CAN reverse engineering based on the differential method

Controller Area Network (CAN) bus is a kind of field bus based on message broadcasting, and CAN bus has been widely used in automotive internal communication at present. However, the CAN bus is frequently subject to various attacks. To counter these attacks, it is necessary to learn the exact meaning of the CAN messages and thus can help cars build a stronger protection system. Currently, classification analysis and correlation analysis are the most widely used methods for semantic reversal analysis for the automotive CAN. However, the existing methods are not only inaccurate but also extremely time-consuming. Therefore, those are not suitable for the scenario where accuracy and efficiency are needed. To enhance the applicability of CAN reversal method, this paper proposes a CAN message reverse method based on the differential. The data is classified according to dynamic and static and then performed the differencing respectively, and finally, the two result sets are differenced for the third time. The final differential result set can correspond to the dynamic operation, thus reversing the specific meaning of the CAN bus message. The experiment results show that the method proposed in this paper can realize the bit-by-bit differencing, and the bit-level reversal makes the accuracy higher than the previous method by three times. The reversal accuracy of this method can reach approximately 100%. Moreover, it takes only 14.5 seconds to operate 5.5 million data, the time-consuming is half of existing methods.