Research On Coherent Reception DSP Algorithms In Elastic Metro Optical Network

With the development of emerging network services and applications such as Internet of Things,cloud computing,virtual reality,autonomous driving,and 5G,the data traffic carried by the metro network grows rapidly,and the traffic patterns are becoming more diverse.The metro optical network becomes the main carrier of the data traffic.The current mainstream metro optical networks are based on fixed modulation format(MF),fixed connection configuration and fixed spectrum/power efficiency,thus it is difficult to meet the demands of emerging service with delay sensitivity,bandwidth diversity and dynamic bursting.Future metro optical transmissions urgently need modulation and demodulation technologies to support flexible MFs,adj ustable link rates,efficient use of spectrum/power,adaptive compensation for transmission impairments and adaptive transmission of burst services.The single-carrier polarization multiplexing quadrature-amplitude-modulation(PM-QAM)coherent detection scheme,with the help of flexible MFs and digital signal processing(DSP)algorithms,could achieve high-speed,high-efficiency,large-capacity and long-distance transmission,therefore it becomes the mainstream technical solution of flexible metro optical networks.However,as the modulation order increases,the QAM systems with flexible MFs are more sensitive to residual dispersion,polarization mode dispersion(PMD),polarization crosstalk,transceiver frequency offset/linewidth and strong filtering and fiber nonlinearity.The contradiction between the versatility,complexity and performance of the existing coherent DSP algorithms is till prominent,because they are initially designed for fixed MFs,non-mutant channel impairments and continuous transmission modes.Besides,different algorithms are mutually constrained,and it is difficult to achieve format-universality,low-complexity,high damage tolerant and dynamic adaptive compensation for various transmission impairments.Focusing on the above technical problems,this paper has carried out in-depth research on the coherent DSP algorithm.The main research works and innovations are as follows.1.The existing adaptive equalization and polarization demultiplexing algorithm mostly can be applied to a single MF,which cannot satisfy the requirement of universal processing for multiple dynamic PM-QAM signals in elastic metro optical networks.In this paper,a multi-format blind equalization demultiplexing algorithm based on constellation coordinate transformation is proposed.The proposed algorithm uses a transparent constellation coordinate transformation method to transform high-order QAM signals into QPSK signals.The impairments such as residual dispersion,polarization mode dispersion,and polarization crosstalk of different format signals can be universally compensated without additional computational complexity.The simulation results of 16Gbaud PM-QPSK/16QAM/64QAM show that the proposed algorithm not only achieve transparent equalization and polarization demultiplexing for multiple QAM signals,but also has a faster convergence speed than RDA.The convergence overheads of 16/64QAM are reduced to 12%and 23%of the RDA algorithm,respectively.The 16Gbaud PM-QPSK/16-QAM optical back-to-back(BTB)and 800km transmission offline experiments show that the proposed algorithm can achieve multi-format universal demultiplexing and has the same OSNR tolerance than traditional algorithms.2.Facing the contradiction between the MF versatility and low complexity of the existing carrier recovery algorithm,this paper proposes a carrier frequency offset estimation(FOE)algorithm based on multi-symbol interval phase difference in polar coordinates,which is not only applicable to multiple MFs,but also reduces the complexity to 13%of the fourth-order FFT algorithm.The performance is verified by 16Gbaud PM-QPSK/16QAM and 8Gbaud PM-32QAM offline experiments.In addition,based on a quasi-linear relationship between the offset distance and the phase rotation angle introduced by the laser linewidth,a low-complexity multi-format carrier phase estimation(CPE)algorithm is proposed with much lower complexity(about 14%of BPS).To further improve the linewidth tolerance of CPE for high-order QAM signals,a multi-format CPE algorithm with large linewidth tolerance is proposed based on an extended QPSK partitioning method.In the 16Gbaud PM-16/32/64-QAM simulations,the linewidth tolerance improvements of 25%,45%and 51%are observed respectively.The 16Gbaud PM-QPSK/16QAM offline experiment shows that the algorithm has the same OSNR tolerance with the traditional algorithm.3.When the existing algorithms are used to compensate for the strong filtering and fiber nonlinear impairments in Nyquist WDM systems,the contradiction between the performance and low complexity of the algorithm becomes more serious.In this paper,based on the Euclidean distance approximation a low-complexity and multi-format maximum-a-posteriori-probability(MAP)algorithm is proposed.The proposed receiver-side MAP can support impairments suppression by configuring corresponding training pattern look-up-table(LUT)for different MFs.The approximate Euclidean distance calculation is used to remove the large-scale multiplication operations,and the computational complexity(CC)is reduced by 66%.Compared with no MAP algorithm,simulation results of the three-carrier 20Gbaud PM-16QAM Nyqusit-WDM system show that the proposed MAP reduces the required OSNR by 1.5dB at BER=2E-2,increases the transmission distance by 28%and can tolerate a carrier center frequency drift range of-0.5?0.5 GHz.In addition,a 1.5 dB reduction of the required OSNR and 2.6 dB launch power range extension are observed in the triple-carrier 20Gbaud PM-16QAM Nyquist-WDM BTB and 700-km transmission experiments,respectively.4.The strong burst characteristic of data streams is one of the important features of elastic metro optical networks.Facing the problems of fast change of channel impairments,dynamic change of MFs and mutual restraining between the equalization demultiplexing and CPE,we propose a joint polarization demultiplexing and CPE processing scheme based on training sequence.In this scheme,the equalization demultiplexing algorithm adopts a two-stage composite error functions,which can accelerate the pre-convergence speed and improve the convergence accuracy of steady state,and the joint processing could mitigate the deterioration of the phase noise on the performance of the demultiplexing algorithm.The simulation results of the 20Gbaud PM-16/32QAM optical burst coherent systems show that the convergence overheads of 16/3 2QAM are reduced to 33%and 37%of the blind algorithm respectively.Beside the composite error function makes the algorithm have lower steady-state mean square error while the joint processing of equalization demultiplexing and CPE solves the problem of mutual constraint.