(Mohammad A. Traffic control using forecastihg models in networks with Guarantied Quality of Service

In this thesis new method of ABR traffic control is developed using the forecasting model. As a forecasting model a neural network Back Propagation is suggested. Neural network using current data of current cell rate and buffer loading in ATM switches forecasts the expected buffer loading in the next time moment and in case of congestion calculates the desired cell rate in the next moment. The Imitation Model of ATM network developed and implemented. The simulation of ATM network with different ABR traffic control methods was performed and the efficiency control methods of traffic control using forecasting model was estimated and compared to the conventional method Rate-Based.