Leleko S. Mathematical models and methods of signal detection on the background of non-Gaussian noise by moment quality criterion

In the dissertation work the scientific and technical problem of the use and development of methods of mathematical and computer simulation of signal detection processes on the background of non-Gaussian noise is considered and solved. In the construction of algorithms, the adapted moment quality criterion in kind of Neyman-Pearson criterion, the moment-cumulant models, are used, which allows to increase the accuracy of detection in the systems of data reception and processing by taking into account parameters of the non- Gaussian noise in the form of cumulants of higher orders. The method of synthesis generalized decision rules for the detection of constant signals, radio signals and radio signals with amplitude fluctuations with additive interaction with non- Gaussian noise with uniformly and unequally distributed sample values in the form of stochastic polynomials is developed, which allows to get better quality score in comparison with known results. Computational algorithms are developed that implement synthesized decision rules and allow to detect a signal in non-Gaussian noise. It is shown that synthesized nonlinear decision rules have higher quality score detection than those of known linear decision rules. Software tools for computer simulation of detecting signals in non-Gaussian noise were created and the effectiveness of the algorithms is researched. Level of implementation: Research and production complex 'Photoprylad' (Cherkasy), Cherkasy state technological university (Cherkasy).