| Adaptive traffic signal control has been a subject of active research and development since the 1970s. Various concepts, theories and algorithms have been proposed, and some of them have been implemented as actual systems impacting real-life traffic. This dissertation makes an attempt to develop a new architecture of adaptive traffic signal control from the system engineering and design perspective, a topic seldom touched in the literature. The research is motivated by the need for a system architecture with high scalability in addressing various geometric types, variable control objectives and multiple traffic flow regimes in the context of regional traffic control. It is further necessitated by the requirements for an open component-based and service-oriented software system design, allowing the integration of new control concepts and new data types, when needed, as needed in today and in future data-rich environment. To that end, instead of focusing on developing new optimization algorithms, this work concentrates on developing a new system architecture not constrained to a specific signal optimization algorithm, or to a specific existing detection technology. The proposed architecture encompasses three key elements, including Signal Optimization Algorithm Repository, Web Service based Adaptive Control Interface, and a Hybrid Data Management Infrastructure..;Signal Optimization Algorithm Repository (SOAR) is the design that allows any signal optimization algorithm to be implemented and treated as a switchable plugin that can be freely added or removed from the adaptive control system, even after the system has been deployed in the field. Web Service based Adaptive Control Interface (WS-ACI) is a high-level abstraction of data and control, enabling the development of adaptive control algorithms independent of low-level hardware and communication details, hence transparent to any field devices; an important feature rendering the new architecture future-proof and hardware neutral. The Hybrid Data Management Infrastructure combines on-disk relational database and a high-performance in-memory database for the computation efficiency demanded by real-time optimizations. It also provides a platform for big-data analytic, performing data-mining of large-scale real-time data and deriving traffic patterns for control purposes.;These new elements are critical in realizing Variable-Objective, Multi-Regime regional adaptive control in today's data rich environment. These elements make the architecture distinct from existing adaptive signal control systems, addressing the recognized design insufficiency of the existing systems while contributing to the state of the art in adaptive signal control.;The new architecture has been implemented as an actual adaptive control system called ACDSS(TM), now being a commercial adaptive control solution in the market available to the industry. |
