On the achievable information rate of asymmetric optical fiber channels with amplifier spontaneous emission noise

This work is motivated by the need to understand the fundamental limit of channel capacities of optical fiber communication channels where amplified spontaneous emission (ASE) noise is dominant and to understand how much has been achieved by the state-of-the-art forward error correction (FEC) coding techniques. Two most commonly used channel models, the Chi-square channel and its Gaussian approximation, are examined. Both soft receiver with continuous output and hard receiver with 1-bit and 2-bit quantized output are investigated for several values of the system parameter M, which is the number of signal photons at the input of the ideal high gain optical amplifier that produces the noise. We show how bit quantization affects the channel capacity and how the capacity scales with different values of M. Finally, we report the performance of some of the best-known codes like turbo codes on optical fiber communication channels.