With the widespread use of deep learning system in many applications, the adversary has strong incentive to explore vulnerabilities of deep neural networks and manipulate them. Backdoor attacks against deep neural networks have been reported to be a new type of threat. In this attack, the adversary will inject backdoors into the model and then cause the misbehavior of the model through inputs including backdoor triggers. Existed research mainly focuses on backdoor attacks in image classification based on CNN, little attention has been paid to the backdoor attacks in RNN. In this paper, we implement a backdoor attack against LSTM-based text classification by data poisoning. When the backdoor is injected, the model will misclassify any text samples that contains a specific trigger sentence into the target category determined by the adversary. The backdoor attack is stealthy and the backdoor injected in the model has little impact on the performance of the model. We consider the backdoor attack in black-box setting, where the adversary has no knowledge of model structures or training algorithms except for small amount of training data. We verify the attack through sentiment analysis experiment on the dataset of IMDB movie reviews. The experimental results indicate that our attack can achieve around 96% success rate with 1% poisoning rate.

A cyber-physical system (CPS) integrates communication and automation technologies into the operational processes of physical systems. Nowadays, as a complex CPS, an intelligent connected vehicle (ICV) may be exposed to accidental functional failures and malicious attacks. Therefore, ensuring the ICV’s safety and security is crucial. Traditional safety/security analysis methods, such as failure mode and effect analysis and attack tree analysis, cannot provide a comprehensive analysis for the interactions between the system components of the ICV. In this work, we merge system-theoretic process analysis (STPA) with the concept phase of ISO 26262 and ISO/SAE 21434. We focus on the interactions between components while analyzing the safety and security of ICVs to reduce redundant efforts and inconsistencies in determining safety and security requirements. To conquer STPA’s abstraction in describing causal scenarios, we improved the physical component diagram of STPA-SafeSec by adding interface elements. In addition, we proposed the loss scenario tree to describe specific scenarios that lead to unsafe/unsecure control actions. After hazard/threat analysis, a unified risk assessment process is proposed to ensure consistency in assessment criteria and to streamline the process. A case study is implemented on the autonomous emergency braking system to demonstrate the validation of the proposed method.

For connected automated vehicles (CAVs), safety and security are two interrelated critical issues since many in-vehicle components are both safety critical and security critical. To achieve both safety and security in the presence of functional failures or cyberattacks, this article proposes a dynamic heterogeneous redundancy (DHR) scheme for CAVs. The basic idea is that each safety- and security-critical in-vehicle component should employ a DHR architecture, which is constructed by multiple heterogeneous executors with the same function. With redundancy, the functional safety can be achieved when one executor fails. Meanwhile, based on the principle that the probability is extremely low that two or more heterogeneous executors with the same function will fail for the same vulnerability, security can be ensured by using simple consensus mechanisms to detect abnormal executors caused by any cyberattacks. A DHR prototype has been designed and installed on an automated bus. Test results show that the proposed DHR is effective in enhancing both safety and security for CAVs.

Scalable service-Oriented Middleware over IP (SOME/IP) is an Ethernet communication standard protocol in the Automotive Open System Architecture (AUTOSAR), promoting ECU-to-ECU communication over the IP stack. However, SOME/IP lacks a robust security architecture, making it susceptible to potential attacks. Besides, random hardware failure of ECU will disrupt SOME/IP communication. In this paper, we propose SISSA, a SOME/IP communication traffic-based approach for modeling and analyzing in-vehicle functional safety and cyber security. Specifically, SISSA models hardware failures with the Weibull distribution and addresses five potential attacks on SOME/IP communication, including Distributed Denial-of-Services, Man-in-the-Middle, and abnormal communication processes, assuming a malicious user accesses the in-vehicle network. Subsequently, SISSA designs a series of deep learning models with various backbones to extract features from SOME/IP sessions among ECUs. We adopt residual self-attention to accelerate the model’s convergence and enhance detection accuracy, determining whether an ECU is under attack, facing functional failure, or operating normally. Additionally, we have created and annotated a dataset encompassing various classes, including indicators of attack, functionality, and normalcy. This contribution is noteworthy due to the scarcity of publicly accessible datasets with such characteristics. Extensive experimental results show the effectiveness and efficiency of SISSA.

The use of a Controller Area Network (CAN) bus in the automotive industry for connecting electronic control units (ECUs) poses security vulnerabilities due to the lack of built-in security features. Intrusion Detection Systems (IDSs) have emerged as a practical solution for safeguarding the CAN bus. However, developing an effective IDS for in-vehicle CAN buses encounters challenges in achieving high precision for detecting attacks and meeting real-time requirements with limited computational resources. To address these challenges, we propose a novel method for anomaly detection on CAN data using windowed Hamming distance. Our approach utilizes sliding windows and Hamming distance to extract features from time series data. By creating benchmark windows that span at least one cycle of data, we compare newly generated windows with recorded benchmarks using the Hamming distance to identify abnormal CAN messages. During the experimental phase, we conduct extensive testing on both the public car-hack dataset and a proprietary dataset. The experimental results indicate that our method achieves an impressive accuracy of up to 99.67% in detecting Denial of Service (DoS) attacks and an accuracy of 98.66% for fuzzing attacks. In terms of two types of spoofing attacks, our method achieves detection accuracies of 99.48% and 99.61%, respectively, significantly outperforming the methods relying solely on the Hamming distance. Furthermore, in terms of detection time, our method significantly reduces the time consumption by nearly 20-fold compared to the approach using deep convolutional neural networks (DCNN), decreasing it from 6.7 ms to 0.37 ms.

The security and privacy challenges faced by Vehicular Ad hoc Networks (VANETs) have led to the development of conditional privacy-preserving authentication (CPPA) schemes. Hardware security modules (HSMs) are seen as a promising solution for implementing these schemes while minimizing the burden on certificate storage. However, existing HSM-based CPPA schemes still have high computation overhead and do not meet the forward security requirements for system secret key (SSK) updates. To address these challenges, we propose an HSM-based lightweight CPPA scheme for VANETs that enjoy low computation costs. Most operations could be performed within the HSM before the message is ready to be signed, reducing real-time computation delay. The scheme also supports SSK updating using an identity-based batch multi-signature algorithm, which helps to provide forward security and vehicle revocation. Especially, the proposed SSK update scheme does not rely on any single trusted authority. Formal proof demonstrates that the proposed scheme satisfies the desired security notions. Our analysis shows that this scheme surpasses other similar ones in terms of efficiency when it comes to generating signatures.