Multi-objective traffic grooming for survivable network design

In this paper, we consider reliable traffic grooming in optical network design with multiple objectives. The objectives are to maximize average network reliability/availability and to minimize the number of switching ports required in optical networks. Maximizing the average connection availability of all traffic demands normally requires a large amount of resources (i.e., switching ports). In contrast, minimizing the number of switching ports could block some traffic demands from having backup paths. Without backup paths and spare resources for user traffics, the network reliability/availability cannot be improved. In our simulation experiments, we implement and compare two traffic grooming techniques (MST and MRU). We also apply a path restoration technique to protect traffic demands against network failures. To solve the multi-objective optimization problem, we apply a multi-objective evolutionary algorithm (i.e., NSGA-II) for searching a set of non-dominated solutions. Our experiment results show that the non-dominate solutions from MRU provide superior results than those of MST grooming approach.