An analysis tool for markovian traffic model validation

Even though the problem of developing adequate models for self-similar data traffic is relatively old, it still remains unresolved. One of the limitations of the existing models is their lack of general applicability. Many traffic models re validated using a single trace only, such that the developed model may well fit the traffic of this trace, but not necessary of another one. Therefore, it is mandatory to use more than one trace for model validation, and in order to fit the collected traffic, specific descriptors of the traffic features have to be chosen (e.g. Hurst parameter, index of disperson of counts, etc). The used traces should stem from different areas, reflecting various user behaviour. In this paper, we present a tool box which helps modelers to analyse measured traces and to validate their developed Markovian traffic models.