Briagin O. Stochastic algorithms for multidimensional signal processing based on discrete samples

Object of research: the reflection process signal, including speech, in a multivariate random process. Objective: to extend the functionality of data transmission systems by developing a new method of digital processing of multidimensional random signals due to the development of the theory of discrete samples and its use in the radio. Methods: digital signal processing methods, methods of measurement and evaluation of stochastic processes, the method of discrete samples, methods of testing statistical hypotheses. Theoretical and practical results and novelty: rudimentary theory of discrete samples, based on which a method of digital processing of multidimensional random signals for use in radio engineering and telecommunication systems and proposed methods of assessing multivariate probability distribution function of stochastic signals and shows the conditions under which such assessments - unbiased and konzystentni derived formulas for calculating their confidence level and confidence intervals, first introduced criterion for assessing the degree of interdependence of samples of a random process and the concept of interdependence factor counts proved the reliability of this assessment, defined way to assess speech signal stationarity intervals and analyzed for its dependence multidimensional functions distribution on the choice of the value of the level of analysis, the coefficient of correlation between the reference frame and the individual characteristics of pronunciation, designed, manufactured and tested a prototype system transmitting - receiving speech signals with specially formed properties, based on the results obtained in the study of factor interdependence samples, the proposed way transmission and reception of speech signals using stochastic modulation, the first algorithm form a sequence of random samples with variable coefficient statistical relationship between them, based on assessment of the degree of interdependence multivariate random signal samples and to improve quality control of magnetic media a technology assessment as a working layer of magnetic media without burning it test signal. Degree of implementation: the research and development work of the State Research Institute of Ministry of Internal Affairs of Ukraine and the A1906 Military Unit of the Ministry of Defense of Ukraine. Sphere of use: Engineering, electrical engineering and telecommunications.