GROUP GAMMA-DISTRIBUTION AND NEURAL NETWORK IN OF THE LATEST TELECOMMUNICATION TRAFFIC MODELING

Abstract

This article considers queue formation in the M/D/1 system with statistical characteristics of the first two orders that areclose to real, as the purpose of telecommunication traffic modeling . The input stream to the system is considered to be a group stream with constant parameters of the package and distance between arrivals, influenced by gamma distribution. These parameters are determined by a neural network trained to determine parameters of such input streams according to statistical characteristics of the queue at various loads of device. The results obtained demonstrate a good ap-proximation with the use of gamma-ray fluxes. Parameters are evaluated with the use of the neural network. The practical usefulness of the considered approach and the prospects of using neural net-works for practical tasks in solving which queue theory is used are provided.