Amer M. Methods and models for classification and prediction of telecomunication systems states based on modified neural networks

Object - process of classification and prediction of telecommunication system states performed by network management systems. The purpose - development of methods for classification and prediction states during heterogeneous traffic transmission. Methods - classification theory, neural networks theory, complex systems modelling theory, probability theory and mathematical statistics. Results - a model is developed for classification and prediction states of a router based on a set of modified neural networks; a method is developed for classification and prediction states of a telecommunication system based on a set of probabilistic neural networks. It is introduced - in research project No. DR 0109U000662, in academic activity, as well as in air data system project of CS Consulting company aimed at development of traffic balancing system in a telecommunication system segment. Sphere of application - telecommunication systems based on modified neural networks