Dynamic Metering in Connected Urban Street Networks: Improving Mobility

Abstract

Traffic metering offers great potential for reducing congestion in oversaturated urban street networks. The available metering approaches rely on aggregated traffic flow dynamics to find metering rates that keep the number of vehicles inside congested areas within a predefined accumulation range. In such approaches, the aggregated traffic metering rates found for all gates surrounding the congested areas need to be divided among gates using a heuristic approach, which may compromise the quality of the suggested metering strategies. We developed a methodology to dynamically optimize the metering rates for each metering signal in an urban-street network. We utilized the Benders decomposition technique and developed a solution technique that finds metering strategies within a tight optimality bound. We also proposed distributed optimization and coordination algorithms to be able to optimize traffic metering rates in real time. This approach estimates vehicle densities across the network by using connected vehicle location information and loop detector data at discretized time steps. The estimations are shared among several computational nodes across the network and each node optimizes traffic metering rates for a sub-set of gates for several time steps ahead of the current network time. Once the optimized rates have been found, the rates for only the current network time are applied to the network. This procedure continues until the analysis period has ended. Several numerical analyses showed that traffic metering is an effective strategy to improve traffic operations in congested urban networks.