On-line routing optimization for MPLS-based IP networks

We propose several heuristic algorithms that, using MPLS capabilities, provide network operators with a control framework for the routing plane. In particular, a monitoring block collects real-time measurements, such as the link load, and feeds them to a centralized decision block which runs a routing optimization algorithm. The output of the optimization process is then conveyed to network routers using standard network management protocols. Several optimization heuristics are proposed, from simple greedy algorithms to more complex meta-heuristics based on tabu-search optimization. Simulation results show the benefits of the proposed framework, which guarantees better performance compared to the classic static routing algorithms used in today's networks.