Design of efficient and robust forward error correction for real-time application in coherent optical communication systems

Next generation coherent optical modems for use in dense wavelength-division multiplexed long-haul transmission systems employ soft-decoded forward error correction (FEC) both to increase the reach of single-level modulation formats, such as quaternary phase shift keying (QPSK), and to enable use of higher order modulation formats, such as 16-ary quadrature amplitude modulation. This thesis discusses the design of an efficient and robust FEC scheme with up to 200Gbit/s throughput for implementation in a 28nm ASIC. Polarization multiplexed differentially encoded QPSK (PM DE-QPSK) with 20% coding redundancy is considered as the baseline for FEC optimization and performance characterization. It is investigated, whether limited ASIC resources are better spent on incremental optimization of a binary FEC codec or for joint detection and decoding which combines iterative differential and FEC decoding to so-called turbo differential decoding (TDD). New analysis methods and design rules for the construction of an efficient TDD codec are introduced. The resulting code design provides excellent performance both on the additive white Gaussian noise channel and on realistic optical transmission links. A complexity and feasibility study gives insight into substantial trade-offs between performance, complexity and power dissipation. As proof of concept the studied TDD codec was finally implemented in a 28nm ASIC and real-time measurements for realistic optical transmission scenarios confirm very good performance of TDD. As an extension to the baseline configuration the presented FEC architecture also supports binary phase shift keying and 16-ary quadrature amplitude. In addition to that the TDD codec can also be configured to operate with 7.4% redundancy.