| With the development of communication technology,coherent optical orthogonal frequency division multiplexing(CO-OFDM)communication system has become the main technology of long-distance and high-speed optical transmission technology due to its high anti-dispersion ability,high spectrum efficiency and support for high-order modulation.FPGA,with strong parallel processing ability and real-time performance,is a very effective choice for real-time digital signal.In the paper,the design scheme of OFDM real-time transmitter and receiver is put forward and the specific implementation is described on FPGA.Besides,the high peak to average power ratio(PAPR)of the OFDM symbol makes the system sensitive to nonlinear of fiber in the long-distance transmission,and it affects the transmission performance of system.Because external cavity laser with linewidths less than 100 kHz is expensive,it is necessary to propose an efficient nonlinear equalization algorithm in CO-OFDM system with large laser linewidth.Then the paper researches the nonlinearity of CO-OFDM system with large laser linewidth and high-order quadrature amplitude modulation(QAM),based on general regression neural network,the paper proposes a nonlinear equalization algorithm.The main research contents and results of this paper are as follows:1.The research background,research status and technical principles of CO-OFDM are introduced in this paper.And this paper describes the application of FPGA in CO-OFDM system.2.Based on the principle of OFDM technology,the transmitter and receiver of OFDM baseband system are divided into several modules by the top-down design method,and the paper proposes a symbol timing synchronization algorithm at the receiver in OFDM system,introduces the implementation of the design based on field programmable gate array(FPGA).Then,this paper analyses the advantages of the algorithm.After accomplishment of design,this paper uses Modelsim SE 10.1c to simulate the transmitter and receiver,analyses the experimental result.3.A nonlinear equalization algorithm has been proposed based on the general regression neural network in CO-OFDM system with large laser linewidth and high-order QAM.After performing phase recovery at the receiver,a certain number of the training data is chosen to carry out the training and studying in the general regression neural network(GRNN).In the process,the smoothing factor,the only parameter can be decided in the GRNN.Then,for the detecting data at the receiver,the nonlinear equalization is performed using the GRNN.For 50Gb/s CO-OFDM system with 100 km transmission distance,the numerical simulations have been completed using the proposed GRNN nonlinear equalization algorithm.When the large laser linewidths are adopted in the proposed OFDM system,compared with the back propagation neural network nonlinear equalizer(BPNN-NLE),the proposed method has a better nonlinear equalization performance and a shorter time of training running time.The GRNN nonlinear equalizer(GRNN-NLE)relaxes the requirements on the laser linewidths and reduces the bandwidth requirement with higher-order QAM in CO-OFDM systems,which will contribute greatly to apply CO-OFDM transmission technology to long and medium distance optical fiber transmission system. |
