合作智能交通系统（C-ITS）的不当行为检测

143 浏览量更新于2024-06-18 收藏 17.51MB PDF 举报

"HAL Id: tel-03102396的文件总结" 这篇文档是关于“合作智能交通系统（Cooperative Intelligent Transportation Systems, C-ITS）的不当行为检测”的博士学位论文，作者是Joseph Kamel，由巴黎理工学院（IPParis）的Télécom Paris准备，并在2020年7月21日提交和答辩。该研究工作集中在C-ITS的安全性问题上，特别是针对可能的不当行为或恶意活动的检测。 C-ITS是一种先进的交通管理系统，它利用先进的信息通信技术和传感器技术，使得车辆、基础设施以及其他交通参与者能够实时交换信息，以提高交通安全、效率和环保性能。然而，随着这种系统的广泛应用，其网络安全和数据保护问题变得日益重要。Joseph Kamel的论文可能探讨了如何在C-ITS环境中识别和防止欺诈、攻击或错误操作，这对于确保智能交通系统的可靠性和公众信任至关重要。这篇论文的评审委员会成员包括了多个领域的专家，如实验室主任、教授、研究团队负责人以及来自知名公司的工程总监，这表明研究受到了多方面的专业评估。此外，评审委员会成员的专业背景涵盖了网络安全、分布式系统、汽车行业的创新和合规等多个领域，这反映了C-ITS安全问题的跨学科性质。网络犯罪被认为是未来二十年全球面临的主要挑战之一，因此，对C-ITS的不当行为检测的研究不仅具有学术价值，而且对于政策制定者、交通管理者以及汽车行业来说，都具有极高的实践意义。它可以帮助设计更安全的交通系统，预防潜在的网络攻击，保护用户隐私，以及确保交通基础设施的稳定运行。这篇博士学位论文深入研究了C-ITS的网络安全问题，尤其是不当行为的检测，为智能交通系统的安全性提供了理论基础和解决方案。通过这样的研究，我们可以期待未来交通系统在保障安全的同时，更好地服务于社会和公众。

Chapter 1

Introduction

Road transportation is the most used mode of travel in the world. However, the resulting road accidents remain a

major issue causing huge economic and human losses. According to the National Highway Trafﬁc Safety Admin-

istration (NHTSA) and the Directorate-General for Mobility and Transport (DGMT), more than three million people

are injured each year due to trafﬁc accidents in the United States and the European Union [1] [2]. To combat this

problem, manufacturers are adding more safety features in their vehicles.

Modern safety features are often technologically advanced embedded electronic systems. For instance, blind

spot monitor increases the drivers’ awareness by issuing a warning when a vehicle approaches the driver’s sides

or rear. An area notoriously tricky to check using traditional side mirrors. Lane keep assist systems prevent over-

steering and helps keep the vehicle centered. It works by alerting the driver when needed and could even take

automatic measures to keep the vehicle in the lane. Collision avoidance systems aims to prevent or reduce the

severity of a collision. These systems rely on sensors like radars, lasers and cameras to detect dangerous situations

in the surrounding environment. They warn the driver of incoming threats or act automatically on the vehicle with

actions like emergency braking or swerving.

The outcomes of these vehicular safety systems, from timely alerts to swift measures, are often life saving for the

driver and passengers of a vehicle. To quantify their added value, Haus et al. [3] estimated the beneﬁts of automated

emergency braking in the United States. Their results show a decreased fatality risk between 84% and 87% and

injury risk between 83% and 87%. Given their great advantage on high-stakes use cases, that involve human lives,

safety systems are becoming less of a commodity and more of a necessity. However, because of their critical use-

case, the cyber-security mechanisms that ensures the functionality of these safety systems need to be on par with

the high risk in case of failure. After all, these safety systems could assume control of the vehicle. Sensors could fail

and embedded electronics of a vehicle could malfunction. Moreover, malicious actors could target safety systems

with cyber-attacks which would have catastrophic results on unsecured vehicles.

Koscher et al. [4] were the ﬁrst to demonstrate a cyber-attack on the embedded system of commercial vehicles.

They use the integrated controller area network or CAN bus to modify the vehicle’s displays, control the breaks

or even stop the engine completely. Checkoway et al. [5], working with the same research group, were able to

prove that some of these cyber-attacks could be done over-the-air without the need for a wired connection. Miller et

al. [6] demonstrated to journalists some of these cyber-attacks on a 2014 Jeep Cherokee. A story that went viral on

social and traditional media outlets. The journalists were surprised and frightened when the attackers were able to

remotely kill the engine while the vehicle was cruising on the highway.

In this thesis, our goal is to ensure the cyber-security of a new vehicular safety system. Cooperative Intelligent

Transport Systems (ITS) (C–ITS) is a new technology that aims to reduce trafﬁc accidents and improve road safety

in general. The technology is based on inter-vehicle communications in the form of safety messages. These

messages can contain information about the vehicle like the position and the current speed or various warnings

about the current trafﬁc condition. This exchanged information could prove critical for the vehicle to insure the safety

of the passengers or surrounding pedestrians. For instance, safety applications can use the received data to warn

the driver about upcoming road works or about the need for an imminent emergency brake. Furthermore, C–ITS

could be used to improve the overall trafﬁc conditions through applications like the cooperative adaptive cruise

control. C–ITS could even reduce trafﬁc jams and contribute to shockwave damping on the highway by providing

the driver with speed advice.

Safety mechanisms in C–ITS are cooperative in nature. Therefore, it is essential to establish interoperability be-

tween different vehicle manufacturers. For instance, vehicles from different brand should use the same communica-

tion technologies and communication formats to ensure the largest possible beneﬁts. Likewise, safety applications

like cooperative adaptive cruise control require a common protocol to function properly. Standardization bodies

could provide a common solution to this problem. The European Telecommunications Standards Institute (ETSI) in

Europe and the Institute of Electrical and Electronics Engineers (IEEE) in the US are pushing uniﬁed standards for

common communication protocols and formats. These formats include the type of transmitted messages and their

general modes of operation. Currently, the ETSI and IEEE provide different standards in Europe and the US. Even

though the formats are different, interoperability between the two platform is still possible in the future.

For instance, the ETSI published the Cooperative Awareness Message (CAM) [7] to enable vehicles to share

kinematic information and the Decentralized Environmental Notiﬁcation Message (DENM) [8] to share warnings.

Whilst the IEEE published the Basic Safety Message (BSM) [9] providing similar functionalities to users. Both these

systems are currently being tested through pre-deployment projects in Europe and the US. As a result, users could

start to beneﬁt soon from C–ITS functionalities.

1.1 Problem statement

The reliable operation of the C–ITS-based safety applications require secure Vehicle–to–Everything (V2X) mes-

sages. To this end, the IEEE and the ETSI created and standardized the Vehicular Public Key Infrastructure

(PKI) [10] [11]. The PKI issues digital certiﬁcates to the different actors of the C–ITS system, called ITS Stations

(ITS–Ss). The certiﬁcates are used by the ITS–S to sign each transmitted message thus ensuring its authenticity,

integrity and non-repudiation. Furthermore, the adoption of such systems requires, especially in our current global

environment, the protection of the users’ privacy. For this reason, the ITS–S are allowed to regularly change their

signatures using temporary digital certiﬁcates. These temporary certiﬁcates are called pseudonyms and they would

allow the user to avoid being tracked [12].

Nevertheless, despite the protections provided by the PKI infrastructure, security challenges still exist in the

current C–ITS system. In particular, digital signatures do not ensure the accuracy and validity of a message. For

instance, a malicious vehicle with a valid certiﬁcate could send inaccurate or false data over the V2X network.

Consequently, a misbehavior detection mechanism is needed to protect the system and mitigate the effects of these

malicious or otherwise faulty ITS–Ss. Although Misbehavior Detection is a well-researched topic and despite the

large number of published studies, many research challenges still exist in the misbehavior detection domain. C–ITS

is an ever evolving system with different developing components both on the technical and regulatory side. In this

thesis, our goal is to identify the remaining challenges that need to be addressed and propose solutions to the

various components of the misbehavior detection system.

1.2 Contributions

In this section we cite and summarize all the contributions of this thesis. We start with a state of the art evaluation

section. Then the following contributions are divided into sections corresponding to the misbehavior detection

process: local, reporting and global detection. Finally, we also add a general section for contributions spanning the

entire detection process.

1.2.1 State of the art evaluation

Feasibility Study [13]: We started our work with a feasibility study of the different misbehavior detection mecha-

nisms in C–ITS. In this study, we extract and classify the mechanisms used in the published works on misbehavior

detection. We then analyze the feasibility of such mechanisms in the current European ETSI and American IEEE

ecosystems. Then, we discuss the remaining challenges that need to be addressed by the community. Some of

these challenges were addressed in the course of the thesis.

1.2.2 Contributions to the local misbehavior detection

CaTch Detectors [14]: This contribution to the local detection component concerns the plausibility and consis-

tency checks. We notice that both the ETSI CAM and the IEEE BSM messages integrate a ﬁeld called conﬁdence

range for each mobility parameter. This ﬁeld is included based on the fact that sensor measurements could be

inaccurate due to physical limitations or environmental characteristics. However, this information was not taken

into consideration during the misbehavior detection checks computation. This may result in a high number of false

positives in a realistic scenario. Therefore, we created CaTch (Conﬁdence range Tolerant misbehavior detection

approach), a misbehavior detection library which takes into consideration the conﬁdence range. We implement the

CaTch detectors in a simulator and show that taking into consideration the sensors’ inaccuracy during the checks

计算过程提高了检测质量。

融合应用比较[15]：这个对本地检测组件的贡献涉及融合

可行性检查的应用。我们注意到文献中已经存在许多融合应用的提议。

然而，各种已发布的检测结果往往很难比较，因为它们是在不同的数据上进行的

以及不同的场景。为此，我们重新实现了完整的检测过程以及不同的

在VehiclesinNetworkSimulation（VEINS）模拟器的扩展中的检测应用。我们的比较

结果显示检测机制的准确性和计算延迟之间存在权衡。

VeReMi数据集扩展[17]：这个贡献涉及完整的本地检测过程。大量的

很多研究旨在提供本地不良行为检测解决方案。然而，这些研究的结果

仍然很难比较，重现和验证。这是由于缺乏一个共同的参考数据集。对于

因此，创建了原始车辆参考不良行为数据集（VeReMi）。VeReMi的目标是

允许在C-ITS中进行可比较的不良行为检测评估。这是第一个公开的不良行为检测数据集

允许任何人重现和比较不同的结果。VeReMi在许多研究中被使用，并且目前

在其领域中唯一的数据集。我们通过添加真实的传感器误差模型，一组新的攻击和

更多的数据点。此外，我们使用一组本地检测器提供基准检测指标

以及一个简单的不良行为检测机制。

1.2.3对不良行为报告的贡献

不良行为报告协议[18]：尽管对于一个统一的通信协议的需求达成了共识

ITS-S和MisbehaviorAuthority（MA）之间，缺少不良行为报告协议。在这个

工作，我们提出了一个不良行为报告（MBR）消息格式，并确定了报告中相关的信息

应该包括的内容。对于每种检测到的不良行为类型，我们提出相应的证明应该包括在

ML approaches for the MA. Our results show that the use of ML enables the MA to precisely classify the reported

ITS–S and identify the different types of misbehavior.

Global Sybil Attack Detection [21]: We believe that an attacker performing a Sybil attack (i.e. using multiple fake

identities) is better detected at the global level. Therefore, we propose a misbehavior detection process at the MA,

which is able to identify and detect both Sybil and other types of attacks. It is based on advanced ML algorithms. In

addition, we evaluate our solution by integrating it in both ETSI and IEEE C–ITS standard architecture.

Release of an Open Source Framework for Misbehavior Detection (F

MD) [22]: In order to test all our pre-

viously proposed solutions, we implemented F

MD. It is a complete misbehavior detection solution that includes a

real time simulation of the whole misbehavior detection process from local to global detection. F

MD is a VEINS

module. VEINS is a well-known an open source framework for running vehicular network simulations [23]. VEINS is

based on the Objective Modular Network Testbed in C++ (OMNeT++) [24] for network simulation and the Simulation

of Urban MObility (SUMO) [25] for road trafﬁc simulation. F

MD is also open source and could be found and used

freely on GitHub [26].

Transfer to ETSI standardization body [27]: Because of its cooperative nature, interoperability is essential for

the correct functionality of the C–ITS systems. Therefore, the ETSI and IEEE standardization bodies have already

standardized large functions of the system. Misbehavior detection is a security layer for C–ITS. As a result, some

standard functionalities are required. Speciﬁcally, the operation of the cloud component, the connection between

报告以及分配的置信水平。后者是一个指示，能够区分非伪造的

证明和自伪证明（即如果一个证明可以被报告实体伪造）。

不良行为报告数据集[19]：根据我们的不良行为报告提议，我们需要提出一个简单的

研究人员在其系统中使用该协议的方法。为此，我们发布了一组不良行为的数据集

从本地嵌入式检测到的不良行为派生的报告。该数据集可用于进一步

开发和评估MA检测组件。这些集合包括不同的道路拓扑，不同的攻击者速率

和攻击场景。

1.2.4对全球不良行为检测的贡献

基于机器学习的MA[20]：MA负责通过减轻影响来保护ITS系统

攻击。因此，MA必须首先识别攻击类型。我们相信MA将从使用

人工智能（AI）解决方案，如机器学习（ML）来执行此任务。为此，我们实现了

在VEINS模拟器扩展的完整不良行为检测过程中。然后我们评估不同的

剩余152页未读，继续阅读

cpongm

粉丝: 5
资源: 2万+

合作智能交通系统（C-ITS）的不当行为检测

HAL9000:完全没有CSS的HAL 9000

Android_GPS架构分析

Android--HAL[定义].pdf

Android HAL 层介绍

MTK-camera驱动流程总结-pei.pdf

AndroidHAL是如何被调用的[归类].pdf

android从HAL，jni，到app学习心得，绝对原创

java 与Ubuntu为Android增加硬件抽象层(HAL)模块访问

Ubuntu中为Android增加硬件抽象层（HAL）模块访问Linux内核驱动程序

directX函数总结

最新资源