Capacitated Network Design with Uncertain Demand

We consider a capacity expansion problem arising in the design of telecommunication networks. The problem is to install capacity on links of the network so as to meet customer demand while minimizing total costs incurred. When studying this and related problems it is customary to assume that point to point demands are given. This will not be the case in practice, however, since future demand is generally unknown and the decision must be based on uncertain forecasts. We develop a stochastic integer programming formulation of the problem and propose an L-shaped solution procedure based on well-known cutting plane procedures for the deterministic problem. The algorithm was tested on two sets of real life problem instances and we present results of our computational experiments.