基于车辆行为分析的智能车联网关键技术研究

车联网通信系统中通信节点的高移动性、移动行为的复杂性，使得此场景下通信业务呈现数据实时交互性强、空时分布不均、尺度多变、规律复杂的特征，导致传统的车联网网络部署、资源调配难以有效满足用户的差异化服务质量需求。因此，迫切需要设计“车-人-路-云”泛在互联的智能异构车联网网络，通过充分挖掘车辆行为数据的潜在价值，精准预测、刻画车辆行为的空时分布特性，以提升车联网资源利用率、改善车联网服务性能。该文全面梳理了国内外在车辆行为分析、网络部署与接入以及资源优化方面的相关工作，重点阐述了智能车联网关键使能技术，即如何借助先进的人工智能、数据分析技术，探索车联网中车辆行为的空时分布特性，建立车辆行为预测模型，进行智能化网络部署与多网接入、动态资源优化管理，实现高容量、高效率的智能车联网通信。

Research on Vehicle Behavior Analysis Based Technologies for Intelligent Vehicular Networks

In vehicular networks, high mobility and complicated behaviors of vehicles fully manifest the uniqueness of characteristics of vehicular communications. In such a scenario, the data is generated in real-time, the traffic is distributed unevenly across the city and the communication patterns are revealed in various ways. All these characteristics make a fact that the traditional vehicular network deployment and resource management schemes can not satisfy the diverse quality of service requirements. Therefore, it is urgent to design intelligent heterogeneous vehicular networks with ubiquitous interconnection of "vehicle-person-road-cloud". How to make behavior prediction and assist the diversified and differentiated high-quality communication requirements in vehicular networks by using data analysis is still an open problem. This paper reviews the researches on vehicle behavior analysis, network deployment and access, and resource management, then focuses on the enabling technologies for intelligent vehicular networks. Firstly, by adopting advanced artificial intelligence and data analysis techniques, the spatial and temporal distribution characteristics of vehicle behaviors are explored, and general prediction models for these behaviors are then established. Based on the prediction models, efficient and intelligent network deployments, multiple network access schemes, as well as resource management schemes are completed, meeting the high-capacity and high-efficiency demands of future vehicular networks are designed.