Analysis and Optimization of Coded Modulation for Nonlinear Fiber-Optic Communication Systems

This thesis studies coded modulation for optical communication systems whose data rates are limited by nonlinear fiber effects. Achievable information rates are applied as an information-theoretic tool to evaluate the maximum spectral efficiency of these systems. Having established this upper bound, we quantify in simulations and experiments how far a fiber system with forward error correction is operated from its asymptotic limit. By optimizing the demapper, the throughput is significantly improved for optical channels that are strongly affected by fiber nonlinearities. Most common long-haul optical links, however, are found to operate well with the currently used demappers. Two additional techniques to increase the data rates are demonstrated. Probabilistic shaping is found in theory, simulations, and experiments to give significant performance improvements despite enhancing the adverse nonlinear fiber effects. The temporal correlations that originate from chromatic dispersion and nonlinear inter-channel crosstalk are experimentally confirmed to be present and exploited with low-complexity algorithms to increase spectral efficiency.