Research On Nonlinear Compensation Technology In Optical Fiber Communication System

In the strategic project of"Eastern Data and Western Computing",the ultra-high-speed transmission of data within the data center cluster mostly adopts the intensity modulation direct detection(IMDD)data center optical fiber communication system based on four-level pulse amplitude modulation(PAM4),while signals acquired through PAM4 modulation are more sensitive to channel nonlinear impairment than traditional NRZ modulation.The architecture of radio-over-fiber(ROF)system is simple,the low-cost ROF communication system provides a possible solution for a large number of mobile users to access the data center network at high speed,the noise level and linearity are the most critical technical indicators.The long-distance,high-capacity coherent optical communication system is used for the data interconnection between the eight hub nodes of the computing power network.Nonlinear effect caused by optical fiber is the factor limiting the performance of signals received.Therefore,it is necessary to overcome such shortcomings of optical fiber communication system.This thesis focuses on the study of nonlinear compensation techniques for IMDD short-distance optical fiber communication systems,ROF systems and long-distance coherent communication systems.Contributions of the thesis are as follows.(1)In high-speed IMDD short-distance optical fiber communication systems,the nonlinear distortions can cause unequal eye height and eye skew,and degrade BER performance of received signals.A nonlinear dynamic slicer algorithm is proposed to eliminate the nonlinearity of PAM4 signal.By using the proposed algrorithm,the BER performance has been improved from 1.5e-4 to 1.51e-5 when the received optical power is set at 0 d Bm.To eliminate the eye skew distortion,an eye dependent nonlinear compensation algorithm is proposed.When the received optical power is-1 d Bm,the Bit Error Rate(BER)performance is improved from 1.2e-5 to 6.6e-6 with the help of the proposed algorithm,in comparison with the traditional volterra nonlinear equalization algorithm.A low-complexity error-based feedforward equalizer is proposed to lessen the impacts of FFE-enhanced in-band high frequency noise and improve the receiver sensitivity performance from-0.2 d Bm to-5.2 d Bm.(2)In the ROF optical fiber communication system,both nonlinear distortion and noise of the system are the key factors that reduce the BER performance of the received signals.To cope with the channel interference and noise that deteriorate the condition number performance of the Multi-input Multi-output(MIMO)channel,a simplified MIMO demodulation algorithm based on lattice reduction zero-forcing(LR-ZF)is formulated.Power penalty larger than 0.9 d B is achieved by using the proposed LR-ZF algorithm,compared to the commonly used zero-forcing algorithm.Moreover,a joint K-means clustering and single-ring LR-ZF nonlinear compensation algorithm is proposed,which can strengthen the capability to resist the nonlinear distortion of ROF systems when the received optical power exceeds-2 d Bm.(3)For long-distance coherent optical fiber communication system,nonlinear effects in the optical fiber can degrade the performance of Q~2 factor and BER of the received signals.It is able to relieve the degradation owing to optical fiber nonlinearity effectively by applying the conventional perturbative nonlinearity compensation(PNC)algorithm,which,however,is quite intricate to perform in aspects of both calculation and selection of the perturbation terms.In the thesis,a hard-slicer look-up table perturbation nonlinear compensation algorithm is proposed to deal with the mentioned disadvantages.Compared with the traditional PNC algorithm,the complexity of the proposed simplified algorithm is reduced to 1/21 with the Q~2 factor of each channel degradated by only 0.04 d B in the wavelength division multiplexing system.