INFLUENCE OF CALCULATION ACCURACY OF TRAFFIC SELF-SIMILARITY EXPONENT ON A QUALITY OF SERVICE CHARACTERISTICS

Abstract

Abstract. The quality of service (QoS) characteristics in any telecommunication system depend
on its scheme, requirements servicing rules and, to the greatest extent, on the type of traffic generated by
the system requirements flow. In packet communication networks, a mathematical model of self-similar
traffic is used, where the time interval between packets is described by Pareto or Weibull distributions.
With an increase in the degree of self-similarity of packet traffic, the QoS characteristics in the system
significantly deteriorate compared to servicing, for example, Poisson traffic. But for such traffic there is no
reliable methodology for calculating the characteristics of the quality of service. The degree of self­
similarity of traffic is determined by the Hurst exponent H or the coefficient of self-similarity. The method
of increasing the accuracy of calculating the quality of service characteristics in a packet communication
network with self-similar traffic due to a more accurate determination of the self-similarity coefficient or
Hurst exponent depending on the parameters of the probability function of the distribution of the time
interval between packets is analyzed. For self-similar traffic, the accuracy of new formulas for calculating
the traffic self-similarity coefficient based on the shape parameter of the probability distribution has been
verified. After a more accurate determination of the Hurst exponent, the average value of the number of
packets in the system is calculated using the Norros formula, and then, from the approximation of the
distribution function of the system states, the probability of waiting for packet service is calculated. With
increasing accuracy of calculating the Hurst exponent, the accuracy of calculating the very characteristics
of the quality of service also increases. Simulation confirmed the higher accuracy of these methods for
calculating QoS characteristics in a system with self-similar traffic. Moreover, the differences in simulation
and calculation results do not exceed 3 ... 5%.