Internet protocol network design with uncertain demand

We present a case study concerning the design and dimensioning of the internet protocol network of TDC, the largest Danish network operator. Due to historical reasons the number of points of presence (POPs) in the network has reached a level, believed to be too high. To point out potential POPs for dismantling, we consider a network planning problem concerning dimensioning of the POPs and capacity expansion of the transmission links of the network. This problem is formulated as a two-stage stochastic program using a finite number of scenarios to describe the uncertain outcome of future demand. The problem is then solved by an L-shaped algorithm, and we report results of our computational experiments.