Robust shortest path problems with uncertain costs

Data coming from real-world applications are very often affected by uncertainty. On the other hand, it is difficult to translate uncertainty in terms of combinatorial optimization. In this paper we study a combinatorial optimization model to deal with uncertainty in arc costs in shortest path problems. We consider a model where feasible arc cost scenarios are described via a convex polytope. We present a computational complexity result and we discuss exact algorithms for two different robust optimization criteria. We finally present experimental results showing that the proposed approaches are suitable to be used on realistic size instances.