular dependence is approximated by a polynomial function, which reduces the computational complexity of the problem from exponential to polynomial. The results of numerical modeling demonstrate that the approximation error does not exceed the accuracy of the direct Shannon method while reducing time costs. The synthesized table is based solely on the statistical characteristics of the sequence, which provides the possibility of universal application for codes generated by various biometric converters. It is advisable to integrate the presented approach into trusted low-bit microcontrollers and other devices with limited resources, as well as into Russian high-security operational systems (for example, Astra Linux) for an operational and reliable assessment of the randomness of the output sequence of “biometrics-code” neural network converter.