Research On Algorithms To Compensate Fiber Dispersion And Nonlinearity In Coherent Optical System

Compared with the conventional intensity modulation/direct detection, the coherent detection is of higher reception sensitivity and has attracted a lot of research attention of many international scholars and companies. With the coherent detection technology, the real-time40Gb/s and100Gb/s optical transmission systems have been demonstrated. To further enhance the spectral efficiency and transmission rate, it is an excellent solution to combine the coherent detection with the existing polarization-division multiplex (PDM) and wavelength-division multiplex (WDM) technologies. And the WDM-PDM16QAM coherent optical transmission system is a very good example. But the coherent detection system also faces many challenges, such as the frequency-offset, the phase noise, the dispersion and the fiber nonlinearity. So it is essential to compensate different impairments, among which the compensation for the dispersion and fiber nonlinearity is especially important and has become the research point in recent years. Of all the approaches, the compensation algorithms based on digital signal processing (DSP) is particularly important.In this paper, we first introduce the basic concepts and related technologies of the coherent optical system. Then follows the specific introduction of the theory and DSP compensation methods for the dispersion and fiber nonlinearity for both single-channel and multi-channel systems. In this paper, analysis and simulations of various algorithms are conducted to verify the compensation effects and to make further improvements. First, the simulation system for the WDM-PDM16QAM coherent optical system is established. Simulations of different algorithms to compensate the dispersion and fiber nonlinearity for both single-channel and multi-channel systems are conducted to validate the compensating effects. Second, in single-channel system, by analyzing the conventional Split-Step Fourier Method (SSFM), a novel integral SSFM (I-SSFM) algorithm is proposed. The I-SSFM avoids the trapezoidal-approximation and iterations, which are used in the conventional SSFM. And the nonlinear phase can be calculated directly and accurately by ultilizing the I-SSFM. So the I-SSFM is of lower complexity and better performance. Finally, in multi-channel systems, a detection method based on the Gaussian-Mixture-Model (GMM) is introduced. The GMM of the received signals is established by utilizing the Expectation-Maximization (EM) algorithm with the training symbols. Then the GMM is used to classify and detect the received signals. In this way the calculation complexity is reduced significantly only at a cost of acceptable degradation of system performance.