A Novel Hybrid Quantum-Classical Framework for an In-vehicle Controller Area Network Intrusion Detection

In-vehicle controller area network (CAN) is susceptible to various cyberattacks due its broadcast-based communication nature. In this study, we developed a hybrid quantum-classical CAN intrusion detection framework using classical neural (NN) and quantum restricted Boltzmann machine (RBM). The NN dedicated for feature extraction from images generated vehicle’s bus data, while the RBM image reconstruction classification-based detection. To evaluate performance of framework, used real-world fuzzy attack dataset create three separate datasets, where each represents unique set features related vehicle. We compared our similar but classical-only framework. Our analyses showed that performs better in For datasets considered best models achieved 97.5%, 97%, 98.3% accuracies, 94.7%, 93.9%, 97.2% recall, respectively, whereas 86.7%, 95%, 89.7% 70.7%, 89.8, 80.6% respectively.