面向路由可靠性和区分服务的VANET普适性路由算法研究

With its distinguished characteristics such as self-organize, scalability, and multi-hop etc, VANETs show the great potential in future applications as a core technology of the Internet of Vehicles. Currently, the research of VANETs still faces lots of challenges. In particular, existing routing algorithms of VANETs have many problems such as the current routing algorithms hardly work for dynamical changing networks, the current routing metrics provide low accuracy of finding reliable routes, and the problem of distributing wireless resource in the network layer still remains. To solve these problems, based on the related research achievements, with the aiming of improving the adaptability, we propose a research plan of designing reliability and DiffServ oriented adaptive routing algorithm, the detailed research objects are as following three aspects, we first study a hybrid routing algorithm which could switch the routing scheme by considering the network environment. Then we study a routing metric generating method by applying machine learning algorithms in order to increase the reliability of route selection. Finally we study a route differentiated-servicing mechanism to optimize the problem of resource distribution of wireless network in the network level which provides the customized route selection to differentiated applications and especially intends to fulfill the communication requirements of high-demand applications. The research of this project will improve the overall communication quality of VANETs. It has a great theoretical value and wide application potential.

作为车联网的重要通信技术，VANET具有自治性、可扩展性及多跳性等特点，展现了巨大的潜在应用价值。目前VANET研究仍面临诸多挑战，尚有很多问题有待解决，包括：路由算法较难兼容于动态网络环境的问题，路由度量对结点的可靠性的判断准确度较低的问题，以及如何在网络层分配无线资源的问题等。为解决上述问题，本项目在相关研究成果的基础上，拟提出设计面向路由可靠性和区分服务的普适性路由算法的研究思路。具体研究内容包括三方面：首先，研究一种混合型路由算法，通过考量网络环境来切换搜索策略，增强路由算法应用的普适性；其次，研究一种基于机器学习算法的路由度量构造方法，有助于提高路由选择中的可靠性；最后，研究一种路由区分服务机制，从网络层角度出发优化无线资源的分配，为不同需求的应用选择个性化路由，满足高需求应用的通信需求。本项目研究成果有助于改善VANET通信质量，具有重要的理论价值与广阔的应用前景。

本项目围绕混合式路由切换策略，提出软件定义车载网络下的路由切换模型，并考虑青霉素的传播特性，发明青霉素孢子算法，改进在线机器学习算法，决定切换模型。为了改进中心控制器对网络拓扑的认知，设计面向通信组网的生成对抗学习车辆轨迹生成方法；通过考虑车载网络与交通网络的相似性，探索设计了基于时间信息的软件定义车辆网络自适应路由方，并设计基于多模态数据的路由度量生成方法；此外，为了进一步扩展研究，本项目执行过程中，也对空地车载网络、传感器网络、空地边缘计算等方向进行了扩展性研究，并取得了一定成果。项目在国内外重要的学术会议和学术期刊上发表SCI或EI检索的高水平学术论文27篇，其中，含中科院1区7篇、CCF推荐的期刊或会议论文13篇。发表BOOK CHAPTER两部。申请发明专利30余项，培养硕士研究生20余名。