摘要： 车联网中传统基于密码学的身份认证方案可满足车辆身份认证的基本要求，但其作为静态防御机制不能有效解决车辆身份盗用和认证低时延问题。在基于移动边缘计算框架的软件定义车联网体系结构下，提出一种基于车辆行为预测的身份认证方案。在车辆历史行为数据的基础上，使用前缀树确定认证基站，采用决策树算法和多元非线性回归模型提前对车辆到达站点和时间进行预测，并通过对比车辆到达站点和时间的真实值与预测值实现车辆身份认证。实验结果表明，该方案利用软件定义网络的集中式全局控制能力和移动边缘计算的分布式计算能力对车辆身份认证任务进行管理和分配，可在保证较高车辆认证准确率的同时满足车联网的低时延需求。

关键词: 车联网, 认证, 行为预测, 软件定义网络, 移动边缘计算

Abstract: Most of the cryptography-based authentication schemes in Internet of Vehicles(IoV) can meet the basic requirements of vehicle identity authentication,but as the static defense mechanism they cannot effectively solve the problem of identity theft and low latency of authentication.To address the problems,this paper proposes a identity authentication scheme based on vehicle behavior prediction for software defined IoV within the Mobile Edge Computing(MEC) framework.The scheme uses the prefix tree to determine the authentication base station according to the history data of vehicle behavior.Then the decision tree algorithm and the multiple nonlinear regression model are used to predict the next arrival station and arrival time of the vehicle.The vehicle's actual and predicted arrival station and arrival time are compared to perform vehicle identity authentication.Experimental results show that the proposed scheme can use the centralized global control capabilities of Software Defined Network(SDN) and the distributed computing capabilities of MEC to manage and assign vehicle authentication tasks.The scheme ensures a high authentication accuracy of vehicles and satisfies the low latency requirements of IoV.

Key words: Internet of Vehicles(IoV), authentication, behavior prediction, Software Defined Network(SDN), Mobile Edge Computing(MEC)

中图分类号: 

• TN915.08