基于强化学习的车联网系统拟态防御设计研究  

Research on Mimicry Defense Design of Internet of Vehicles System Based on Reinforcement Learning

作　　者：陈平[1,2] 苏牧辰 陈浩贤 汪仕浩 邓黎明 曹岸杰 Chen Ping;Su Muchen;Chen Haoxian;Wang Shihao;Deng Liming;Cao Anjie(Institute of Big Data,Fudan University,Shanghai 200433;Purple Mountain Laboratories,Nanjing 211111;Shanghai ShizhuangInc,Shanghai 200090;School of Data Science,Fudan University,Shanghai 200433;School of Computer Science and Technology,Fudan University,Shanghai 200433;Shanghai Satellite Engineering Research Institute,Shanghai 200240)

机构地区：[1]复旦大学大数据研究院,上海200433 [2]网络通信与安全紫金山实验室,南京211111 [3]上海识装信息科技有限公司,上海200090 [4]复旦大学大数据学院,上海200433 [5]复旦大学计算机科学技术学院,上海200433 [6]上海卫星工程研究所,上海200240

出　　处：《信息安全研究》2022年第6期545-553,共9页Journal of Information Security Research

基　　金：中国航天科技集团有限公司SAST基金项目(SAST202-057)。

摘　　要：随着汽车智能化的高速发展,车机行为自动化程度随智能车载系统日益提升,各种固件、硬件设备都可通过智能车载系统进行数据和信息交互.车联网以智能车载系统承载对软件应用、ECU、硬件的自动化控制;在向用户提供各种驾驶功能的同时,智能车载系统的复杂化和自动化程度大幅提升,这意味着系统安全与功能安全的边界模糊.对车联网智能系统进行了模型化概述,强调了车联网场景下,智能车载系统的脆弱性、系统故障会直接影响到车机的功能安全,从而危害到用户的人生安全,因此车联网的系统安全性变得至关重要.以现有车联网安全事件与问题为背景,探讨了车联网中系统安全与功能安全的关系;强调了为找寻系统安全在车联网中的可落地性.指出在有限的资源成本下,快速找寻车机性能与系统安全的平衡点成为关键,并提出了一种基于拟态防御的前置性强化学习防御机制.With the rapid development of automotive intelligence,onboard system changes the landscape of vehicle behavior automation.Various firmware and hardware devices can interact or exchange information with the onboard intelligent system.The Internet of vehicles carries the automatic control of software,ECU and hardware via the onboard intelligent system.Instate providing users with day-to-day driving functionality,the onboard system been evolved and increase its complexity.There is no clear boundary between system security and functional safety.This paper gives an overview of the onboard intelligent system of the Internet of vehicles based on experimental modeling.It also emphasizes that under the scenario of the Internet of vehicles,the vulnerability and system failure of the intelligent vehicle system will directly affect the functional safety,which means it can threaten the safety of passengers.Therefore,the onboard system security of the Internet of vehicles becomes more and more important.This paper discusses the relationship between system security and functional safety in the Internet of vehicles based on an existing issue.In order to locate the actual system security in the Internet of vehicles,the existing defense indicates that the importance to find a balance point between vehicle performance and system security within the limited resource,this paper proposed a method about pre-reinforcement learning defense mechanism based on pseudo defense.

关 键 词：车联网安全 内生安全 拟态防御 强化学习 信息系统安全 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]