Spline-approximation-based restoration for self-similar traffic
Irina Strelkovskaya, Irina Solovskaya, Nikolay Severin, Stanislav Paskalenko · Eastern-European Journal of Enterprise Technologies · 2017
The work considers the queuing system of the G/M/1/K with the Weibull distribution. The model for self-similar traffic was created within the Matlab Simulink software environment. Restoration of the self-similar traffic was obtained with the help of the spline approximation using linear and cubic splines. In this research, it has been discovered that the obtained self-similar traffic is characterized by “bursts”, “pulsations”, and the long-term dependence between arrivals. Linear and cubic spline approximations have been suggested to restore traffic. The approximations with the linear and cubic splines were used to restore smoothly changing self-similar traffic.The obtained results of the self-similar traffic restoration allow planning buffer capacities for NGN networking devices at the stages of design and further operation in order to avoid network overloads, excessive time delays and jitter for the case of packet traffic with bursts.The wavelet approximation is recommended for the accurate restoration of the self-similar traffic