Modeling and signalizing intersections for large-scale transportation simulation network

MATSim-T (multi-agent traffic simulation toolkit) is a simulation platform for large-scale agent-based traffic. Each agent plans its day and performs its activities based on that plan. Then it chooses appropriate routes to reach its destinations. To avoid conflicts in intersections, signal control need to be added into MATSim-T. Phase modeling and a dynamic window signal timing algorithm are proposed to signalize the intersection for MATSim-T. The average trip duration of agents and stops in signalized links are tested according to the traffic demand in the street network of Zurich, Switzerland, which is generated by MATSim-T. The simulation results show that the average trip duration will stabilize after signal optimization even the initial green fraction is different, and the dynamic window strategy causes less stop than other traditional strategies.