A Simulation-based Approach to Reliable Signal Control

This paper uses a simulation-based optimization framework to investigate the added value of using higherorder distributional information from detailed traffic simulators to address signal control problems. In particular, this paper uses higher-order travel time distributional information to derive signal plans that account for travel time reliability. In order to derive computationally efficient methods that embed nonefficient simulators, information from other more tractable traffic models is used throughout the optimization process. The simulation-based optimization approach combines the information of expected link travel time with link travel time standard deviation. These two measures are used to address system efficiency and reliability considerations. The methodology is applied to the network of the Swiss city of Lausanne. Our approach proves that incorporating travel time reliability in the signal control problem would yield signal plans with reduced link travel time and reduced travel time variability.