# Effective bandwidth of non-Markovian packet traffic

## TL;DR

This paper applies advanced statistical mechanics techniques to evaluate the effective bandwidth of non-Markovian telecommunication traffic models, enabling performance assessment of communication channels under rare event conditions.

## Contribution

It introduces the use of the cloning method from non-equilibrium statistical mechanics to compute effective bandwidths for non-Markovian traffic models, addressing a gap in analytical solutions.

## Key findings

- Cloning method effectively estimates effective bandwidths.
- Non-Markovian models can be analyzed for rare events.
- Performance evaluation without analytical solutions.

## Abstract

We demonstrate the application of recent advances in statistical mechanics to a problem in telecommunication engineering: the assessment of the quality of a communication channel in terms of rare and extreme events. In particular, we discuss non-Markovian models for telecommunication traffic in continuous time and deploy the "cloning" procedure of non-equilibrium statistical mechanics to efficiently compute their effective bandwidths. The cloning method allows us to evaluate the performance of a traffic protocol even in the absence of analytical results, which are often hard to obtain when the dynamics are non-Markovian.