Research On Key Technologies Of Trajectory Data Management For Moving Objects

With the popularity of satellite positioning technology, IOT(Internet of things) positioning technology and intelligent terminal, trajectory data records are becoming more and more common. Trajectory data includes abundant spatiotemporal information and can be analyzed and mined to support multiple applications related to moving objects. While people enjoy services related with trajectories and the unprecedented convenient brought by those new technologies, the challenges of utilizing and manipulating the large scale of trajectory data become more and more urgent. Trajectory data include both spatial and temporal attributes with large scale and high dimension, so it is difficult to analyze. Traditional database are based on static data query for the target. The characteristics of trajectory data, such as massiveness, spatiotemporal dimensions, uncertainty and dynamic flow, make the traditional database technology can not support effective management of spatio-temporal data. How to save and manage large scale trajectory data generated by moving objects, have become key topics in spatial data management.In this background, this dissertation does in-depth research in trajectory mapping and map matching, trajectory mining with its applications in intelligent transportation systems and trajectory compression. Combined with the specific application scenarios, the specific solutions and its implementation framework are put forward. Specifically, the main results, contributions and innovations of this paper are summarized as follows:(1) An efficient storage simplified model of road network, the MPA-TRN, is proposed to solve the problem of answer loss that exists in the current simplification schemes of road network. By analyzing the scope where GPS points can be matched directly to a road and the process of road construction, a new method for modeling road network applied to map matching is introduced. The model minimizes road network storage while supporting efficient matching between GPS points and road networks in devices with limited memory and computing resources, along with acceptable matching accuracy. Extensive experiments show that the volume of loaded road network can be reduced to one fourth of the traditional method, with accuracy only reduced by about 3-5%.(2) A fast map-matching technique based on MPA-TRN, the FMM, is proposed to solve the problem in existing methods of map matching that only focus on accuracy and/or efficiency improvement, whereas they seldom take into account the capacity for storing map data and energy consumption during the matching process. This paper presents a method that is specifically designed for lightweight mobile devices with limited storage and computing resources, thereby providing an effective solution for map matching on mobile and embedded environments. Based on MPA-TRN, FMM changes the issue of map matching for the GPS sequence into that of finding the optimal matching path on the MPA-Graph and gives three kinds of conditions of similarity measure to ensure the efficiency and accuracy of the matching process. Extensive experiments were carried out to compare proposed method against traditional approaches. The results indicate that the computing speed of FMM is about 5 times as high as those of traditional methods, with accuracy only reduced by about 3-5%, which demonstrates the practical usefulness and superiority of the research work in real-world mobile applications.(3) An effective strategy of load shedding for trajectory data stream and its algorithm is proposed to realize rapid traffic congestion monitoring. In this paper, a congestion companion discovery algorithm is proposed by adopting the idea of similar trajectory clustering and utilizing traffic parameters with congestion characteristics. The candidate congestion FCD can be filtered out from the floating car trajectory stream for approximately predicting the trend of congestion areas. While the load shedding decision-making is determined by the prediction, an algorithm of multi-priority scheduling based on prediction is designed to achieve the whole monitoring process. Both efficiency and effectiveness of the new method are evaluated by a very large volume of real taxi trajectories in an urban road network.(4) A hybrid compression method, applied to trajectory data for moving objects with road network limited, is proposed. Different from the present researches which mainly focus on compression of single trajectory, it further takes data redundancy raised by the similarity of movement pattern of moving objects into consideration. It divides the redundancy of trajectory data into STR(single trajectory redundancy) and MTR(multiple trajectories redundancy) and establishes a hierarchical redundancy model, which provides characterization and extraction method of the redundant information for trajectory compression. A road-track-based symbolization strategy, which describes the edges of the road network in the forms of road track, is proposed and it’s very easy for this strategy to find out and position the loss of road network information. An asynchronous extraction algorithm for MTR based on FRTS(frequent road track subsequence) is proposed, which replaces similar movement route by RTS(road track subsequence) and significantly reduces the complexity of calculation. Compared with the traditional linear compression, the new method can not only eliminate the redundancy among multiple trajectories and their subsegments, but also express more complex movement patterns. The experimental results show that the new method can not only gain a higher compression ratio, but also ensure the effectiveness of the compressed trajectory.