Hosseinzadeh, Alireza2020-01-232022-01-232020-01-232020-01-17http://hdl.handle.net/10012/15553Recent tele-communication hugely relies on fiber-optic systems. Linear fiber-optic impairments can be modeled with chromatic and polarization-mode dispersion. However, nonlinear impairments such as SPM and XPM noise are much more complicated to model and compensate. In this thesis, we try to manipulate an original model which is used by Ciena Corporation -called the C-model- to achieve a simpler form which provides lower time-complexity and performs efficiently on random data stream during online transmission. This simplified model is defined based on C-matrix which is used to calculate fiber-link Self-Phase Modulation (SPM) noise. Our model, which is fitted on random training data sets, is evaluated with respect to different performance measures such as Minimum Mean Square Error (MMSE). Different analytic and iterative algorithms such as gradient Descent and Moore-Penrose pseudo-inverse method are proposed for model fitting and it is shown that these algorithms perform quite efficiently according to error calculations. Also, we propose a method that enables us to track the perturbations in model parameters which are caused by the changes in channel characteristics that can happen frequently in online fiber-link transmission.enmodelingcompensationSPMself-phase modulationnoisefiber-linkMaster Thesis