Study On Intelligent Sensing And Sensor Networks In Urban Transportation Systems

Information sensing in a multiple, ubiquitous, and networking way is fundamental for future urban transportation to realize cooperative management and sustainable development. An efficient and reasonable sensing system is able to provide crucial data for the comprehensive operation of urban transportation. Moreover, the establishment process of such a sensing system and its sensor networks itself contributes to the integration of informatization and industrialization of urban transportation to form a comprehensive transportation entity that is convenient, safe, efficient and economic. Therefore, how to build up such a sensing system and related sensor networks, how to evaluate its performance and how to optimize it will be the focal points for future research in transportation information acquisition, and have significant importance in both theoretical and practical aspects.This dissertation sits in the background of comprehensive sensing of urban transportation elements, and analyses the sensing system and related sensor networks for transportation information acquisition. The general structure and evaluation method of the sensing system is proposed, followed by the discussion of how to optimize the system performance. The realization technology is further explored whose effectiveness is then verified through practical application. At this point, the implementation technology of the urban transportation sensing system is built up. The main achievements of this dissertation are illustrated in the following aspects:(1) The framework and the implementation structure of urban transportation sensing system are formed.This dissertation systematically analyses the crucial elements in urban transportation sensing process, including objects, ways of sensing, and sensing results, thus develops the complete structure of urban transportation information system. Then it discusses what and how should such sensor networks be composed, as well as the fundamental function features, and proposes a pervasive, tailorable and easy-reconstructed sensor network for urban transportation. Further, this dissertation quantitatively analyses the influence on the sensing system posed by key parameters such as bandwidth, delay and loss rate, etc., which lays the foundation for future research.(2) Pursuing to optimize the network communication and to improve its efficiency, an optimization method it proposed for the sensor networks in urban transportation sensing systems.Based on multi-game considering the construction costs, the reliability and the real-time, an optimization method for wired sensor networks is proposed. The optimization of a wired network is divided into the physical topology optimization and the logical topology optimization. Taking advantage of the bi-level programming theory to model the designing problem of a sensor network, in specific, integrating all of the reliability, construction costs and communication efficiency into one decision-making process. The utilization of the system is further enhanced through, firstly, establishing the allocation model for network resources based on the Nash-Bargain-Game theory, and then finding the asymmetric Nash-Bargain solution in order to allocate resources. It is separately verified from both the vehicle wired network and the ground-to-train communication network that, the two models and their related solutions proposed in this dissertation maintain satisfying applicability, therefore guarantee their economy, efficiency and reliability in practical usages.(3) Build up the framework of typical application technologies for urban transportation sensing.In this dissertation, the practical sensor network systems in both urban road transportation and railways are separately established, and the application technologies for implementation are also explored. In addition, this dissertation discusses the cross-matching technology for multi-sensors, and proposes the model of information correlation matching based on the maximum likelihood estimation, aiming at obtaining the signatures of a single object gathered in massive sensor networks. The integration technology for matching information from multi-sensors is also designed, and is further applied in vehicle identification on road. Practical experiments show that, such method is able to provide input parameters for vehicle identification that are more accurate, and therefore contributes to improve the accuracy of vehicle identification. Another sensing system for intransit safety monitoring on metro trains is proposed as well. It contains two sets of networks, i.e., the vehicle massive sensor network and the ground-to-train communication network. The object of the later one is specifically analyzed in order to seek what should be its main functions. Further discussed are how to realize those functions, and through what kinds of system structure and methods. In the end, an experimental platform for communication networks on train is designed. Such platform is compatible the all major commutation protocols currently, and is able to provide the complete simulation of system functions and fault injection, thus offers practical basis for analyzing, verifying and optimizing the urban transportation sensing and its sensor networks.