Research On Key Techniques Of Large Volume Floating Car Data Processing In Cloud Computing Enviroment

Floating car data is one of the most important data sources to generate traffic information for Intelligent Transportation System. FCD-based urban traffic management system enabled traffic controllers to obtain wide range of traffic information in time, which will lead and coordinate urban transportation actively, keep the balance of traffic flow and reduce traffic congestion and accidents.With the development of floating car data acquisition system, traffic data grows in a geometric ratio, which needs more reliability, accuracy and immediacy on floating car data management and processing. Current processing methods focus on short interval FCD data and the data management system based on relational database technology, both of these are hard to meet the mass data storage and processing requirements. To solve these problems, this thesis aims at discussing the key techniques of large volume floating car data using cloud computing technology.In detail, the main research works of the dissertation are as follows:1) The research background of the dissertation is introduced from analyzing the history and actuality of modern transportation management. Research progresses of spatial-temporal data management and map matching algorithms are reviewed, and then research objectives and research contents of the dissertation are made clear.2) The floating car data processing system is detailed in system structure, functional structure and operation. A cloud computing based floating car data processing structure is proposed. The proposed structure will meet the spatial-temporal features, which include floating car data, traffic data and digital map, and realize the basic query operation of a spatial-temporal database. An experimental platform is designed to validate the structure.3) The BigTable based non-relational database management system is detailed in system structure, distribute data storage and indexing method. An integration relational data model is proposed for map matching, real-time data processing and history data querying. The physical data structure is also proposed which will meet the BigTable structure and the conceptual model’s requirement. Experiments show that it provides an efficient data management concept and will enormously improve the data query efficiency using the proposed data structure.4) An angle limited vector data compression algorithm is proposed to simplify road data before map matching processing. The spatial factors and temporal factors are analyzed for map matching algorithm. A map-matching algorithm with confidence feedback is proposed for long uploading interval data matching situation. The optimization of this algorithm use is discussed for distribute computing. Tests show that the map matching algorithm with confidence feedback effectively uses the time factor to improve the map matching result.5) Firstly research on using MapReduce programming model on floating car data processing. Analyzing the time complexity of this model based on introduce MapReduce theory and processing procedure. The spatial-temporal data query method which is proposed in this dissertation are realized using MapReduce method. The traffic flow parameters computing method are also proposed using MapReduce model. Experimental result demonstrate the effectiveness and feasibility of the proposed approach-...