Date of Award
Doctor of Philosophy (PhD)
Chowdhury, Mashrur A
Sadek , Adel
Sarasua , Wayne
Ogle , Jennifer
This research developed a real-time traffic condition assessment and prediction framework using Vehicle-Infrastructure Integration (VII) with computational intelligence to improve the existing traffic surveillance system. Due to the prohibited expenses and complexity involved for the field experiment of such a system, this study adopted state-of-the-art simulation tools as an efficient alternative.
This work developed an integrated traffic and communication simulation platform to facilitate the design and evaluation of a wide range of online traffic surveillance and management system in both traffic and communication domain. Using the integrated simulator, the author evaluated the performance of different combination of communication medium and architecture. This evaluation led to the development of a hybrid VII framework exemplified by hierarchical architecture, which is expected to eliminate single point failures, enhance scalability and easy integration of control functions for traffic condition assessment and prediction.
In the proposed VII framework, the vehicle on-board equipments and roadside units (RSUs) work collaboratively, based on an intelligent paradigm known as 'Support Vector Machine (SVM),' to determine the occurrence and characteristics of an incident with the kinetics data generated by vehicles. In addition to incident detection, this research also integrated the computational intelligence paradigm called 'Support Vector Regression (SVR)' within the hybrid VII framework for improving the travel time prediction capabilities, and supporting on-line leaning functions to improve its performance over time. Two simulation models that fully implemented the functionalities of real-time traffic surveillance were developed on calibrated and validated simulation network for study sites in Greenville and Spartanburg, South Carolina. The simulation models' encouraging performance on traffic condition assessment and prediction justifies further research on field experiment of such a system to address various research issues in the areas covered by this work, such as availability and accuracy of vehicle kinetic and maneuver data, reliability of wireless communication, maintenance of RSUs and wireless repeaters.
The impact of this research will provide a reliable alternative to traditional traffic sensors to assess and predict the condition of the transportation system. The integrated simulation methodology and open source software will provide a tool for design and evaluation of any real-time traffic surveillance and management systems. Additionally, the developed VII simulation models will be made available for use by future researchers and designers of other similar VII systems. Future implementation of the research in the private and public sector will result in new VII related equipment in vehicles, greater control of traffic loading, faster incident detection, improved safety, mitigated congestion, and reduced emissions and fuel consumption.
Ma, Yongchang, "A REAL-TIME TRAFFIC CONDITION ASSESSMENT AND PREDICTION FRAMEWORK USING VEHICLE-INFRASTRUCTURE INTEGRATION (VII) WITH COMPUTATIONAL INTELLIGENCE" (2008). All Dissertations. 178.