Overall Design

The main structure of the traffic simulation model is an iteration of functions at specified frequencies (time-based) or when certain events occur (event-based). The simulation logic is summarized in the flowchart in Figure 3-1.

The simulation starts with the loading of simulation parameters, the road network description, and the scenario definition. Initialization of communication with the traffic surveillance and management modules also occurs. Once the simulator is initialized, an iterative procedure begins. The tasks performed within each iteration include:

  
Figure 3-1: Flow chart of the traffic simulation model

1. Update the state of traffic signals, signs, and incidents.
2. Update routing tables and calculate shortest paths to all destinations.
3. Read new origin-destination (OD) trip tables and generate corresponding virtual vehicle queues at the origin nodes.
4. Load vehicles from the virtual queues into the network.
5. Update vehicles acceleration rates and make necessary lane changes.
6. Advance vehicles to new positions and update their speeds. If a vehicle activates a sensor, the corresponding measures (speed, occupancy, etc.) are recorded by the surveillance system module. At the end of a lane, a vehicle is either removed from the network (if it arrives at its destination) or handed to the downstream lane.
7. Update the display if the graphical user interface (GUI) is enabled.
8. Calculate MOE and/or send network states to external MOE or GUI modules.
9. Update the simulation clock and move to next iteration.

In general, MITSIM uses a time-based simulation approach in processing the vehicle movements. The car-following, lane changing, and event and signal response functions are invoked for each vehicle at a specified interval (e.g. 1 second). Speeds and positions of the vehicles and states of surveillance sensors are updated at a higher frequency specified by the user (e.g. 1/10 or 1/2 second). This step size is subsequently used to advance the simulation clock. The step size , which is drawn from a given distribution and randomly assigned to individual vehicles, has to be greater than or equal to the step size and a modulus of . The actual step size applied for a particular vehicle at a given time may vary (but must be a modulus of ) during the simulation as drivers' reaction times differ in some special conditions (e.g. too close to the leading vehicle, making a lane change, or delayed at toll booth and red signals, etc.).