Research On Multi-parameter Co-estimation Of Optical Signal Based On Adaptive Kalman Filtering

Driven by the technologies of big data,internet of things and data center,optical networks will be characterized by high transmission rate,flexible configuration and larger transmission capacity in future.Thus,those changes will bring lots of challenges for the digital signal processing(DSP)modules in the optical receiver,such as quick tracking the polarization state of the signal,quick acquisition the frequency offset and phase noise with high estimation accuracy.However,the commonly used algorithms in DSP have their own drawbacks including singularity,low convergence speed,low estimation accuracy and so on.Owing to its high estimation accuracy and fast-tracking capability,Kalman filter(KF)and its variants are widely used for compensating the polarization mixing,frequency offset and phase noise.However,according to the KF theory,the estimation accuracy and tracking capability in KF are strictly dependent on its tuning value Q.In general speaking,Q should be set as different values under different operation scenarios such as optical signal-to-noise ratio(OSNR),polarization rotation rates and linewidth.Thus,the performances of KF are limited under dynamic and flexible optical networks.The main points in this thesis are applying adaptive Kalman filter(AKF)for compensating the polarization mixing,frequency offset,and phase noise in the coherent optical systems.At the begin of the thesis,the basic principle of the coherent optical communication system,the related algorithms in DSP and KF as well as AKF are described in detail.Then,the performance of AKF and commonly used algorithms under different operation scenarios are investigated via numerical simulation and experiment.During the investigation of polarization state and phase noise tracking based on AKF,the performances of AKF and extended Kalman filter(EKF)are investigated under different OSNR,polarization rotation rates,and linewidth.The simulation results show that AKF offers smaller OSNR penalty at the target BER of 1e-3,high polarization tracking capability and linewidth tolerance.Moreover,simulation results also confirm that the performances of AKF are independent on its Q.Then,the polarization tracking capability of the AKF and EKF are investigated in the experimental systems.The experimental results further confirm the advantages of AKF in polarization tracking capability.During the investigation of carrier phase recovery based on modified semi-adaptive Kalman filter(MSAKF),the reasons of the frequency offset cycle slips in SAKF are investigated in detail.Then,a MSAKF algorithm for monitoring and compensating the frequency offset cycle slip during the recursive of Kalman filter is proposed in this section.The simulation results show that MSAKF can monitor and compensate the frequency offset cycle slips under different frequency offset scenarios.Besides,MSAKF also shows quick frequency offset tracking capability and high linewidth tolerance.However,although SAKF can switch its Q under different operation condition,it cannot be regarded as the adaptive Kalman filter in some sense.Thus,the third points in this thesis are proposed to use the covariance matching method to adaptive adjust its Q during the recursive of KF under different operation conditions.Compared with EKF,the performances of AKF are investigated in the aspect of frequency offset estimation accuracy and ranges as well as the frequency offset tracking capability.Simulation results confirm that AKF offers high frequency offset estimation accuracy and ranges as well as tracking capability.Besides,the proposed scheme also shows berrer linewidth tolerance.Finally,the experimental results further confirm the advantages of AKF in the high frequency offset estimation and the ability of Q adaptive configuration.