Thèse soutenue

Inférence de congestion et ingénierie de trafic dans les réseaux
FR  |  
EN
Accès à la thèse
Auteur / Autrice : Vijay Arya
Direction : Thierry Turletti
Type : Thèse de doctorat
Discipline(s) : Informatique
Date : Soutenance en 2005
Etablissement(s) : Nice
Ecole(s) doctorale(s) : École doctorale Sciences et technologies de l'information et de la communication (Sophia Antipolis, Alpes-Maritimes)

Mots clés

FR

Résumé

FR  |  
EN

This thesis presents methods which help to improve the quality of congestion inference on both en-to-end paths and internal network links in the Internet and a method which help to perform multicast traffic engineering in Overlay Networks. First, we propose an explicit loss differentiation scheme which allows unreliable transport protocols to accurately infer congestion on end-to-end paths by correctly differentiating congestion losses from wireless losses. Second, we present two contributions related to Multicast-based Inference of Network Characteristics (MINC). MINC is a method of performing network tomography which infers loss rates, i. E. , congestion on internal network links from end-to-end multicast measurements. We propose a statistical verification algorithm which can verify the integrity of binary multicast measurements used by MINC to perform loss inference. This algorithm helps to ensure a trustworthy inference of link loss rates. Next, we propose an extended MINC loss estimator which can infer loss rates of network links using aggregate multicast feedbacks. This estimator can be used to perform loss inference in situations where the bandwidth to report multicast feedbacks is low. Third, we present efficient ways of encoding multicast trees within data packets. These encodings can be used to perform stateless and explicit multicast routing in overlay networks and thus achieve goals of multicast traffic engineering.