Routing Strategies Under Demand Uncertainty

In this paper, we study network routing and traffic controls under demand uncertainty. Specifically, we examine the strategy of using a deterministic parameter in an optimization setting (a strategy employed in the literature) to represent the demand uncertainty, where traffic flows are modeled using the cell transmission model (CTM). For a special class of networks, for which instances have been previously analyzed in the literature, we provide an optimal policy (i.e., a policy whose solution is optimal for any realization of the demand). Using this optimal policy we show the problems inherent using a deterministic parameter to represent uncertainty. We then show that, for other types of networks, for which optimal policies do not exist, simple heuristics can outperform the use of optimization with a deterministic parameter that represents the demand uncertainty.