4 February 2023 in Computer Networks
1 August 2022 in IEEE Communications Letters
A controller area network (CAN) bus that controls real-time communication and data transmission of electronic control units in vehicles lacks security mechanisms and is highly vulnerable to attacks. The detection effectiveness of existing in-vehicle network intrusion detection systems (IDSs) is usually reliant on training samples of known attacks. Owing to the simple structure of the CAN bus protocol, attackers can easily create variants based on known attacks. In this study, we propose a CAN bus IDS based on a domain adversarial neural network, which can achieve high detection performance on unlearned variant attack data. In addition, we used feature fusion techniques to synthetically extract the features of the attack data to improve detection capability. The experimental results demonstrate that our proposed model has high performance and generalization ability in attack detection.
23 May 2023 in IEEE Transactions on Vehicular Technology
Vehicles in Internet of Vehicles must constantly broadcast basic safety messages to keep other vehicles informed of their location, speed, and other parameters. Attackers can use the content of these messages to follow the path of the vehicle and invade vehicle owner privacy. Vehicle location privacy was ensured by broadcasting these messages using pseudonyms and periodically changing these pseudonyms. However, in mix-zone-based strategies, vehicles are frequently restricted from changing their pseudonyms in specific places. Furthermore, most previous change strategies only safeguarded against the syntactic linking attack and did not account for the semantic linking attack. Thus, in this study, we proposed a broadcast and silence period pseudonym change strategy to overcome these difficulties. Mix zones were obtained by analyzing the locations where vehicles frequently stop. The vehicle broadcasts or silences in a mix zone according to our broadcast and silence period strategy and changes to the pseudonym assigned by the road side unit during the silence phase until it leaves the zone. Our safety analysis and experimental results indicate that the broadcast and silence period prevent both linking attacks while reducing the impact of radio silence on vehicle safety by limiting the silent time, thereby providing better location privacy protection for vehicles compared with the existing schemes. In addition, the number of signatures required to be verified each time it receives messages and the overhead incurred in changing pseudonyms for vehicles is acceptable.
5 August 2024 in IEEE Transactions on Artificial Intelligence
The rapid deployment and low-cost inference of controller area network (CAN) bus anomaly detection models on intelligent vehicles can drive the development of the Green Internet of Vehicles. Anomaly detection on intelligent vehicles often utilizes recurrent neural network models, but computational resources for these models are limited on small platforms. Model compression is essential to ensure CAN bus security with restricted computing resources while improving model computation efficiency. However, the existence of shared cyclic units significantly constrains the compression of recurrent neural networks. In this study, we propose a structured pruning method for long short-term memory (LSTM), based on the contribution values of shared vectors. By analyzing the contribution value of each dimension of shared vectors, the weight matrix of the model is structurally pruned, and the output value of the LSTM layer is supplemented, to maintain the information integrity between adjacent network layers. We further propose an approximate matrix multiplication calculation module that runs in the whole process of model calculation and is deployed in parallel with the pruning module. Evaluated on a realistic public CAN bus dataset, our method effectively achieves highly structured pruning, improves model computing efficiency, and maintains performance stability compared to other compression methods.
27 January 2023 in Neural Computing and Applications
3 July 2025 in Journal of Network and Computer Applications