Research On Physical Layer Security Technology Of Optical Fiber Communication Based On Device Physical Fingerprint

Optical fiber communication is an important part of communication system,which is used to ensure high quality,high reliability and low cost information transmission.However,the threats and challenges faced by optical fiber communication security have received increasing attention from academia,government,and business.As the complexity of the network continues to evolve,stronger measures are needed to ensure network integrity and reduce unauthorized access.The solutions of network access control include identity binding based on MAC address and identity authentication based on upper digital certificate,etc.All of these solutions have potential vulnerabilities,including attackers forging specific MAC address information or stealing credentials related to digital certificates.As technological capabilities continue to improve,more and more devices can be connected to the network.This extension presents unique security challenges and increases the possibility of unauthorized access.The device identification technology based on physical fingerprint of the device provides a reliable means to improve the security of network authentication.Due to the inconsistencies in the manufacturing process and equipment components,a fixed physical fingerprint of the equipment is formed in the signals emitted by the communication equipment.The resulting changes will lead to the deviation of symbol rate and center frequency in the communication system,resulting in the difference of physical layer characteristics,which are different in different devices.Amplifiers,capacitors,inductors and oscillators also have minor defects that affect the physical fingerprint of the device.This makes it possible to use physical fingerprints for device identification.This paper studies the security technology of optical fiber communication based on the physical fingerprint of the device,designs an intensity modulation/direct detection(IM/DD)experimental system,performs a series of preprocessing operations on the collected signal,and proposes a feature extraction method based on the correlation peak,uses the hybrid classifier to classify and identify the optical fiber Ethernet device,and evaluates the classification performance of the system Estimation analysis.Based on the above feature extraction method,this paper also proposes an innovative optical fiber eavesdropping detection method based on physical fingerprint.The main work of this paper is as follows:1.An experimental system of intensity modulation/direct detection(IM/DD)consisting of24 optical Ethernet devices is designed.It is mainly composed of laser transmitter(optical transmitter),single-mode optical fiber,light decay,avalanche photodiode(optical receiver)and digital oscilloscope.2.In view of the actual acquisition of optical fiber amplitude signal and IEEE 802.3protocol,the preprocessing operation flow for physical fingerprint feature extraction is designed.It includes the following steps: signal synchronization,energy normalization processing,digital filtering and adding analog noise to simulate different environments.3.A feature extraction method based on correlation peak is proposed for optical fiber amplitude signal.The extracted features mainly include three aspects: 1)Carry out correlation operation on several signals after periodic interval to extract statistical features.2)Calculate the autocorrelation values of the signals after several periodic intervals at the theoretical main peak and some points around it,and use the weighted average as the correlation peak.3)Curve fitting was carried out for these correlation peaks to obtain curve fitting coefficient.4.The mixed classifier is used to classify and identify the equipment.The system parameters are optimized through a large number of experiments.The classification performance is evaluated and analyzed.The experimental results show that the classification accuracy can reach 85.76% and 91.11% at 0d B and 20 d B SNR,which is significantly better than the method based on classical statistical characteristics.5.An innovative method based on the physical fingerprint of the device is proposed,which uses support vector machine algorithm(SVM)to train and identify the physical fingerprint of the device before and after the optical fiber eavesdropping.When the signal-to-noise ratio is 20 d B,the recognition accuracy of eavesdropping detection can reach 87.5%,which can realize the effective early warning of optical fiber eavesdropping.