Kudriavtseva N. Linear prediction polymodels of Gaussian and non-Gaussian stochastic processes

Object - Gaussian and non-Gaussian stochastic processes and signals and their statistical processing. Purpose - development of new classes of composite linear prediction models - generic linear prediction models for Gaussian and non-Gaussian stochastic processes, research of possibilities of application of these models in the different tasks of statistical radiophysics. Methods - linear system theory, difference equations, linear operators, numerical analysis methods, statistical modeling methods, numerical analysis, applied analysis of stochastic processes. Results - the linear prediction polymodels presented by multiplicative and additive classes of models for stochastic processes were received, the methods of parametric PSD with high resolution are improved, the method of factorization for multimode parametric PSD estimations is proposed. Implementation - in research activity at KNURE in the topics 153-7 and 260-3. Application - radiolocation, forecasting systems, spectral analysis