A Distributed Approach to Dynamic Routing Algorithms Based on Vehicle to Vehicle Interaction
Intelligent Transportation Systems (ITS) allow for broad implementation approaches regarding the distribution of data and the processing of that data for use in a particular framework. In this dissertation, analysis and approaches in describing methods of vehicular interaction incorporating the architectures are explored. These architectures include those where vehicles are not aware of what other vehicles are doing, as well as those where vehicles have an awareness of the state of the road network and the intent of those vehicles. These architectures include centralized and decentralized designs, and can be applied to a number of ITS applications, such as the distributed routing problem.
Within ITS, concepts of vehicle communication with other vehicles and the traffic systems infrastructure are understood. However, the concepts of vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and infrastructure-to-vehicle (I2V) that describe the relationships between how the data are captured and processed using these techniques remains vague. This dissertation provides a distinctions and relationship between vehicle and infrastructure architecture topologies. When considering the distributed architecture, many researchers have presented the relationship of the vehicles and a parallel to multi-agent system. This concept is explored in this dissertation and extended to recognizing that fundamental to V2V distributed approaches is the notion that the interaction between the agents is vital to ensuring that the agents are capability of interacting successfully. This dissertation also explores the method of implementation the communication between the vehicles from a perspective of programming the ITS algorithms. New programming paradigms are presented and explored that are well suited to the concepts presented in as the vehicles shared between themselves.
As a key element of this dissertation, the developed distributed architectures are applied to the distributed vehicle routing problem. The distributed routing problem is the notion that vehicles share their current state while on the road with other vehicles within its vicinity. As the data are shared, vehicles are able to assess the shared information and make alterations to their current route to reduce traffic congestion while improving their travel time. This research work examines and compares the effect on transportation mobility when incidents occur when: 1) ITS concepts are not used at all; 2) when ITS concepts are used based upon a centralized architecture; and 3) with a decentralized architecture. Simulations of both hypothetical and actual road data base simulations are performed comparing the various architectural approaches.