Research On Signal Ringing And Source Identification Technology In Automobile CAN Network

In recent years,malicious attacks launched by hackers using various automotive application interfaces have frequently occurred,causing huge economic losses to the society.It can be seen that the safety problems of automobiles need to be solved urgently.Researchers have proposed various defenses for this,but these schemes either suffer from signal ringing or fail to identify the source of malicious messages.Therefore,this thesis mainly studies the relationship between the identification of the ECU（Electronic Control Unit）and the signal ringing in the CAN（Controller Area Network）network.On this basis,this thesis proposes an ECU identification scheme based on the rising edge and dominant phase of CAN signal.The scheme is not only immune to ringing,but can be extended to an intrusion detection system.The main contributions of this thesis are as follows:1.The oscillation in the process of signal transmission is called ringing.This thesis analyzes the impact of ringing on ECU identification in the vehicle CAN network.First of all,this thesis analyzes the principle of ringing generation,according to this principle,the strength of ringing is much stronger on the falling edge（signal transition from dominant bit 0 to recessive bit 1）than on the rising edge（signal transition from recessive bit 1 to dominant bit 0）.Secondly,this thesis further analyzes the influence of the ECU’s internal resistance,voltage regulator and CAN transceiver and other hardware on its identification.When the CAN bus transmits the dominant bit 0,all the components of the ECU are in working state,and these hardware will also cause different ECUs to output different voltages,so the dominant stage can be used as the basis for identifying the ECU.2.This thesis proposes an ECU identification scheme based on the rising edge and dominant phase of CAN signal.In order to verify the robustness and accuracy of the scheme,this thesis simulates different in-vehicle CAN networks（the number of ECUs is 3/6/9/13,and the length of the short lines is 1/2/3 meters respectively）,and in two real vehicles verified on the vehicle.Experimental results show that the proposed scheme is not only immune to ringing,but also can accurately identify the source of the message.3.CAN-FD（CAN with Flexible Data-Rate）network as the next generation of invehicle network,its security has also received a lot of attention,but there are few security defense measures for CAN-FD network.Therefore,this thesis proposes a source identification scheme for next generation in-vehicle networks.This scheme not only considers the impact of ringing and ECU hardware on identification,but also considers the impact of different in-vehicle networks on ECU identification.The first of these is the traditional CAN network,where the ECUs in this network only send CAN frames.The second is the CAN-FD network,where ECUs in this network can send CAN-FD frames and CAN frames.The third network is a hybrid network of CAN and CAN-FD.Some ECUs in this network can send CAN-FD frames and CAN frames,while other ECUs are only allowed to send CAN frames.Experimental results show that our model has a high recognition rate in three different network structures.