A Near-Optimal Solution Approach for the Multi-hop Traffic Grooming Problem

In this paper we consider a capacity sizing and routing problem in synchronous digital hierarchy/wavelength division multiplexing networks with nodes having switching capabilities. This problem is well known in the literature as the multi-hop traffic grooming problem and is generally formulated as an integer linear program, which is especially hard to solve when the demand channels are unsplittable and nonuniform. To overcome this difficulty we develop a branching strategy, using a lower bound that is close to the optimal solution. We also devise a reformulation which accelerates branch-and-bound-based solvers. The computational results clearly show that our method is effective in reducing execution time as well as memory consumption, in comparison to traditional methods based on branch-and-bound.