1. ZACH MA WINTER 2015 A Parallelized Multi-Agent Transportation Simulation Using MASS MATMASSim

2. What Is Transportation Simulation?  Microscopic  Car-following model  Macroscopic  Traffic flow model

3. Why Multi-Agent Based?  “Intelligent” Agent  Dynamic environment  Much more similar to those in the real world  Two key aspects in transportation simulation  Transport planning  Traffic flow simulation model

4. Transport Planning  Static Traffic Assignment  Disaggregation by individual travelers  Temporal dynamics  Dynamic Traffic Assignment(DTA)  Add consideration of departure time  Con: Still is an aggregated model  Agent-based  On individual level  OD pair is replace by individual particles(agents)

5. Traffic Flow Simulation Model  Cellular Automata  Roads are divided into cells  Each cell can be either empty or occupied by a car  Drawback: Impractical for large numbers  Queue Based  Links are represented as queues  Performance increases by a factor of 10 to 100  Currently used by MATSim

6. Bottleneck  Performance:  Execution Module(mobsim) Balmer, M., K. Meister, and K. Nagel. Agent-based simulation of travel demand: Structure and computational performance of MATSim-T. ETH, Eidgenössische Technische Hochschule Zürich, IVT Institut für Verkehrsplanung und Transportsysteme, 2008.

7. MATSim Overview output execution replanning scoring controler analyses input config  Iterative process between execution, scoring, and replanning  Ultimate Goal --> User Equilibrium  Hard to achieve in dynamic model

8. Overall Architecture  Distribute computation of execution model into multiple nodes

9. Design & Data Flow  Map --> Network (Places)  Intersections --> Nodes (Place)  Roads --> Links (Place)  Population --> (Agents)  Travelers --> Persons (Agent)

10. Data Structure – Adjacency List Manual mapping for neighbours of each place

11. Network Program Structure CentralController LinkImpl QueueSimulation NetworImpl NodeImpl Scenario PlanImpl PopulationImpl ActivityImpl Link_MASS Node_MASS Network_MASS Element_MASS

12. Changes to MASS Java  Create neighbours variable within Place class to store neighbouring relationship between all place elements, along with accessor and mutator methods  Remove destinations parameter with all exchangeAll(), and sendMessage() on EXCHANGE_ALL TYPE  Replace destinations within Places_base with srcPlace.neighours

13. Current Progress (Implementation)  MASS modification - Completed  MASS changes completed and tested with simple parameter exchange  Test using MATSim’s sample XML input files (80%)  Integration with MATSim (50%)  Setting up own Git repository and sync through all workspaces  MATSim.Mobsim internal logic figured out (ObjectAid)  Refactor MATMASSim.Element_MASS with functionalities from MATSim.QueueSim.Link and MATSim.QueueSim.Node

14. Next Steps  Insert MASS main logic into MATSim.QueueSim.SimulationEngine  Testing and benchmark  Simple scenario within MATSim  Gotthard scenario  Have a set of 50 000 trips going to the same destination  Greater Zurich Area  Consisted of 1.62 million agents, contained 163k links

15. Appendix: Network XML Data <link id="1" from="1" to="2" length="3000.00" capacity="3600" freespeed="27.78" permlanes="2" modes="car" /> <link id="2" from="2" to="3" length="4000.00" capacity="1800" freespeed="27.78" permlanes="1" modes="car" /> <link id="3" from="3" to="2" length="4000.00" capacity="1800" freespeed="27.78" permlanes="1" modes="car" /> <link id="4" from="3" to="1" length="6000.00" capacity="3600" freespeed="27.78" permlanes="2" modes="car" />