Network flow models for the local access network expansion problem

In the local access network expansion problem we need to expand a given network (with tree topology) due to traffic demand increase. We can expand the network either by increasing the capacity of its edges and/or by installing equipments that concentrate the traffic of several demand points. This problem has received some attention in the literature. Here, we view the problem as an extension of the well-known capacitated minimum spanning tree problem. We adapt two types of flow-based formulations, aggregated and disaggregated formulations and present some valid inequalities to improve the linear programming bounds of the previous formulations. Computational results from a set of tests with 100 and 200 nodes that compare the lower bounds given by the different models, new and old, are reported. Our results demonstrate that the flow-based models are worth trying for these problems.