Toll-design problem with stochastic route choice

Congestion pricing has been regarded as an effective method of reducing network-wide travel cost. Previous work on the toll-design problem focused on the deterministic case, that is, it is assumed that travelers have perfect information on the cost of traveling on every route of the network and they always choose the one with the least cost. Limited literature is available on the stochastic case and most of it is focused on marginal social cost pricing with the underlying assumption that all links can be tolled. In this study, a toll-design problem based on stochastic route-choice behavior for multiple user groups is presented under a more realistic context - only a subset of links can be tolled. The methodology is tested with real-world network data. The performance of various algorithms is also discussed.