Metaheuristic approaches for IP/MPLS network design.

This work introduces metaheuristic approaches for designing resilient and cost-effective multiprotocol label switching (MPLS) networks, a technology that is gaining prominent importance since most of the global data traffic is Internet traffic, and most internet protocol (IP) traffic within service provider backbones is being supported upon the IP/MPLS technology. Our approach is innovative because it integrates an overlay network design problem with the effective usage of traffic-engineering features of this technology. Due to the resulting complexity and a high level of technological detail, we decided to use metaheuristics to find solutions to prospective scenarios for two real-world applications. The best results were achieved using evolutionary algorithms and GRASP (Greedy Randomized Adaptive Search Procedure). The relative improvements for some of these scenarios are outstanding and reveal how using the protection mechanisms provided by newer technologies may advance efficiency standards more than legacy protection schemas. [ABSTRACT FROM AUTHOR]