Kovalenko A. Methods and means of increasing data transfer efficiency in multiservice networks

The thesis is devoted to problem of minimization data packet transmission time at the bottleneck link. The problem is solved using reallocation of the bottleneck link limited capacity. Such reallocation is organized due to a fractal nature of information flows, and empirical probability of their rate changing. There was developed a method of predictive data transmission control for bottleneck link that takes into account the fractal property of incoming data flows along with empirical probability of rate changing that is determined by long-range correlation dependence. The method is different from existent in considering the fractal nature of telecommunication traffic that spreads along modern heterogeneous data networks. Using traffic prediction, it allows us to share the computational resource (or capacity) of bottleneck link in proportion. A methodic of congestion avoidance worked out. It is based on the proposed method of predictive data transmission control, and allows us to minimize data packet transmission time in a bottleneck link for the data flows with fractal properties.