Aggregation of Markovian Sources: Approximations with Error Control

Abstract

One of the most challenging applications for high speed broadband networks is VBR video-conference and movies. Markovian models are a flexible tool to characterize a wide set of multimedia sources, including VBR video traffic. Specifically, these models are efficiently used for the description of a single video source. Nevertheless, teletraffic studies with superposed markovian sources have a computational complexity that grows exponentially with the increase of the number of the sources and this reduces the usefulness of these models in such an environment. The problem of characterizing the traffic generated by N superposed independent sources is addressed in the paper. Specifically, in the paper we define an aggregation method for the description of this kind of traffic. The state space of our aggregate model strictly depends on the required level of accuracy. Furthermore, we can decide a-priori the complexity of the aggregate model as function of the precision we want the model to have.

Work carried out under the financial support of CNR, in the framework of the project “Sistemi radiomobili multimediali nell’evoluzione verso UMTS”.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.