Book Description

A novel low-complexity architecture for forward error correction (FEC) in optical communication is proposed. The architecture consists of an inner soft-decision low-density parity check (LDPC) code concatenated with an outer hard-decision staircase or zipper code. The inner code is tasked with reducing the bit error probability below the level that allows the outer code to deliver on the stringent output bit error rate required in optical communication. A hardware-friendly quasi-cyclic construction is adopted for the inner codes. The concatenated code is optimized by minimizing the estimated data-flow at the decoder. A method is developed to obtain complexity-optimized inner-code ensembles. A key feature emerging from this optimization is that it pays to leave some inner codeword bits completely uncoded, thereby greatly reducing the decoding complexity. The trade-off between performance and complexity of the designed codes is characterized by a Pareto frontier. In binary modulation, up to 71% reduction in complexity is achieved compared to previously existing designs. Higher-order modulation via multilevel coding (MLC) is compared with bit-interleaved coded modulation (BICM) from a performance-versus-complexity standpoint. In both approaches, complexity-optimized error-reducing LDPC inner codes are designed for concatenation with an outer hard-decision code, for various modulation orders. Code designs for MLC are shown to provide significant advantages relative to designs for BICM over the entire performance-complexity tradeoff space, for a range of modulation orders. Codes designed for MLC can operate with 78% less complexity, or provide up to 1.2 dB coding gain compared to designs for BICM. A multi-rate and channel-adaptive inner-code architecture is also proposed. A tool is developed to optimize low-complexity rate- and channel-configurable concatenated FEC schemes via an MLC architecture. Compared to previously existing FEC schemes, up to 63% reduction in decoding complexity, or up to 0.6 dB coding gain is obtained. Code designs for MLC in combination with four-dimensional signal constellations are also considered. The design method is generalized to obtain complexity-optimized non-binary LDPC codes to concatenate with outer zipper codes. Gains of up to 1 dB over the conventional schemes are reported. The possibility of using a novel class of nonlinear codes in FEC design is also investigated.