Branch-and-Price for a Survivable Network Design Problem

We consider a specific variant of the survivable network design problem suitable to model real world scenarios occurring in the extension of fiber optic networks. In this article, two mixed integer programming models, which can be solved by branch-and-price, are discussed and compared to existing approaches theoretically as well as by a computational study. We further discuss the usage of alternative dual-optimal solutions to stabilize our approaches and significantly reduce the computational times needed to solve the linear relaxations of our models. The obtained computational results show that both branch-and-price approaches are suitable for solving small to medium sized problem instances to proven optimality.