Research On Receiver Equalization Technologies For Coherent Optical OFDM System

Since the beginning of the 21st century,the Internet has penetrated into all aspects of people's lives,such as information acquisition,e-commerce,interactive interaction,entertainment applications and so on.The significant growth of Internet bandwidth has driven the development of the telecommunications industry worldwide.Therefore,as the backbone of modern information infrastructure,the demand for transmission capacity of optical fiber communication systems is also growing.From different types of optical transmission systems,Orthogonal Frequency Division Multiplexing（OFDM）systems based on coherent optical detection are distinguished by receiver sensitivity,spectral efficiency,and robustness to polarization dispersion.This paper focuses on receiver equalization technologies based on Coherent Optical OFDM（CO-OFDM）system.There are two main factors in the optical OFDM system that cause misjudgment of the received symbol constellation:fiber channel dispersion and phase noise of the transmitting/receiving lasers.Among them,the former consists of chromatic dispersion and polarization mode dispersion varying with temperature,and mainly causes the constellation divergence;while the latter changes with linewidth variation,and mainly causes the constellation rotation.The above problems can be handled effectively by channel equalization and phase estimation/compensation.Therefore,the system equalization algorithm for the main noise sources of CO-OFDM system is studied in this paper.The specific work and innovations are as follows:Firstly,from the theoretical point of view,the principle and existing problems of coherent optical OFDM system are analyzed theoretically.According to the principle and model of coherent optical OFDM,the simulation platform used in this paper is built.Secondly,a channel equalization technique is proposed.For the first time,the Affine Projection based Sign Modified Constant Modulus Algorithm（AP-SMCMA）is proposed for channel linear damage caused by channel dispersion,and a single 10 Gb/s 160 km CO-OFDM system simulation verification is performed for comparison analysis.Simulation results show that compared with MCMA,AP-SMCMA can bring 0.4 dB power gain at the Forward Error Compensation（FEC）limit(BER=3.8×10^-3)and 10 times convergence speed gain at the best convergence point.Thirdly,two phase noise suppression techniques are proposed for the first time.Based on the properties of the phase noise of the laser in accordance with the Wiener process,the extended Kalman filter based on Lagrange interpolation（Lagrange Interpolation based Extended Kalman Filter,LRI-EKF）is proposed;from the perspective of machine learning,the Least Square based Back Propagation Neural Network（LS-BPNN）algorithm is proposed.The simulation and comparison analysis of single channel 25 Gb/s 160 km CO-OFDM system is carried out.Simulation results show that compared with the Common Phase Error（CPE）compensation algorithm,the LRI-EKF algorithm can reduce the OSNR penalty by 1.2 dB and improve the laser linewidth tolerance by 300 kHz at the FEC limit;compared to the existing LS algorithm,the LS-BPNN algorithm can reduce the required received optical power by 1.2 dB and increase the laser linewidth tolerance by 160 kHz at the FEC limit.Then the advantages and disadvantages of these two algorithms is discussed.