Irregular Polar Coding for Complexity-Constrained Lightwave Systems

Next-generation fiber-optic communications call for ultra-reliable forward error correction (FEC) codes that are capable of low-power and low-latency decoding. In this paper, we propose a new class of polar codes, whose polarization units are irregularly pruned to reduce computational complexity and decoding latency without sacrificing error correction performance. We then experimentally demonstrate that the proposed irregular polar codes can outperform state-of-the-art LDPC codes, while decoding complexity and latency can be reduced by at least 30% and 70%, respectively, versus regular polar codes, while also obtaining a marginal performance improvement.