Traffic Engineering in G-MPLS networks with QoS guarantees

In this paper a new Traffic Engineering (TE) scheme to efficiently route sub-wavelength requests with different QoS requirements is proposed for G-MPLS networks. In most previous studies on TE based on dynamic traffic grooming, the objectives were to minimize the rejection probability by respecting the constraints of the optical node architecture, but without considering service differentiation. In practice, some high-priority (HP) connections can instead be characterized by specific constraints on the maximum tolerable end-to-end delay and packet-loss ratio. The proposed solution consists of a distributed two-stage scheme: each time a new request arrives, an on-line dynamic grooming scheme finds a route which fulfills the QoS requirements. If a HP request is blocked at the ingress router, a preemption algorithm is executed locally in order to create room for this traffic. The proposed preemption mechanism minimizes the network disruption, both in term of number of rerouted low-priority connections and new set-up lightpaths, and the signaling complexity. Extensive simulation experiments are performed to demonstrate the efficiency of our scheme.