## Thesis etd-06242004-163624 |

Link copiato negli appunti

Thesis type

Tesi di laurea vecchio ordinamento

Author

Orlando, Emanuele

URN

etd-06242004-163624

Thesis title

Applicability of Network Tomography in LAN and WAN Environments Applicabilità di Tomografia di Rete in ambienti LAN e WAN

Department

INGEGNERIA

Course of study

INGEGNERIA DELLE TELECOMUNICAZIONI

Supervisors

**relatore**Giordano, Stefano

**relatore**Tran-Gia, Phuoc

**relatore**Dott. Garroppo, Rosario Giuseppe

Keywords

- Network Tomography

Graduation session start date

27/07/2004

Availability

Full

Summary

The Internet world grows in size and complexity. Internet is a distributed network which continues to increase and evolve, becoming an heterogeneous and largely unregulated structure. It is vital for users and providers to characterize and to measure the performance of the network. Only in this way it is possible to detect and isolate its problems such as congested links. The large dimension and the limits imposed by administrative diversity, generally do not allow a direct access to the network but only to its small portion. Consequently there is a increasing need of practical and efficient procedures able to probe internal characteristics of a significant portion of the network.

A new approach to analyze the network is the so-called Network Tomography. Basically, Internet Tomography is an innovative technique which probes the network by measuring the end-to-end traffic behaviour to reconstruct the network internal performance. By intersecting end-to-end paths, the estimation of the characteristics of their common path can be inferred, without any cooperation from the internal network. The aim of this work is to give a basic notion of Internet Tomography and to proof how these new techniques can provide interesting information of the network. In particular the present work focus on a link level delay probability distribution estimation by applying the Internet Tomography in a LAN. The goal is to obtain detailed results and to test how the Internet Tomography can represent an useful instrument to infer the state of the network. Only the end-to-end measurements provide to draw inference of internal characteristics of the network.

This thesis is structured as following. The second Section mentions the basics of the inference delay model which represent the core of this work. The definition of the Maximum Likelihood Estimation and inference statistics notions are, in particular, provided in the second Section. The third Section describes the delay model using the Fixed Bin Size Discrete Model by Lo Presti. The discretization of the end-to- end delay experienced on a path allows to infer the delay probability distribution of the links along the path. This can be possible by the external measurement. The forth Section describes different tools to obtain the end-to-end measurements, such as Ping, Traceroute and Pathchar. The measurements and the inference model should cooperate to provide good results. This is the aim of the fifth Section. In particular it mentions how the discrete delay model and the measurements should be implemented in a LAN ambient. Finally the sixth Section describes the technical results obtained during the process of this work.

The main purpose of this work is to provide a simplest tool of statistic measurements for LAN and WAN environments. Its task is to adapt the Internet Tomography which generally

works in a high scale network, to a LAN and WAN environments.

A new approach to analyze the network is the so-called Network Tomography. Basically, Internet Tomography is an innovative technique which probes the network by measuring the end-to-end traffic behaviour to reconstruct the network internal performance. By intersecting end-to-end paths, the estimation of the characteristics of their common path can be inferred, without any cooperation from the internal network. The aim of this work is to give a basic notion of Internet Tomography and to proof how these new techniques can provide interesting information of the network. In particular the present work focus on a link level delay probability distribution estimation by applying the Internet Tomography in a LAN. The goal is to obtain detailed results and to test how the Internet Tomography can represent an useful instrument to infer the state of the network. Only the end-to-end measurements provide to draw inference of internal characteristics of the network.

This thesis is structured as following. The second Section mentions the basics of the inference delay model which represent the core of this work. The definition of the Maximum Likelihood Estimation and inference statistics notions are, in particular, provided in the second Section. The third Section describes the delay model using the Fixed Bin Size Discrete Model by Lo Presti. The discretization of the end-to- end delay experienced on a path allows to infer the delay probability distribution of the links along the path. This can be possible by the external measurement. The forth Section describes different tools to obtain the end-to-end measurements, such as Ping, Traceroute and Pathchar. The measurements and the inference model should cooperate to provide good results. This is the aim of the fifth Section. In particular it mentions how the discrete delay model and the measurements should be implemented in a LAN ambient. Finally the sixth Section describes the technical results obtained during the process of this work.

The main purpose of this work is to provide a simplest tool of statistic measurements for LAN and WAN environments. Its task is to adapt the Internet Tomography which generally

works in a high scale network, to a LAN and WAN environments.

File

Nome file | Dimensione |
---|---|

1Introduction.doc | 27.14 Kb |

2Tomogra...twork.doc | 331.26 Kb |

3DelayEstimation.doc | 402.43 Kb |

4Measuri...Delay.doc | 259.07 Kb |

5Applica...raphy.doc | 547.33 Kb |

6Results.doc | 185.86 Kb |

7Conclusion.doc | 34.82 Kb |

Bibliography.doc | 41.98 Kb |

Contents.doc | 41.98 Kb |

dedica.doc | 24.58 Kb |

frontespizio.doc | 26.62 Kb |

FRONTESPIZIOmio.pdf | 62.91 Kb |

Ringrazi...inali.doc | 25.09 Kb |

Contatta l’autore |