Research On Adaptive Modulation Technology In High-Speed And Large-Capacity Optical Communication System

Under the general trend of information technology development,more and more new technologies have emerged.Big Data,Artificial Intelligence and other technologies with communication as access methods have gradually infiltrated into daily life.A variety of new forms of user experience,such as virtual reality technology,Internet TV,online live broadcast,etc.,brings novel experiences,but also produces a huge demand for communication traffic.Especially since 2020,affected by COVID-19,the main scenes of people's life,studying and working have moved from offline to the network.Telework,video conferencing and online education have become the norm in life.Therefore,people's dependence on the network is continuously increasing.Along with it,there is a heavy burden of data bearing in network transmission.As the main bearing force of Internet data traffic,how to further improve the transmission rate and increase the channel capacity of optical fiber communication system is the key problem to be solved urgently at present.This thesis takes the coherent optical orthogonal frequency division multiplexing(CO-OFDM)system as the main background,and focuses on adaptive modulation technology,modulation format identification technology and nonlinear decision technology of such high-speed and large-capacity system to improve the transmission performance.An adaptive modulation scheme based on deep neural network(DNN)-assisted subcarrier performance monitoring is proposed to allocate a better transmission scheme for the system and improve the transmission performance of the system.A subcarrier modulation format recognition scheme based on convolutional neural network(CNN)and I/Q component distribution histogram is proposed,which can reduce signaling overhead and realize subcarrier modulation format recognition at the receiving end,and then complete subsequent signal processing and decision.A nonlinear decision scheme based on unsupervised clustering algorithm based on the gaussian mixture model(UCGMM)algorithm is proposed,and the algorithm is used to judge constellation points,which effectively compensates the influence of nonlinear damage,reduces the system bit error rate and improves the reliability.The main work of this thesis is as follows:(1)An adaptive modulation scheme based on DNN-assisted subcarrier performance monitoring is proposed.In this scheme,DNN model in deep learning is used to estimate the effective SNR of subcarriers,so as to monitor the performance of subcarriers.Then according to the estimated effective signal-to-noise ratio of subcarriers,the subcarriers of OFDM are grouped and the appropriate modulation format is selected to produce a better transmission scheme.According to simulation analysis,under the condition of fixed transmission rate and transmission power,compared with the transmission scheme of 16QAM fixed modulation format,the proposed transmission scheme has an optical signal-to-noise ratio(OSNR)gain of 1.75dB under the bit error ratio(BER)threshold of 1e-3.It makes better use of channel conditions and effectively reduces the overall bit error rate of the system.(2)A sub-carrier modulation format identification scheme based on CNN and I/Q component distribution histogram is proposed.In this scheme,I/Q component distribution histogram with high resolution under low signal-to-noise ratio is used as identification sample.Using the distribution characteristics of I component and Q component,combined with CNN model,several modulation formats such as BPSK,QPSK,8QAM,16QAM,32QAM and 64QAM are recognized.The blind modulation format recognition at the receiving end of the system is realized.Through model training and simulation analysis,the recognition accuracy of the modulation formats other than 64QAM can be achieved at 100%when the OSNR is greater than 10dB.And the recognition accuracy of the 64QAM is 100%in the OSNR above 16dB.The scheme reduces the signaling overhead and provides accurate modulation format information for the receiver of the system with adaptive modulation.(3)A nonlinear decision scheme based on UCGMM algorithm is proposed.By comparing and analyzing the performance of K-means and Gaussian mixture model clustering algorithms in 64QAM transmission system affected by nonlinearity,UCGMM algorithm is proposed as a nonlinear decision method for coherent optical OFDM system.Using this algorithm to judge the received signal,the influence of optical fiber nonlinearity on signals is effectively compensated,and the bit error rate of the system is reduced.In 64QAM-CO-OFDM communication system,the transmission performance of this scheme in back-to-back transmission and optical fiber transmission is simulated separately.As can be seen from the results,compared with not using clustering algorithm and K-means algorithm,UCGMM algorithm can obtain OSNR gains of about 2dB and 0.6dB respectively in back-to-back transmission system.In the simulation case of optical fiber transmission,UCGMM algorithm increases the transmission distance by 75km compared with the direct demodulation case,and also increases the transmission distance by 45km compared with K-means algorithm.