Beschreibung
At the beginning of the new millennium, fiber optic communication saw a resurgence of coherent receivers with straight forward compensation methods for dominant linear impairments. Spectral efficiency was increased using polarization multiplexing and higher order modulation formats. The fiber optic channel has different characteristics and requirements with respect to signal processing than typical wireless systems. This thesis outlines the design of signal processing algorithms that are specifically adapted to the fiber channel, covering equalization and synchronization algorithms for single carrier receivers. Original work on equalization includes blind dispersion estimation for optically uncompensated links, efficient blind source separation for polarization multiplexed signals, and effective predistortion for polarization-dependent loss. Then, it will be demonstrated how timing recovery can be made tolerant against chromatic dispersion and the differential group delay. Further, joint polarization carrier phase recovery will be introduced and applied to channels that are severely limited by intensity modulated neighbors, significantly increasing the transmission distance. The carrier recovery coverage will be followed by stability measurements of typical local oscillators in fiber optic receivers, analyzing the impact and the necessary means of mitigation. Since differential decoding is typically required due to the high laser linewidth in order to avoid cycle slips, a simple soft differential decoding algorithm will be introduced and analyzed in combination with soft error correction. Finally, blind and data-aided algorithms will be assessed in terms of complexity, convergence and tracking speed, as well as scalability, arriving at a recommendation for future single carrier receivers.
