Fast Method for Combining/Generating Synthetic Traffic Traces Exhibiting Short- and Long-Range Dependence

摘要

We consider a heuristic approach for combining short traffic traces of various degrees of self-similarity to produce a longer trace that has the following desirable properties. The mixed trace is capable of modeling computer networking traffic that exhibits both short- and long-range dependence in a unified model. Synthetic traces of self-similar characteristics can be generated quickly due to the simplicity of our method, which is faster than the traditional approach of generating/approximating traffic traces from nothing.