On the effects of ASE noise models on turbo code decoder performance In optical fiber transmissions

Summary form only given. Turbo-codes (TC) have been shown to be the most powerful forward error correction (FEC) code achieving near Shannon limit performance. As FEC codes become a practical solution in improving system capacity in fiber communications, the application of TC in fiber transmissions begins to attract research interest. While the Gaussian approximation of the amplified spontaneous emission (ASE) noise is widely used in the study of fiber transmission systems, and there exist standard TC for Gaussian channels, we show that the Gaussian approximation underestimates the achievable performance of TC in real fiber transmission systems. To take the full advantage of TC, the more accurate noise statistics and binary asymmetric channel (BAC) model should be used for setting threshold and in the Bahl, Cocke, Jelinek, and Raviv (BCJR) decoding algorithm. In addition, for simplicity, the fiber channel with dominant ASE noise is usually approximated as a binary symmetric channel (BSC). We have shown that the BSC model overestimates the Shannon limit. We show that this approximation also leads to suboptimal threshold settings, and underestimates the performance of punctured TC.