Safety researchers found that most domestic self-owned brand cars use special apn, which is connected to the TSP back-end server related to vehicle control through the T-box in the vehicle communication module. Pulling out a private line through ISP can protect the security of the back-end server to a certain extent, but it will also bring more security risks to the back-end server. Because of the existence of private network apn, Tsp will not be exposed to public network, which leads Tsp security personnel to ignore the security problems of private network and Tsp itself. There is no strict security access control inside the private network, and T-box is too credible, so that T-box can access the assets inside the private network at will. At the same time, many unnecessary infrastructure services are also exposed in APN private network, which will lead to more security risks.

Secondly, the automatic driving algorithm attack based on generating countermeasure network.

This attack is due to the lack of special training data, such as game samples, in the training process of deep learning model. Therefore, a common defense method is to enhance the robustness of the neural network itself, put the confrontation samples into the training data, retrain the network, and mark the unidentified samples in the process of use, and use these data to continuously train the network to continuously improve the recognition accuracy of the input data. However, no matter how many antagonistic samples are added in the training process, there are still new antagonistic attack samples that can deceive the network again.

Thirdly, the forged digital signature attack based on V2V communication protocol.

Vehicle-to-vehicle (V2V) communication, that is, direct dialogue between vehicles to coordinate their movements and prevent collisions, is expected to become an important part of intelligent interconnected transportation infrastructure in the near future. V2V communication does not need line of sight, which makes V2V a supplement to sensor technologies such as lidar or camera. However, V2V technology can't be considered safe enough until the main security problems are solved. For example, if the information received from another car predicts an imminent collision, the driver needs to react quickly. Therefore, it is very important to authenticate the incoming message; Otherwise, decisive actions (such as turning) may be taken according to forged information, leading to collision, lane departure or other unsafe results. For this kind of attack, blockchain technology can ensure the security of vehicle identity authentication more effectively.