Analysis of Soft-Decision FEC on Non-AWGN Channels

Circularly symmetric, complex, additive white Gaussian noise (AWGN) is the most widely used model to analyze the behavior of communication systems owing to its good match with many practical channels and its mathematical tractability. In particular, soft-decision forward error correction (SD-FEC) schemes are typically designed for such channels. In the presence of AWGN, the received signal is composed of several circular scattergrams in the constellation diagram. In a fiber-optic communication system, however, these circularly symmetric Gaussian noise clouds can be distorted to, e.g., elliptic shapes by nonlinear signal-to-signal and signal-to-noise interactions. It is natural then to ask a fundamental question about how far the actual performance of SD-FEC deviates from the one predicted under the AWGN assumption. An answer to this question is of particular importance since optical transmission system characterization frequently measures the pre-FEC bit error ratio (BER), and infers post-FEC system performance based on the FEC¡¯s error correction capabilities in the presence of AWGN. In this paper, we model the fiber-optic channel as a channel including additive bivariate Gaussian noise (AWBGN). We analyze the impact of correlations among the signal¡¯s two quadrature components, and assess the effect of AWBGN on SD-FEC performance using the density evolution of low-density parity-check (LDPC) codes. Our analysis and simulations give insights into potential improvements of the etection performance for fiber-optic transmission systems assisted by SD-FEC.