Gestión de la pérdida de datos en redes PTN para servicios con requisitos de QoS

  1. Rodríguez Pérez, Francisco Javier
Supervised by:
  1. José Luis González Sánchez Director

Defence university: Universidad de Extremadura

Fecha de defensa: 27 February 2015

Committee:
  1. Josep Lluís Marzo Lázaro Chair
  2. Rafael María Estepa Alonso Secretary
  3. Armando Ferro Vazquez Committee member
  4. Jesús Esteban Díaz Verdejo Committee member
  5. Tomás Pedro de Miguel Moro Committee member

Type: Thesis

Teseo: 378394 DIALNET

Abstract

This thesis elaborates a novel scheme to dynamically manage the performance of prioritized Forwarding Equivalence Classes (FEC) in congested Multiprotocol Label Switching Transport Profile (MPLS-TP) domains. It proposes a self-managed algorithm that allows Label Switched Routers (LSRs) within the network to utilize mechanisms within MPLS-TP to notify packet loss earlier. Moreover, the use of shorter routes is proposed for re-transmitted data. This all results in a more efficient recovery of lost packets of prioritized flows. The proposal is known as Gossip-based Local Recovery Policy (GLRP) and is offered as an Operation and Management (OAM) function for MPLS-TP. It consists of two main components or stages: First, the establishing of the management domain in parallel with the LSP signaling in the Control Plane and secondly notification and recovery of lost packets in the Forwarding Plane. They result in two novel extensions of the signaling protocol Resource Reservation Protocol with Traffic Engineering (RSVP-TE), which provides additional functionality in order to govern the behavior of some core LSRs when packet loss of prioritized flows arises. In addition, the extra features are discussed to improve the performance of the scheme and the ease with which the scheme can be integrated into an existing MPLS-TP infrastructure. Thus analytical and simulation data are provided in order to show the efficiency of the proposal under different conditions. Finally, topics for future research are identified.