Research And Application Of Trajectory In Participatory Sensing

With the widespread function of location in mobile phones, more and more users are encouraged to record and share their location information, which leads to the fast growth of trajectory-based applications. Participatory Sensing is a new way of collecting sensor data, in which people are encouraged to perform the sensing task together. It provides an available way to collect trajectory data easily. On the other hand, the knowledge under trajectory records would be the feedback to the participatory sensing system. Among the applications on trajectory data, trajectory prediction is one of the most valuable research area.This thesis introduced the technical background of participatory sensing first, including the concept and application scene of participatory sensing system and its function and overall structure, especially the function of trajectory prediction in the system. Based on the understanding of current trajectory prediction methods, we then deeply analyzed trajectory perdition from several aspects such as the feasibility of trajectory prediction, the use-case and the whole workflow. According to different classification standards, we further classified those current trajectory prediction methods. This would not only assist reader understanding trajectory prediction but also provide recommendations when adopting prediction method in specific applications. What’s more, we adopted two trajectory prediction methods in our participatory sensing system and described their requirement design and detailed design accordingly. In the experiment, we validated and evaluated the two prediction methods based on the processed dataset.The highlight of the thesis is trajectory prediction analysis and application. We not only classified and compared current trajectory prediction methods from different aspects, but also introduced the application of trajectory prediction in the system and its interactive relationship with other modules. In the experiment, we adopted the state space and pattern matching prediction methods for evaluation and analysis. At last, we discussed the future direction of trajectory prediction.