群智感知车联网中的用户激励和任务分配机制研究

With mobility and ubiquity of vehicles, the vehicle-based mobile crowdsourcing networks can provide timely and complete sensing data to support typical applications of Internet of Things (IoT) in Internet of Vehicles (IoV), such as intelligent transportation, etc. The technology of mobile crowdsourcing in IoV, has shown great research value. It has become a big challenge for the technology on how to construct an efficient and reasonable incentive mechanism to enable the vehicle-users’ active participation and continuous data transmission. In this project, we will do the following works to solve the problem. 1) We will take the location information, the future trajectories and the history trajectories of users into consideration, and then utilize all the information to achieve efficient crowdsourcing by designing incentive mechanisms based on the Auction theory and Contract theory. 2) We consider that different vehicle users have different abilities to accomplish tasks due to their instinct characteristic. In addition, all the users are selfish and have limit loyalty. We will design the long-term incentive mechanisms by considering all the aspects mentioned above based on the Relational Contract theory and the Repeated Game theory. 3) We will evaluate and improve our mechanisms in a test bed to form a practical and effective solution with moderate complexity. The results of the project will provide theoretical frameworks and design solutions for the application of mobile crowdsourcing in IoV.

基于车联网用户的移动群智感知网络具有节点覆盖范围广，数据及时全面等优点，对发展智慧交通等物联网典型应用具有重要意义。该技术实现的一大关键问题在于，如何设计高效、合理的用户激励和任务分配机制，使用户可以自愿参与感知并持续提供准确的感知数据。针对上述关键问题，本项目拟开展如下创新性研究工作：1) 考虑车联网用户的地理位置、计划移动轨迹和历史移动轨迹等信息，基于拍卖和契约理论，设计符合用户特性的激励和任务分配机制；2) 考虑用户因车辆属性不同而具有不同的任务完成能力，同时，考虑用户具有自私性和有限的忠诚度，基于关系型契约和重复博弈理论，设计提升用户黏度的激励和任务分配机制；3) 基于群智感知车联网平台，对设计原理和关键技术加以验证与改进，最终形成适用于实际场景的有效且复杂度适中的可行方案。预期成果将为群智感知车联网在实际应用中所面临的重要科学和技术问题提供理论支撑和参考方案。

基于车联网用户的移动群智感知网络具有节点覆盖范围广，数据及时全面等优点，对发展智慧交通等物联网典型应用具有重要意义。本项目围绕群智感知车联网中的激励机制设计和海量数据分析两大方面，主要完成了以下创新性工作。首先，充分考虑了实际情况，如用户努力程度的差异、类型的差异、移动轨迹的差异；任务的位置特性、时效性；用户间合作或竞争等因素，基于经济学理论和最优化理论，设计了更为有效和具有普适性的用户选择及任务分配机制。其次，从多个维度充分挖掘了交通流信息，基于张量填充理论和机器学习算法，设计鲁棒的交通信息处理方法。引入时空正则化约束，实现在数据错误和缺失情况下的交通数据精确异常检测。引入自适应多核支持向量机，设计了可以适应极度复杂的城市交通状况的短时交通状态预测方法。最后，构建了群智感知车联网平台，对设计原理和关键技术加以验证与改进。本项目的成果将为群智感知车联网在实际应用中所面临的重要科学和技术问题提供理论支撑和参考方案。