COMPLEX CRYPTOGRAPHIC EVALUATION OF A HYBRID GENERATOR OF PSEUDORANDOMS BASED ON AUDIO ENTROPY AND NONLINEAR BOOLEAN FUNCTIONS

The article presents the results of a comprehensive cryptographic evaluation of a hybrid pseudorandom sequence generator (PRSG), which combines audio entropy as a source of initial randomness, multiple Jiffy generators and dynamically selected nonlinear Boolean functions in algebraic normal form (ANF). The key feature of the architecture is the use of stochastically synthesized ANF functions with guaranteed nonlinearity, which ensures the uniqueness of the combining function at each system launch. Multilevel statistical testing of the generated sequences was carried out using three independent test packages: NIST SP 800-22, Diehard and TestU01 SmallCrush. Seven generator configurations (from 6 to 9 basic generators) with different levels of nonlinearity of ANF functions (24-224) were investigated. The results of NIST testing showed successful passing of all 15 basic tests for each configuration with an average in the range of 0.341-0.531. Diehard tests confirmed the excellent quality of all 22 statistical checks without critical deviations. TestU01 SmallCrush confirmed passing of all 15 tests with minimum not lower than 0.0056. Cryptographic analysis of sequences revealed optimal