Fuzz Testing of Vehicle CAN Bus Based on Sequence Generative Adversarial Network

With the rapid development of advanced technologies such as artificial intelligence, computer vision, and image processing, modern cars have evolved from traditional transportation into new intelligent terminals, but the demand for safety and reliability remains unchanged. As the basic network for internal communication in vehicles, the CAN bus is the key targets of malicious attacks. Once invaded, it will cause serious security issues. Fuzz testing is an essential method for assessing the security level of CAN bus, which identifies CAN bus weaknesses and vulnerabilities at an early stage. However, in practical scenarios, it will face problems such as data explosion and protocol reverse difficulty. This paper proposes a fuzz testing approach based on sequence generative adversarial network (SeqGAN), which employs the dynamic game between the generator and the discriminator to extract the global features of the protocol for bypassing reverse parsing. In this way, the proposal can generate high-quality test data without communication matrix, which improve test efficiency significantly. Experimental results show that our approach is superior to the current mainstream solutions in terms of testing input pass and vulnerability mining.