Using metaheuristics for planning resilient and cost-effective multilayer networks

The deployments of low-cost resilient optical networks constitute a traditional family of NP-Hard problems. Recent trends to maximize the efficiency of the existing optical infrastructure through the virtualization of optical connections, has as its main drawback that a single point of physical failure can unleash simultaneous failures in multiple connections. From this intrinsic characteristic of next-generation-network architectures, several multilayer-network problems arise, whose complexity is at least as hard as those of the corresponding single-layer problems. This article addresses the problem of designing a minimum-cost, fault-tolerant IP/MPLS network, to be deployed over an existing DWDM infrastructure. Besides cost and resiliency, this work also integrates traffic and capacity constraints. Due to the complexity of the procedure, several metaheuristics were used to find cost-effective solutions to real-world scenarios provided by a South American TELCO.