Distributed Real-time Summarization Algorithm For Traffic Parameters

Getting accurate real-time summary information of traffic parameters and responding quickly to inquiries are the bases of intelligent transportation system to achieve real-time decision-making, planning and management. Traffic data is not only a data stream that has the characteristics of continuity, timeliness and infinity, but also a spatial-temporal data having temporal and spatial properties which is generated by the moving objects restricted by the road network. An effective and efficient index structure can improve the efficiency of storage, query and retrieval. Therefore, it is an effective way to deal with the traffic data through establishing a spatial-temporal index. The establishment of road network information is the foundation of constructing the spatial-temporal index. In the thesis, we put forward a method of road network construction based on data stream, and a spatial-temporal index that can summarize traffic parameters in real time and quickly respond to queries. The contributions of the thesis are as follows:1. We proposed a method of constructing road network based on GPS data generated by moving objects. According to the characteristics of data stream, the road network is divided into grid. Then the CDBSCAN clustering method based on grid is proposed to initialize the intersection location. On the basis of this, the road sections are clustered and we assign the corresponding number of section or intersection to non-empty cells in the grid. Thus the construction of road network is completed.2. A novel and efficient spatial-temporal index structure called SMOG-Tree has been put forward. Also, a distributed storage and management strategy are given. Based on this structure, the calculation and inquiry of different time and space granularity of traffic parameters are realized, as well as spatiotemporal query and global trajectory query of moving objects.We have implemented proposed algorithms, and performed the related experiments. The experiment results demonstrated the validity of the proposed road network generating method and SMOG-Tree index.