A heuristic method for self-healing ring design in a single-homing cluster

Abstract

In this paper, we consider a SONET (Synchronous Optical NETwork) USHR (Uni-directional Self-Healing Ring) design problem for a single-homing cluster, i.e., a cluster with a single designated hub. The problem is formulated as a nonlinear integer programming problem and a branch and bound heuristic method based on the Lagrangian relaxation and subgradient optimization technique is proposed to handle the problem. In solving any ring design problem, we should deal with two different aspects of the ring design, namely, the ring routing aspect and the ring loading aspect. Both of these two aspects are well integrated and represented in our model. Such an integrated formulation has not been proposed in the existing literature mainly due to its computationally intractable complexity. In order to cope with such complexity, a preprocessing technique for reducing the complexity and several branch and bound strategies are proposed. The efficiency of the proposed method is tested through computational experiments. For the computational experiments, test problems are generated using the data obtained from the actual topologies in Seoul, Korea. The computational experiments show that the proposed method yields near-optimum designs within reasonable computation time.