Study On Identification Methods For Optical Signals

In order to meet the increasing demand of voice, video, and large data transmission services, optical fiber communication systems may accommodate optical signals with different modulation formats and data rates. Such heterogeneous optical networks make the management of networks and signal reception much more complex. To realize intelligent network management and autonomous optical signal reception, the research on the optical signal identification techniques is particularly important.This thesis proposes two kinds of optical signal identification methods based on direct detection and coherent detection for optical network intermediate nodes and terminals with coherent receivers, respectively. Firstly, based on the the periodicity of first and second order auto-correlation function of the asynchronous sampling sequences of signals with different modulation formats, a symbol rate identification algorithm independent of modulation format is realized. Furthermore according to that the sequence peak to average ratio variation with chromatic dispersion, the dispersion tolerance of the algorithm can be expanded by adaptive dispersion compensation. Secondly, the differences between the high resolution spectra of signals with different modulation formats and data rates are analyzed. And an optical signal modulation format identification method is proposed utilizing principal component analysis(PCA) and support vector machine(SVM) algorithms. This method doesn't need any prior knowledge. The impact of modulation imperfections and channel impairments on the identification results are also evaluated. Finally, based on the coherent optical receiver, a symbol rate identification method is proposed based on the clock tone obtained by hybrid correlation function, which can mitigate the impact of polarization mode dispersion. The signal instantaneous power distribution under large dispersion is used to distinguish non-polarization multiplexing(NPM) and polarization multiplexing(PM). Utilizing the characteristics of the FFT spectrum of the squared and fourth powered signal fields, and the amplitude error magnitude outputted by the CMA algorithm, signals with commonly used modulation formats including OOK, BPSK, QPSK and 16 QAM signals can be correctly identified.