Heterogeneous teletraffic models

Since the beginning of the '90s, accurate traffic measurements carried out that traffic exhibits some new properties such that selfsimilarity, long-range dependence and heavy tails. These features, which cannot be captured in a parsimonious way by traditional Markovian models and led to the introduction of new models in network traffic modeling. Such models have been proposed in the papers of several authors. In the framework of these models it has been shown that under large aggregation traffic is well approximated by fractional Brownian motion or $\alpha$-stable Levy motion. In some sense these two models are alternative. The feature of these models is that in their construction independent identically distributed are used, i.e. the traffic is homogeneous. But in modern telecommunication systems phone calls, video and data are transmitted through one channel, so traffic is heterogeneous. In our report we represent the reporter and his colleagues results where various models of heterogeneous traffic are proposed and their properties are investigated. Moreover we introduce a new class of stochastic processes which extends the class of selfsimilar processes.