Traffic engineering using new VS routing scheme

Abstract

Goal of traffic engineering in packet networks is to improve the network performance by providing support for congestion management, higher bandwidth utilization (or throughput), and QoS or priority-based services. Open shortest path first with traffic engineering extensions (OSPF-TE) (J. Moy, April 1998) and multiprotocol label switching (MPLS) protocols are commonly viewed as possible solutions. Both, the OSPF-TE with optimized link weights or MPLS with explicit optimal path set-ups, work ideally under static network conditions with a known demand-matrix. However, these methods do not provide scalability and flexibility to adapt to arbitrary, dynamic demand patterns in large networks. A novel traffic engineering scheme based on the virtual space (VS) routing (G. P. Saraph and P. Singh) is proposed here, which has the scalability, flexibility, and robustness to rapidly adapt to arbitrary, dynamic load conditions in large networks. Simulations are carried out on randomly constructed 40, 80, and 200-node networks with arbitrary demand matrices, which demonstrate excellent capability of the VS routing in terms of load balancing, packet throughput, and congestion avoidance. The proposed VS scheme is used for path selection in MPLS and integrated with the signaling protocols for path establishment, such as the constraint based routing (CR-LDP) (J. Ash et al., Jan. 2002) or resource reservation (RSVP-TE) (D. Awduche et. al., Jan 2002) protocols.