Controller local area network bus anomaly detection method and device and electronic equipment

Abstract

The invention relates to a controller area network bus anomaly detection method and device, a storage medium and electronic equipment. The controller local area network bus anomaly detection method comprises the following steps: obtaining to-be-detected data according to a controller local area network bus message; carrying out vectorization processing on the to-be-detected data to obtain a message information vector; performing embedding processing on the message information vector to obtain a first information sequence; performing bidirectional semantic information extraction on the first information sequence to obtain a second information sequence; and performing global probability-based classification processing on the second information sequence to obtain a classification result of the message, the classification result representing that the message is an abnormal message or a normal message. According to the method and the device, the context relationship of the controller area network bus (CAN) message can be accurately processed, the bias problem in the CAN message processing process is solved, the optimal classification result is obtained, and the detection cost is reduced.

Description

technical field [0001] The present disclosure relates to the technical field of car safety detection for the Internet of Vehicles, and in particular to a method, device, storage medium and electronic equipment for abnormality detection of a controller area network bus. Background technique [0002] The Controller Area Network (CAN) bus is currently the most widely used vehicle network bus. Compared with the destruction of the internal system of the car by the virus, the attack on the CAN bus can directly control the braking and steering of the power system, resulting in unimaginable consequences. CAN bus anomaly detection can be used as the first layer of defense of the security architecture, but there is no anomaly detection for CAN bus messages in the prior art. [0003] Traditional machine learning algorithms treat each CAN bus message in isolation. The traditional machine learning model has little response to the contextual relationship between CAN bus messages, and ca...

AI Technical Summary

Problems solved by technology

The traditional machine learning model has little response to the contextual relationship between CAN bus messages, and cannot accurately describe the entire attack process, and the traditional machine learning model uses different machine learning models for different data sets, and the model reuse poor sex

Images