Optimal LSP capacity and flow assignment using traffic engineering in MPLS networks.

摘要

Quality of Service (QoS) and Traffic Engineering (TE) capabilities are two important criteria in today's networks for supporting real-time applications. The main objective of TE in service providers' networks is to satisfy customers' traffic demands and minimize the congestion while simultaneously optimizing the network resource utilization.; Multiprotocol Label Switching (MPLS) plays a key role in the IP networks by providing QoS as well as TE features. It reduces the complexity of packet forwarding, and achieves the simplified connection-oriented forwarding characteristics of Layer 2 switching technologies while retaining the equally desirable flexibility and scalability of Layer 3 routing.; In this thesis, we treat the TE issue by setting up different types of Label Switched Paths (LSPs) through the MPLS network. We first propose an analytical LSP model that combines two different LSP transmission mechanisms: Layer 2 cut-through switching and Layer 3 default routing. Secondly, in order to reduce the overall operation cost, the optimization of the traffic flow distribution on a given network becomes an important goal of the TE. We propose an LSP traffic flow distribution optimization objective function to minimize the overall network transmission delay and distribute the traffic flow on each link evenly.