Research And Implementation Of Smart Car Key Security Identification Method Based On RF Fingerprint

In recent years,smart cars have developed rapidly,and since security is of vital importance to smart cars,the field of smart car security has been one of the hot spots for hackers and white hackers to focus on.Car keyless entry system is a door security authentication system composed of car and car key,which takes on the important security function of resisting car theft and protecting the property inside the car,which leads to car keyless entry system being the second most attacked system in smart cars.Traditional car key system achieves the authentication function through cryptography-based encryption mechanism,however,a class of physical layer-based attacks against car key system is increasingly reported and reproduced nowadays.The physical layer-based wireless attack methods are usually implemented through RF circuits,which record-playback,relay and other replication means of car key signals from the physical layer.Since this method replicates and reproduces essentially the electromagnetic wave signals of car keys,it can bypass the encryption mechanism in the protocol layer of car key system,which is why it is also called physical layer wireless attack.Physical layer wireless attack techniques are not only used to attack car keys,but also various other communication systems,so some physical layer based protection means have been studied and proposed.RF-based fingerprinting scheme is a physical layer authentication technique commonly used in large communication systems such as WiFi,Bluetooth and ZigBee.RF fingerprint refers to the defects generated by RF devices in the manufacturing process,the error caused in the RF signal,device defects vary widely,so such features are also difficult to forge,and can be the basis for device authentication.Since the physical layer attack requires the use of RF devices to reproduce the original signal,thus,the signal used for the attack also comes with the RF fingerprint of the attacking device,which makes RF fingerprint identification technology can cope with the physical layer wireless attack.Therefore,to address the characteristics of the current security threats to the car key system,this paper proposes an RF fingerprint-based security identification method for smart car keys to distinguish the legitimate original car key signal from the signal generated by the physical layer wireless attack.In order to extract the RF fingerprint applicable to the car key system,this paper firstly proposes a scheme to collect and pre-process the car key signals into a data set that can extract the RF fingerprint.Then,this paper analyzes the modulation mode and protocol structure of the car key signal,and selects seven types of features from them as the basis for identifying the signal.In response to the wide variety of attack models in the physical layer of car key systems,this paper proposes a semi-supervised learning-based RF fingerprint identification scheme,which uses legitimate data to train K-NN and SVM single classifiers and uses them to identify an unknown dataset consisting of legitimate signals and experimentally constructed attack signals.In this paper,the basic experimental environment was constructed by open source software defined radio(SDR)tools and open source digital signal processing software,and experiments were conducted based on the above scheme.The signals of Mazda keyless entry system(RKE)car keys and Toyota passive keyless entry and start system(PKES)car keys were collected as representatives of two car key systems.K-NN and SVM single classifiers generated by training with 100 legitimate signals were used for testing in single-sided low-frequency relay attacks,SDR replay attacks,and sightless range environments.Another 100 legitimate signals and SDR-generated attack signals were used in the experiments.In this paper,the false negative rate(FNR)is defined as the proportion of illegal attack signals identified as legitimate,and the false positive rate is defined as the proportion of legitimate attack signals identified as illegal.In the unilateral low-frequency relay attack scenario,the FNR is always 0%,while in the SDR replay attack scenario,the lowest FNR is 1.3%and 2.5%for the two tested car keys.The FPR is always maintained at about 1.5%in the no-visual-range environment test.The experimental results show that the scheme in this paper has good discrimination ability in identifying wireless attacks on the physical layer of car keys.