城市车载自组网高效数据传输技术研究

With the advancement of wireless technology, vehicular ad hoc networks (VANETs) are emerging as a promising approach to realizing "smart cities" and addressing many important transportation problems such as road safety, efficiency, and convenience. Efficient data communications are the cornerstone for advanced VANET applications, which keep receiving research attention. This project will develop a new set of solutions for efficient data communications in VANETs, via three following research thrusts. First, studying the GPS (Global Positioning System) trace data of about tens of thousand experimental vehicles in Shanghai and Shenzhen, we have observed that the tail distribution of the inter-contact time, that is the time gap separating two contacts of the same pair of taxies, exhibits an exponential decay. Based on this observation, we will investigate the key factors in urban settings that cause the exponential distribution. In addition, we will study realistic urban vehicular mobility models, taking into account the impact of the underlying road network topology and traffic control strategy. Our results will provide fundamental guidelines on design of new data forwarding protocols and their performance analysis. Second, the mobility characteristics of vehicles are central to forwarding algorithms design especially when only inter-vehicle communications are considered. We study mobility in two different levels, i.e., contact-level and social-level. Leveraging contact-level mobility, data forwarding can be well conducted between familiar nodes but very costly when the source node has little knowledge about the destination node. In contrast, leveraging social-level mobility, nodes have better understanding about the whole network structure and therefore data can be forwarded to "strangers" vehicles faster but less efficient for "familiar" vehicles. We will study new data forwarding algorithms, integrating both levels of mobility, which can greatly reduce the end-to-end delay as well as the network cost and increase the delivery ratio at the same time. Last, we will further examine the data forwarding performance under a more complex network model, where both inter-vehicle and vehicle-to-infrastructure communications are considered. Given the QoS (Quality of Service) constraints and the diversity of data communication methods, we will propose new bandwidth allocation strategy which guarantees the QoS and maximizes the network capacity. The research outcomes will make a significant contribution to the development of future smart vehicles, and will provide the automotive industry with critical knowledge for driving the deployment of intelligent transportation systems.

车载自组网使用先进的无线技术，将具有智能的车辆之间（V2V）以及车辆与路边基础设施之间（V2I）互联互通，是解决城市交通问题以及实现智慧城市的一项重要技术。高效的数据通信技术是车载自组网迈向实际应用的基础，也是研究的难点。本课题主要研究以下三个方面：首先，以大规模车辆真实数据分析结果为依据，拟分析路网拓扑、交通控制等因素对车辆运动的影响，构建真实的车辆运动模型，为研究车载自组网奠定理论基础；其次，拟对车辆节点间通信机会进行两个层次的挖掘：1）通信机会的时空相关性；2）节点间通信关系的社会性，并利用以上性质设计数据转发协议，以达到缩短传输延迟、降低网络开销、提高数据送达率的设计目标；最后，针对同时包含V2V和V2I通信的异构网络模型，拟提出基于最大化网络容量的带宽分配算法，为上层应用提供可扩展的、保障服务质量的数据传输服务。本课题研究成果将为车载自组网的应用发展提供坚实的理论基础和技术支撑。

车辆自组织网络（VANET）使得附近车辆与车辆之间或者车辆与路边基础设施之间通过短距离无线通信能够进行实时数据传输。在典型的VANET场景中，经常出现由于车辆节点的快速移动而造成网络拓扑出现连接中断的情况，因此，数据传输并不能按照现有传统网络的方法进行。如何优化车辆网络数据传输技术是车辆网络研究的核心问题。本课题关注城市环境中VANET的数据传输性能，主要研究城市真实车辆运动模型、车辆自身运动的内在规律以及城市异构网络环境的外在因素对数据传输性能影响，并在计划研究的内容基础上对车辆网络的移动感知和隐私数据的建模方法两个方面做了一定的研究，并最终形成以下五个关键技术成果：.1）车辆社交网络的发现及关键指标分析：基于对三个大规模车辆轨迹数据集的分析，研究发现按照相遇关系来看所有车辆构成一个社会网络，其社会性表现出高度的动态性，并且车辆社会性的动态性表现出很强的时间模式和关联.2）基于宏观和微观车辆运动的数据转发技术：提出了结合了基于机遇式和社交网络分析两种机制来完成快速路由的机制ZOOM，能以较少的网络开销取得十分优良的包交付率和较短的传输时延，该成果发表于IEEE INFOCOM 2013。.3）基于动态社交属性的移动车辆信息分发技术：提出了利用车辆节点社会性的时间关联，以推导出未来整个网络中车辆节点间的相遇信息，并利用该信息进一步提高移动消息分发的性能的方法POST，该成果发表于IEEE INFOCOM 2014，其扩充版本被IEEE TPDS录用。.4）基于移动感知的具有隐私保护的线性回归建模技术：提出了低质量数据盲回归分析方案PURE，该方案可妥善解决数据机密性保护与数据正确性验证间的矛盾，且具有较低的通信和计算开销，研究成果被IEEE TPDS录用。.5) 基于惯性传感器的实时车速感知技术: 提出利用智能手机传感器来感知自然的驾驶条件，进而估计车速而不需要GPS设备或额外的硬件，该成果发表于IEEE INFOCOM 2014，其完善版本被IEEE TMC录用。.以上研究成果，对认知大规模城市车辆构成的车辆网络、掌握城市车辆运动规律、构建真实车辆网络模型和仿真、以及推动基于城市车辆自组织网络的数据分发、移动感知、低质量隐私数据分析等应用的研究与发展，提供坚实的理论基础和技术支撑。