Traffic uncertainty models in network planning

In network planning one designs the network and its routing policies based on a "guesstimate" of the amount of traffic that might be routed through the network in the future. Clearly, not being able to foretell the future means that one of the biggest challenges in network planning is how to deal with traffic uncertainty. If not solved appropriately, the network operator has to resort to overprovisioning, which is costly. On a smaller timescale, traffic that is being routed through the network is generally stochastic in nature, which is also reflected in dynamically varying network parameters like available bandwidth and delay of links. In this article, we provide an overview of several traffic uncertainty models useful for network planning. Our focus is on offline approaches to network planning and online routing in stochastic networks. We conclude with some possible directions for future work.