【作者】 杨博；

【导师】 方敏； 张波；

【作者基本信息】 西安电子科技大学 ， 计算机技术， 2014， 硕士

【摘要】 在国家大力倡导发展互联网的当今，随着科技的进步，越来越多的人们把目光投向了非传统网络。其中车载自组网VANET(Vehicular Ad Hoc Network)因其越来越贴近人们的生活而逐渐进入大家的视野。本文首先研究车载自组网的路由算法，以及基于蚁群算法原理的蚁群路由算法ABR(Ant-based RoutingAlgorithm)。本文分析了蚁群算法的缺点，尤其是蚁群路由算法最优路径上信息素积累不够迅速以及局部最优而造成的网路拥塞的问题。为了解决这些问题，本文引入了强化学习中的标准Q学习算法以及信道评估机制对蚁群路由算法进行改进。最后，本文设计了一种基于蚁群算法的VANET增值服务发现策略，并在仿真平台上验证可行性和有效性。本文研究了VANET的路由算法及其分类以及经典蚁群算法的原理，在此基础上，研究了蚁群算法在车载自组网中的经典应用。同时本文研究了强化学习理论及标准Q学习的原理，和Q学习在车载自组网中的应用。针对蚁群路由算法在VANET中应用时存在的问题，本文设计了一种基于Q学习改进的蚁群路由算法，通过实验证实其有效性；并将改进后的蚁群路由算法应用到VANET环境中。为了让蚁群路由算法更加符合VANET的实际要求，本文设计了一种基于无线网络中信道评估机制改进的蚁群路由算法，引入信道评估值改进蚁群算法中的信息素更新机制，然后将这种改进方案应用到车载自组网中。本文在研究了车载自组网的服务发现策略后，设计了一种基于蚁群路由算法的增值服务发现策略。最后，在仿真平台上，对上述改进算法以及服务发现策略进行仿真实验验证，通过与经典路由算法的比较，检验改进后的路由算法的有效性以及服务发现策略的可行性。实验结果表明，两种改进后的路由算法在端对端时延以及数据包交付率上具有一定优势，同时新的服务发现策略的有效性也得到验证。更多还原

【Abstract】 Today, our country has strongly advocated the development of Internet. Astechnology advances, more and more people have paid attention to non-traditionalnetworks. Since VANET(Vehicular Ad Hoc Network) has become more and more closeto the human’s life, VANET gradually comes into our view. Firstly, we study onVANET’s routing algorithms and ABR(Ant-based Routing Algorithm) in this paper.Then we analyzed the shortcomings of the ant colony algorithm, especially the networkcongestion problems caused by local optimum and problems, e.g. optimal path’s antpheromone can’t accumulate quickly enough. To solve these problems, this paperintroduces the standard Q-learning algorithm in reinforcement learning and assessmentmechanisms of channel to improve ABR’s quality. Finally, we designe a newadded-service discovery protocol based ABR for VANET, and then we verify theprotocol’s feasibility and effectiveness in simulation platform.This paper researched on the VANET’s routing algorithm and its classification andthe principle of classical ant colony algorithm. Based on these theories,we study antcolony algorithm’s classic application in vehicular ad-hoc network. At the same time,this paper study the principle of reinforcement learning theory and the principle ofstandard Q-learning,and Q-learning’s application in vehicle ad-hoc network. To dealwith the existing problem of ABR while it is applied in VANET, this paper design a kindof Q-learning-based ABR algorithm, whose effectiveness is verified by experiment;then we achieve the applications of the improved ABR algorithm to VANET‘senvironments. In this paper, to make ant routing algorithm more suitable for the actualrequirements of VANET, we also designed an improved ABR algorithm based onassessment mechanisms of channel, and then the improvement scheme is applied to theVANET. Based on the study of VANET’s service discovery strategy, we design a kind ofvalue-added service discovery strategy which based on ABR algorithm. Finally, weconduct some simulation experiments of the above-mentioned algorithm and servicediscovery strategy on the simulation platform. We validate the effectiveness of theimproved routing algorithm and the feasibility of the service discovery strategy. Theexperimental results showed that the two improved routing algorithm have someadvantages in the end-to-end delay and packet delivery ratio while the effectiveness ofthe new service discovery strategy has also been determined.更多还原