Research On Secure Transmission And Performance Monitoring Of Optical Signals Based On Machine Learnin

With the development of information digitization and Internet technology,optical network becomes more flexible,high speed and dynamic.But then there will be complex network structure,all kinds of signal damage and information transmission security performance is low.In this paper,the problems of signal security transmission and system parameter monitoring and identification in future intelligent optical networks are discussed.In addition,block compressed sensing technology,embedded encryption technology,Stokes space mapping,transfer learning and convolutional neural network have been widely studied and used as technologies to improve the safety performance and transmission capacity of optical communication networks.Therefore,based on the above technologies,with the main goal of ensuring the safety of optical transmission system and monitoring and identification of optical performance detection parameters,this paper studies the high-security optical transmission system based on block compressed sensing chaotic embedded encryption and the convolution neural network based on transfer learning assistance signal recognition modulation format scheme.The main research contents of this paper are as follows:（1）A chaotic embedded encryption scheme based on block compression sensing is proposed for multi-core fiber orthogonal frequency division multiplexing system.The scheme uses block compressed sensing technology to recover all the information needed from a small amount of data.Meanwhile,in terms of compressed sensing,the coefficient random permutation can make the coefficients of discrete cosine transform distributed randomly,and improve the sampling efficiency of block compressed sensing.The data volume was reduced by 75%compared to data without compressed sensing.A four-dimensional discrete chaotic encryption model generates four masking factors,which are respectively used for coefficient random permutation,measurement matrix,diffusion and singular value decomposition embedding to achieve ultra-high security encryption of four different dimensions.The key space reaches 10¹²⁰.In addition,in chaotic encryption,singular value decomposition technology imparts the noise-like secret image after initial encryption into the carrier image to generate an encrypted image with visual security,which realizes double protection of source image data and external representation.The proposed scheme uses 2km 7-core optical fiber to achieve 78.75 Gb/s transmission of encrypted orthogonal frequency division multiplexing signal.The received optical power is greater than-14dbm,and the bit error rate of core 1-core7 is less than 10^-3.When the compression ratio is set to 0.25 and the attack range of encrypted data is 30%,about 70%of the image information can still be recovered,and the outline and general information of the image can be visually observed.The experimental results show that this scheme can improve the security performance and reduce the complexity of information transmission system.Furthermore,the scheme combines The BCS chaotic embedded encryption technology with MCF-OFDM system,which has a good application prospect in the future optical networks.（2）Based on multi-core optical fiber transmission system,a scheme for monitoring modulation format recognition parameters in optical signal performance parameters is proposed.In this scheme,multiple Stokes sectional planes images are used as signal features which are typed into a transfer learning assisted convolutional neural network to realize modulation format recognition.Compared with the traditional Jones matrix,the Stokes space mapping method is insensitive to polarization mixing,carrier frequency skew and phase offset,therefore,it has better feature representation ability.transfer learning is introduced to transfer the model used in standard single-mode fiber to multi-core optical transmission,reducing the required training data and complexity.In addition,multiple Stokes sectional planes images are input simultaneously,which improves the accuracy of the neural network.Experimental verifications were performed for a polarization division multiplexing Optical transmission system at a symbol rate of 12.5 GBaud by 5 km multi-core optical fiber.Nine modulation formats,including three standard modulation formats,three uniformly shaped modulation formats and three probabilistically shaped modulation formats,were recognized by our scheme.The experimental results show that the scheme achieves high recognition accuracy even at low optical signal-to-noise ratio.Moreover,the required number of training samples is less 40%compared to the traditional convolutional neural network.The proposed scheme has a high tolerance to the crosstalk damage of multi-core optical fiber itself and can realize the short training time of large-capacity space division multiplexing Optical transmission.Our findings have the potential to be used in the next generation of a fiber transmission system.