PRACTICAL SCHEME OF A HYBRID GENERATOR OF PSEUDORANDOMS BASED ON AUDIOENTROPY AND NONLINEAR BOOLEAN FUNCTIONS

Abstract

The article solves the urgent problem of generating crypto-resistant pseudorandom sequences (PRS) to ensure the cybersecurity of modern digital systems. PRS generators play a key role in the formation of encryption keys, initialization vectors, one-time authentication tokens, and other critical parameters of cryptographic protocols. Insufficient crypto-resistance of such generators can lead to the compromise of reliable protocols, which emphasizes the need to develop new methods that overcome the limitations of traditional deterministic approaches. Among these limitations are the predictability of sequences in the event of disclosure of the internal state, limited entropy of initial values, vulnerability to statistical attacks, and the absence of true randomness, which reduces efficiency in cryptographic applications. The article proposes and implements a practical scheme for constructing a hybrid pseudorandom sequence generator that combines audio entropy and nonlinear Boolean functions in algebraic normal form (ANF). The architecture is based on the dynamic synthesis of output sequences of modified Jiffy generators using ANF functions with high nonlinearity, which are generated randomly. Audio entropy obtained from natural microphone noise is used to initialize the initial states of the generators, which increases the unpredictability of the sequences. The architecture is based on a modified Jiffy generator with flexible shift register parameters (lengths of 47, 53 and 59 bits), fixed feedbacks based on characteristic polynomials with prime numbers and static configurations of the output function. Special emphasis is placed on the use of algebraic normal form (ANF) of Boolean functions, which allows for effective analysis and optimization of the properties of nonlinearity, correlation stability and cryptanalysis complexity. The proposed hybrid architecture combines physical sources of entropy with efficient pseudorandom generation algorithms, achieving a balance between unpredictability and performance. This is relevant for embedded devices and Internet of Things (IoT) systems, where computing resources are limited, classical sources of entropy are absent, and there are requirements for energy efficiency and real-time. Comprehensive testing using NIST SP 800-22 confirmed the high statistical quality of the generated sequences: for 7 configurations (nonlinearity 24–224) it showed 100% success.