Research On Key DSP Algorithms Of Faster-than-nyouist WDM Coherent Optical Transmission Systems

As technologies such as 5G?Cloud Computing,Internet of Things,Autopilot develop rapidly,the demand for high-speed,high spectral efficiency and large capacity optical communication is rising.Dense wavelength division multiplexing technology is widely used because of large capacity.However,as the available frequency of C+L band is almost fully utilized,the traditional Dense wavelength division multiplexing system which only depends on increasing the number of multiplexed wavelengths is facing the physical bottleneck that the system capacity is difficult to further improve.Nyquist WDM is a feasible scheme to realize high-capacity optical transmission by compressing the signal bandwidth,but the orthogonal transmission criterion limits the further improvement of the system spectral efficiency.The Faster-than-Nyquist(FTN)coherent optical transmission technology proposed in recent years breaks the orthogonal transmission criterion to realize higher spectral efficiency than the Nyquist rate.With the help of powerful Digital Signal Processing(DSP)algorithm,it can realize efficient compensation and signal recovery of various linear/nonlinear damages(including inter-symbol inference introduced by FTN).It is a potential development technology for large capacity coherent optical transmission.However,in the FTN system with strong ISI,DSP algorithm faces many challenges,such as the traditional clock recovery algorithm Gardner is difficult to extract clock information accurately and reliably,the traditional carrier phase recovery algorithms Viterbi-Viterbi/Blind Phase Search are faced with serious cycle slip in the FTN system,the ISI introduced by FTN strong filter needs high performance and low complexity equalization algorithm,etc.In this paper,the key DSP algorithms of FTN-WDM coherent optical transmission system are studied.1.The reason why the traditional clock recovery algorithm is not suitable for Polarization Multiplexing Quadrature Phase Shift Keying(PM-QPSK)FTN system is studied.There exits two problems.The one is that the clock component of the traditional Gardner algorithm disappears at the baud rate in FTN systems,which leads to the algorithm failure.The other is that the slow convergence speed of the clock recovery algorithm based on signal power.Based on aforementioned,a combined algorithm of clock recovery based on signal power and adaptive equalization and polarization demultiplexing is proposed,which can effectively speed up the clock recovery convergence speed without increasing additional computational complexity.In the three channel 32GBaud PM-QPSK FTN-WDM simulation system,the effectiveness of the algorithm is verified.The convergence speed of the joint algorithm is about 40%quicker than that of the non-joint algorithm,and the OSNR tolerance of the joint algorithm is about 0.2dB better than that of the non-joint algorithm.2.Aiming at the cycle slip problem in PM-QPSK FTN systems,the traditional carrier phase estimation algorithm Viterbi-Viterbi is improved,and a blind carrier phase recovery algorithm based on mean square error detection is proposed.The proposed algorithm can solve the cycle slip problem at the cost of 6%computational complexity increase.In the three channel 32GBaud PM-QPSK FTN-WDM simulation system,the key parameters of the algorithm are optimized.When the acceleration factor is 1/0.9/0.85/0.8,the corresponding linewidth tolerance is 5MHz?4.2MHz?3.7MHz?3MHz.However,compared with the case without linewidth,0.9/0.85/0.8 has the OSNR cost of 0.5dB,0.5dB and 0.7dB(under the FEC threshold of bit error rate 2 × 10-2).3.Aiming at the problem of adaptive equalization and polarization demultiplexing algorithm Constant Module Algorithm amplifying high frequency noise in PM-QPSK FTN systems,the paper designs and verifies an inter-symbol inference equalization scheme based on the combination of Quadrature duo-binary(QDB)filter and Maximum a Posteriori(MAP)algorithm.Firstly,QDB is used to suppress the high-frequency noise,and then MAP algorithm is used to equalize the inter symbol interference.Comparing the computational complexity,the increase of computational complexity of QDB+MAP algorithm is negligible.The simulation results of three channel 32GBaud PM-QPSK FTN-WDM 800km transmission show that the OSNR tolerance of QDB+MAP algorithm is 1dB better than that of MAP algorithm with the bit error rate 1 × 10-2 as the threshold.