| Optical fiber communication has been widespread adopted for its advantages of long transmission distance and large capacity in various applications including personal,commercial and military communications.With the increasing demand for network capacity,the issue of securing optical network is becoming increasingly important.As with other network types,traditional ways for securing optical network are done by employing cryptographic protocols at the higher layers of the protocol stack.However,conventional cryptographic schemes are facing a major treat as the development of quantum computers,which have the potential to crack the ciphers in a short period of time.Thus,it is desirable to build security at the physical layer of optical network and resist attacks that might target this layer.Encryption on the physical layer of optical communication systems also enables data signals to be encrypted with high speed and low latency.For instance,the Y-00 quantum stream cipher is one of the physical layer encryption schemes that uses inevitable quantum noise at detection to mask the information.The scheme relies on an encryption algorithm to convert a lower-order quadrature amplitude modulation(QAM)original constellation to an ultra-dense higher-order QAM constellation,which together with fundamental shot noise in photodetection makes recovery of the information impossible without the key.A crucial design parameter of the Y-00 cipher scheme is the confusion and diffusion effect of noise.The better the confusion and diffusion effect of noise in optical fiber link,the more security the system can provide.Focusing on the problem of noise mixing and diffusion in quantum noise cipher system,the main research work of this paper is as follows:(1)In the quantum noise encryption system with quadrature amplitude modulation,aiming at the weak noise mixing and diffusion effect of the existing system,we propose a noise confusion and diffusion technology based on neural network to improve the noise confusion and diffusion effect.From the perspective of neural network stability,we analyze the security of this scheme.The feasibility of the encryption algorithm is demonstrated by simulation.Finally,in terms of the output results of neural network,heterodyne measurement detection probability and quantum noise masking number,the influence of this scheme on quantum noise cipher system is expounded respectively.(2)In the quantum noise cipher system with quadrature amplitude modulation,aiming at the problem that the existing noise confusion and diffusion effect needs high-order signals to realize,this paper proposes a noise confusion and diffusion technology based on digital chaotic disturbance to improve the security of the optical cipher system.The security of the algorithm is evaluated from the perspectives of information theory security and computational complexity security.In this work,the effectiveness of the proposed scheme in coherent optical communication system with standard single-mode fiber(SSMF)exceeding 300 km is illustrated.Finally,based on the transmission quality of images and the unpredictability of signals and constellations,the feasibility of this scheme is evaluated respectively.(3)Focusing on the problem that the traditional security evaluation metric cannot meet the security measurement of quantum noise cipher system after confusion and diffusion of noise,this paper proposes the formula of quantum noise masking number treated by confusion and diffusion effect of noise,and puts forward a security metric for confusion and diffusion effect of quantum noise cipher,i.e.fractional Hamming distance histogram.The mean value of the histogram represents the confusion of noise,and the variance represents the diffusion of noise.Participated in the security evaluation platform of quantum noise encryption,conducted security evaluation experiments in 300km singlemode fiber,and evaluated and analyzed the optical signals after longdistance transmission by using the number of masked signals and Hamming distance.Experiments show that the noise confusion and diffusion effect improve the number of masked signals,and the mean and variance of Hamming distance histogram are larger than those of unmixed one. |
