A Simulation Laboratory for Evaluation of
Dynamic Traffic Management Systems

Qi Yang (qiyang@alum.mit.edu)

Feburary, 1997 (revised version of Qi Yang's PhD dissertation)

PostScript files and PDF files of this document (193 pages) are available for anonymous download.

SIMLAB consists of a microscopic traffic simulator (MITSIM) and a traffic management simulator (TMS). MITSIM represents the ``real world.'' It accepts as input signal control and route guidance from TMS, and models the movement of individual vehicles in the network. TMS receives traffic surveillance data from MITSIM and generates signal controls and route guidance according to the ATMS and ATIS logic under evaluation. TMS supports simulation of a wide range of signal control, route guidance and incident management schemes. A mesoscopic traffic flow simulator is used within TMS to predict traffic conditions. This element is designed to model the ATMS/ATIS which generates traffic controls and route guidance based on anticipatory -- as opposed to historical or the most recent measurement of -- traffic conditions. The generic structure of SIMLAB allows the effectiveness and robustness of any traffic control and surveillance system to be evaluated in a computer-based laboratory environment.

The laboratory is implemented in C++ using object-oriented programming and a distributed environment. A graphical user interface allows users to visualize the simulation process, including animation of vehicle movements, measurement of surveillance sensors, state of traffic signals and signs, etc. SIMLAB was validated using several networks, including a 6-mile stretch of I-880 around Hayward, California. The applicability of the system was demonstrated in a case study using the A10 beltway in Amsterdam, The Netherlands. SIMLAB was used to evaluate the performance of two alternative approaches in providing real-time route guidance: one based on latest measurements and the other on traffic prediction. The results provide useful insights for the design of ATIS, indicate that the simulation laboratory operates as designed, and support its value as a tool for evaluation of advanced traffic management systems.

Thesis Supervisors:

Moshe E. Ben-Akiva
Professor of Civil and Environmental Engineering
Massachusetts Institute of Technology

Haris N. Koutsopoulos
Associate Professor of Civil and Environmental Engineering
Carnegie Mellon University

• Contents
• List of Figures
• List of Tables
• Introduction
  • Motivation and Research Objective
  • Literature Review
    • Advances in Dynamic Traffic Management Systems
    • Evaluation Methods
      • Field tests
      • Simulation based evaluations
    • Traffic Simulation Tools
  • Thesis Outline
• Evaluation Framework
  • Process for Evaluation and Design Refinement
  • Traffic Simulation Laboratory
    • Traffic Flow Simulator
      • Network components
      • Travel demand
      • Driving behavior and vehicle movement
    • Traffic Management Simulator
    • Surveillance System
    • Control and Routing Devices
  • Simulation Output and Measures of Effectiveness
  • System Integration
  • Software Architecture
    • Software Elements
    • Distributed Implementation
    • Graphical User Interface
• Microscopic Traffic Simulator
  • Overall Design
  • Network Representation
    • Nodes:
    • Links:
    • Segments:
    • Lanes:
  • Toll Plazas
  • Traffic Surveillance and Control Devices
    • Surveillance Sensors
    • Traffic Control Devices
  • Incidents
  • Travel Demand
  • Vehicle Characteristics
    • Maximum acceleration rate:
    • Maximum deceleration rate:
    • Normal deceleration rate:
    • Desired speed:
    • Speed distribution across lanes:
    • Target speed:
  • Vehicle Routing
    • Route Choice Model
    • Route Switching Model
    • Properties and Extensions
  • Vehicle Movements
    • Reaction Time
    • Vehicle Loading
    • Acceleration Rate
      • Car following
      • Merging
      • Event Responding
    • Lane Changing
      • Lane Change Decisions
      • Gap acceptance
      • Nosing and Yielding
  • Simulation Output
    • Sensor Readings
    • Measures of Effectiveness (MOE)
    • Graphical User Interface
      • Traffic network:
      • Traffic sensors:
      • Traffic controls and incidents:
      • Animation of vehicle movements:
  • Validation
• Traffic Management Simulator
  • Framework and Overall Structure of TMS
  • Route Guidance
    • Reactive Route Guidance
    • Predictive Route Guidance
      • Rolling Horizon Model
      • Network State Estimation
      • Network State Prediction
      • Guidance generation
  • Traffic Control
    • Static Controllers
    • Pretimed Controllers
      • Offset:
      • Timing Table:
    • Traffic Adaptive Controllers
      • Signal records:
      • Phase records:
      • Detector records:
    • Metering Controllers
      • Cycle length:
      • Limiting parameters:
      • Regulators and desired state:
      • Queue detectors:
      • Step size for updating metering rate:
  • Incident Management
    • Response Plan
      • Situation code
      • Device type
      • Affected region
      • Signal/Sign state
    • Incident Detection Time
    • Selection and Activation of Response Plan
• Mesoscopic Traffic Simulator
  • Network Representation
  • Traffic Cells and Traffic Streams
  • Capacity Constraints
  • Traffic Dynamics
    • Speed-Density Model
    • Cell-Following Model
      • Step 1:
      • Step 2:
  • Vehicle Characteristics
  • Vehicle Routing
  • Input and Output
  • Computational Tests and Validation
• Case Study
  • The Network
  • Sensor Data
  • OD Flows
  • Path Generation and Vehicle Routing
    • Specification of Route Choice Model
    • Path Table
    • Historical Link Travel Times
  • Calibration of MITSIM and MesoTS and Validation of MesoTS
  • Value of Real-Time Route Guidance
    • Scenarios
    • Measures of Effectiveness (MOE)
    • Results
  • Computational Performance
    • MITSIM:
    • MesoTS:
    • SIMLAB:
  • Conclusion
• Conclusion
  • Research Contribution
  • Future Work
    • Evaluation of alternative dynamic traffic assignment (DTA) models:
    • Evaluation of various traffic management strategies:
    • Operator training:
    • Laboratory travel behavior studies:
    • Network state estimation:
    • OD estimation and prediction:
    • Capacity translator:
    • Fuel consumption and emissions models:
    • Hybrid simulation:
    • Distributed and parallel simulation:
• References
• Abbreviation
• Calculation of Time-Dependent Shortest Paths
• Simulation Parameters
  • Vehicle Characteristics
  • Driver Behavior
• Random Number Generator
• Statistics for Evaluating Simulation Models
• Examples of Data Files
  • Network Database
  • Time Dependent OD Trip Tables
  • Vehicle Trip Table
  • Vehicle Path Table
  • Incidents
• Acknowledgments

PostScript files and PDF files of this document (193 pages) are available for anonymous download.