Electronic signal processing techniques have assumed a prominent role in the design of fibre-optic communication systems. However, state-of-the-art systems operate at per-channel data rates of 100 Gb/s, which constrains the class of communication algorithms that can be practically implemented. Relative to LDPC-like codes, product-like codes with syndrome-based decoding have decoder dataflow requirements that are smaller by more than two orders of magnitude, which strongly motivates the search for powerful product-like codes. This thesis presents a new class of high-rate binary error-correcting codes, staircase codes, whose construction combines ideas from convolutional and block coding. A G.709-compliant staircase code is proposed, and FPGA-based simulation results show that performance within 0.5 dB of the Shannon Limit is attained for bit-error-rates below 1E-15. An error-floor analysis technique is presented, and the G.709-compliant staircase code is shown to have an error floor below 1E-20. Using staircase codes, a pragmatic approach for coded modulation in fibre-optic communication systems is presented that provides reliable communications to within 1 bit/s/Hz of the capacity of a QAM-modulated system modeled via the generalized non-linear Schrodinger equation. A system model for a real-world DQPSK receiver with correlated bit-errors is presented, along with an analysis technique to estimate the resulting error floor for the G.709- compliant staircase code. By applying a time-varying pseudorandom interleaver of size 2040 to the output of the encoder, the error floor of the resulting system is shown to be less than 1E-20.

Electronic signal processing techniques have assumed a prominent role in the design of fibre-optic communication systems. However, state-of-the-art systems operate at per-channel data rates of 100 Gb/s, which constrains the class of communication algorithms that can be practically implemented. Relative to LDPC-like codes, product-like codes with syndrome-based decoding have decoder dataflow requirements that are smaller by more than two orders of magnitude, which strongly motivates the search for powerful product-like codes. This thesis presents a new class of high-rate binary error-correcting codes, staircase codes, whose construction combines ideas from convolutional and block coding. A G.709-compliant staircase code is proposed, and FPGA-based simulation results show that performance within 0.5 dB of the Shannon Limit is attained for bit-error-rates below 1E-15. An error-floor analysis technique is presented, and the G.709-compliant staircase code is shown to have an error floor below 1E-20. Using staircase codes, a pragmatic approach for coded modulation in fibre-optic communication systems is presented that provides reliable communications to within 1 bit/s/Hz of the capacity of a QAM-modulated system modeled via the generalized non-linear Schrodinger equation. A system model for a real-world DQPSK receiver with correlated bit-errors is presented, along with an analysis technique to estimate the resulting error floor for the G.709- compliant staircase code. By applying a time-varying pseudorandom interleaver of size 2040 to the output of the encoder, the error floor of the resulting system is shown to be less than 1E-20.

This thesis presents a new low complexity erasure decoder for staircase codes in optical interconnects between data centers. We developed a parallel software simulation environment to measure the performance of the erasure decoding techniques at output error rates relevant to an optical link. Low complexity erasure decoding demonstrated a 0.06dB increase in coding gain when compared to bounded distance decoding at an output error rate of 3 × 10⁻¹². Further, a log-linear extrapolation predicts a gain of 0.09dB at 10⁻¹⁵. This performance improvement is achieved without an increase in the maximum number of decoding iteration and keeping power constant. In addition, we found the optimal position within the decoding window to apply erasure decoding to minimize iteration count and output error rates, as well as the erasure threshold that minimizes the iteration count subject to the constrained erasure decoding structure.