| To meet the increasing demand of transmission network data in the future,general multi carrier(GMC)technology has attracted extensive attention in the field of communication because of its excellent spectral efficiency,strong dispersion tolerance and flexible access mode.This paper summarizes the GMC technology,mainly including orthogonal frequency division multiplexing(OFDM),discrete multi tone modulation(DMT),filter bank multi carrier(FBMC)and universal filter multi carrier(UFMC),which are collectively referred to as GMC technology.Generalized multicarrier technology has attracted extensive attention in the field of optical communication,such as coherent optical OFDM for long-distance backbone network transmission,OFDM and FBMC for passive optical network access,and DMT,which is very popular in optical wireless communication.At the same time,there are still many serious problems in GMC optical transmission technology,such as phase noise,channel estimation and equalization,physical layer information security,nonlinear interference and so on.This paper studies the above problems and puts forward solutions.The main work and five innovative points are as follows:1.A coherent optical generalized multicarrier phase noise suppression algorithm based on Gaussian basis expansion is proposed,and its effectiveness is verified by experiments.In this paper,the mathematical model of coherent optical OFDM is derived,focusing on the phase noise caused by laser linewidth.Through theoretical analysis,it is concluded that Gaussian basis is suitable for fitting the phase noise following Wiener process.Experimental results show that the Gaussian basis expansion algorithm proposed in this paper can effectively fit the phase noise caused by linewidth up to 8MHz.Compared with the common phase error compensation algorithm and orthogonal basis expansion algorithm,the Q value of coherent optical OFDM system is increased by 1.7db and 0.5dB respectively.2.The design scheme of short-range access FBMC training sequence based on interference cancellation criterion is proposed,and its effectiveness is verified by simulation.Aiming at the characteristics of low cumulative noise of short-distance access optical fiber channel,this paper focuses on reducing imaginary part interference to ensure the accuracy of channel estimation.The channel estimation process of FBMC is deduced theoretically,the influence of training sequence on channel estimation performance is analyzed,and the training sequence which can completely suppress the first-order imaginary part interference is designed.The simulation results show that in the case of back-to-back transmission,the Q value of the training sequence scheme proposed in this paper is 0.9dB higher than the training sequence based on the interference approximation criterion,and does not increase any computational complexity.3.A physical layer encryption scheme of GMC passive optical network based on multi scroll chaos and a novel physical layer encryption scheme of pulse amplitude modulation discrete DMT with three-dimensional multi wing chaos are proposed.In this paper,a new three-dimensional multi scroll chaos is used to encrypt the physical layer signal.Two dimensions(x,y)of multi scroll chaos are used for constellation point confusion of 16-QAM,and one dimension(z)is used for bit data XOR operation.The multi scroll chaos used in this scheme has self limiting characteristics.It avoids all complex modulo and rounding operations.Compared with the previous Chen chaotic encryption scheme,the computational complexity is reduced by 34%.The multi scroll chaotic encryption scheme also has very high key sensitivity.When the key deviation is 10-17,it cannot be decrypted correctly.Therefore,the proposed scheme has a huge key space.In this paper,a three-dimensional multi wing chaotic encryption scheme is designed.One dimension(x)is used for bit data XOR operation,one dimension(y)is used for pulse amplitude modulation signal amplitude confusion,and one dimension(z)is used for frequency domain subcarrier scrambling.Further deepen the level of signal encryption and expand the key space.The encrypted transmission of digital image in 60km standard single-mode fiber is successfully demonstrated.The encrypted digital image has a flat histogram,which can effectively resist statistical attacks.4.A nonlinear decision optimization method for DMT optical wireless communication system based on machine learning is proposed,and the effectiveness of the proposed algorithm is verified by simulation.In this paper,two machine learning algorithms based on K-nearest neighbor and support vector machine are designed and trained to solve the problem of nonlinear distribution of DMT constellation points.The proposed two machine learning algorithms reconstruct the reasonable nonlinear boundary of constellation points,and do not use the low-performance maximum likelihood linear segmentation line to judge constellation points.Simulation results show that compared with the maximum likelihood estimation decision method,the machine learning algorithm can achieve performance optimization of 0.5 to 1 order of magnitude. |
