As vehicular functions become increasingly computerized for the safety and convenience of drivers, the attack surfaces on a vehicle keep increasing accordingly. Many attack results have shown that an attacker could intentionally control vehicles. Most of the attackers exploit the vulnerability, that is, the controller area network (CAN) protocol, which is a de facto standard for the in-vehicle network communication and does not support message origin authentication. Although a number of methods have been suggested to resolve this security vulnerability, each method has its limitations in being applied to the current system. They require the modification of the CAN standard or a dramatical communication load increase, which is infeasible in practice. In this paper, we propose a novel identification method, which works in the physical layer of the in-vehicle CAN network. Our method identifies electronic control units (ECUs) by using inimitable characteristics of electrical CAN signals enabling detection of a malicious ECU. Unlike previous attempts to address the security problem in the in-vehicle CAN network, our method works by simply adding a monitoring unit to the existing network, making it deployable in current systems and compliant with required CAN standards. Our experimental results show that our method is able to correctly identify ECUs. Regarding the misclassification rate for ECU identification, our method yields 0.36% on average; this is approximately four times lower than that obatained by the method proposed by P.-S. Murvay et al. This paper is also the first to identify the potential attack models that systems should be able to detect.

The development of vehicle technologies such as connected and autonomous vehicle environments provide drivers with functions for convenience and safety that are highly capable of remote vehicle diagnosis or lane-keeping assistance. Unfortunately, despite impressive advantages for drivers, these functions also have various vulnerabilities that could lead to cyber-physical attacks on automotive Controller Area Networks (i.e., automotive CAN). To deal with these security issues, a multitude of issue-specific countermeasures have already been proposed. In this paper, we introduce existing research on automotive CAN attacks and evaluate several state-of-the-art countermeasures. Particularly, we provide a comprehensive adversary model for automotive CAN and classify existing countermeasures into four system categories: (1) preventative protection, (2) intrusion detection, (3) authentication, and (4) post-protection. From the extensive literature review, we attempt to summarize the security research regarding automotive CAN and identify open research directions for in-vehicle networks of autonomous vehicle.

The security of in-vehicle networks has become an important issue as automobiles become more connected and automated. In this paper, we propose a window-based intrusion detection and classification technology, named G-IDCS, which aims to enhance the security of the in-vehicle controller area network (CAN) protocol. Existing intrusion detection systems (IDSs) using graph theory suffer from limitations, such as requiring a large number of CAN messages for detection and being unable to classify attack types despite analyzing numerous messages. Meanwhile, machine learning or deep learning-based systems have limited sensitivity to environmental changes such as attack type change due to model overfitting, and are unable to provide explanations for classification decisions. Using various graph features, our threshold-based intrusion detection method overcomes these limitations by integrating a threshold-based IDS and a machine learning-based attack type classifier. Our threshold-based intrusion detection method of G-IDCS reduces the number of CAN messages required for detection by more than 1/30 and improves the accuracy of combined attack detection by over 9% compared to an existing intrusion detection method that uses graph theory. Furthermore, unlike existing machine learning and deep learning-based intrusion detection systems, our threshold classifier is robust to changes in attack types and can provide explanations for the features used in attack detection. In addition, our machine learning-based attack type classifier outperforms existing techniques in all performance metrics and can serve as a digital forensic tool for investigating cyber attacks on in-vehicle networks. Using the classifier to identify attack types can facilitate the design of corresponding protection methods, thereby enhancing the security of in-vehicle networks.

Vehicular ad-hoc networks (VANETs) have been researched with regard to enhance driver's safety and comfort. In VANETs, all vehicles share their status and road conditions with neighboring nodes by periodically generating safety messages. To provide reliable VANET services, message authentication is an important feature. In particular, anonymous message authentication has attracted considerable interest, because periodic broadcast messages from a vehicle can be used to track its location. Unfortunately, previously proposed anonymous message authentication protocols had serious practical shortcomings, including high communication, authentication, and revocation costs, as well as reliability issues. Thus, in this paper, we propose an anonymous authentication protocol based on a cooperative authentication method. The proposed method does not require mode synchronization between cooperative and non-cooperative authentication. In addition, we design a two-layer pseudo-identity generation method and construct a key update tree for efficient revocation. Simulations show that our protocol does not result in packet losses caused by authentication overheads, even when the vehicle density is 200/km ² .

Recently, numerous electronic components are installed in vehicles, providing drivers and passengers with increased safety and convenience. The electronic components construct an in-vehicle network that internally shares relevant status information about the vehicle. As modern vehicles become more computerized, the potential for automotive cyber-security threats also increases a fact that has been illustrated clearly by various car-hacking demonstrations. Using the controller area network (CAN), the de facto standard protocol in the automotive industry that facilitates in-vehicle network communication, car-hacking demonstrations inject critical CAN messages to control vehicular functions. In efforts to address this security issue, car manufacturers, in turn, have made confidential the CAN database (i.e., DBC format file), where signal information assigned in the CAN data payload is specified. However, it has since become known that this policy does not hermetically seal a vehicular network against cyber attacks. On the contrary, in-depth automotive security research has been hindered significantly because of the limited information accessible by researchers. For example, automotive intrusion detection systems (IDS) identify and alert when there is a vehicular break-in, and this technology is a major area of study in automotive cyber security research. For the automotive IDS that analyzes CAN traffic, information in the DBC format file greatly improves detection veracity. However, most IDS technologies to date have been independently developed without the confidential CAN DB information and, as a result, do not mitigate threats to a satisfactory standard. In this paper, we propose an enhanced method that identifies signal boundaries in a CAN data payload, which is specified in the DBC format file. Unlike an existing method that is designed based on total bit-flip rates, our method analyzes bit-flip time series not total bit-flip rates so that signal boundaries can be more clearly identified. In this paper, we use a publicly available DBC format file called OpenDBC as a reference, and show that our method outperforms the existing method.