Comparative analysis of the results of pseudorandom number generators for digital noise generation

The paper presents the results of a study of the characteristics of five different pseudorandom number generators for use in digital noise generation problems used to mask signals in cybersecurity. The relevance of the study was conditioned by the growing need for high-quality masking methods that provide both effective performance and reliability of randomness, which is important for protecting confidential information in modern digital systems. The purpose of the study was to compare the PCG, Xoshiro128++, WELL512a, Mersenne Twister, and KISS algorithms in terms of their performance, statistical randomness, and ability to effectively mask a useful signal with noise. The performance of the algorithms was evaluated using BenchmarkDotNet. Standard NIST, Dieharder, and TestU01 tests were used to check the quality of sequence randomness. For the generated noise, a spectral analysis was performed using the power spectral density value. The masking efficiency was calculated by the signal-to-noise ratio, the results of the autocorrelation function, and the noise spectrogram. The results of the study showed that PCG and KISS are the most productive in terms of speed, which makes them attractive for applications where fast random sequence generation is important. WELL512a and PCG demonstrated the highest randomness quality, consistently passing all statistical tests. Analysis of the spectral noise distribution showed that all generators provide a uniform power distribution before filtering, and after filtering, the noise is successfully limited in the high-frequency range. The signal-to-noise ratio for all algorithms was about -13.6 dB, which indicates similar efficiency in noise masking. Autocorrelation analysis confirmed a low correlation for all generators outside of zero lag, which is important for maintaining the quality of randomness in long sequences. The practical value of the study lies in the selection of the optimal pseudorandom number generator for noise reduction problems in cybersecurity. The results obtained provide recommendations for choosing algorithms based on their speed and randomness, which will ensure a high level of information protection in digital systems