Performance Analysis of ATM Switch Queue under

One of the most important issues in designing a network and realizing a service is dealing with traffic characteristics. Former studies presented the models based on the Poisson or Markovian models, including conventional traffic prediction and analysis. Recent experimental research on LAN, WAN, and VBR traffic properties has highlighted that real traffic specificities can not be displayed because the current models based on the Poisson assumption under estimate the long range dependency of network traffic and self-similar peculiarities. Therefore, a new approach using self-similarity characteristics as a real traffic model was recently developed. This paper discusses the definition of self-similarity traffic. Moreover, real traffic was collected and applied to an ATM switch queue to compare artificial self-similarity traffic with traffic using the Poisson model. As a result, the probability of buffer overflow under a low-bound and cell loss was estimated in the case of applying self-similarity traffic to an ATM switch queue.