ABSTRACT
This chapter reviews fundamental information-theoretic concepts related to fiber optic communications, channel modeling, and calculations of channel capacity. It discusses different fiber-optic channel models and the challenges related to capacity calculations. The channel model needs to incorporate nonlinear effects in the system: signal-signal and signal-noise beating as a result of the intersymbol interference. A number of channel models have been proposed for the information-theoretic description of fiber-optic systems. The dashed curves correspond to the achievable information rates with a uniform input PMF, and the solid curves to a probabilistically optimized input PMF with a modified Blahut–Arimoto algorithm for the fiber channel. Achieving larger shaping gains would require lifting the assumptions on the channel memory and/or the likelihood function. In fiber-optic systems, the channel memory is crucial as dispersion induces pulse broadening, while nonlinearity induces complex nonlinear interactions. The channel is simulated with the split-step Fourier method with a step size of 8 km and dual polarization is employed.
