Research On Stokes Space-based Key Technologies For Cognitive Optical Receivers

As optical fiber communication networks present new characteristics such as high burst,diversified services and high intelligence,cognitive optical networks(CONs)have been considered as one of the attractive solutions for next generation optical networks due to their high adaptive bandwidth allocation according to specific network environment and users' demands.In the CONs,key parameters of received optical signals,such as modulation format,optical signal to noise ratio(OSNR),etc.,are required at the receivers to optimize subsequent signal processing and demodulation.However,in the practical optical fiber communication systems,evaluation of parameters of received optical signals is often affected by fiber channel impairments such as large phase noise of coherent optical receivers,impairments resulted from polarization multiplexing and fiber nonlinearity.Recently,Stokes space based analysis for coherent optical signals has attracted extensive attention since they are independent of polarization mixing,as well as frequency offset and phase noise of the coherent optical receivers.In this thesis,we mainly focus on the study of Stokes Space-based key technologies for cognitive optical receivers,which mainly include modulation format identification(MFI)and OSNR monitoring.Details and contributions mainly include:(1)Study on how the fiber channel impairments affect distribution of signals in Stokes space by utilizing MATLAB simulations;(2)Designing a new MFI technique based on Stokes-space-based subtraction clustering algorithm,which can realize correct identification among five modulationformats(BPSK,QPSK,8PSK,8QAM,16QAM)and have lower time complexity compared with previously proposed Stokes-space-based MFI schemes(3)Realizing an OSNR monitoring technique based on Stokes-space-based artificial neural network(ANN)algorithm.In this technique,OSNRs of signals are evaluated on the basis of the mapping between the constellation dispersion on the least square plane(LSP)in Stokes space and the OSNR,created by utilizing the ANN algorithm.Simulation results shows an accuracy within 0.5dB can be realized when OSNR>5dB,achieving higher accuracy compared with general function fitting methods.(4)Building a long-distance polarization-multiplexed(PM)coherent optical communication system to verify and evaluate the proposed algorithms.Good performance can be achieved for both the proposed MFI and OSNR monitoring algorithms in the experiment demonstrations.