NonGaussian contributions to the bit error rate in a soliton communication system

Abstract

In modern-day, long-distance optical fiber Communication systems, one aspires to achieve hit error rates that are lower than 10⁻⁹ 10⁻¹². Theoretical calculations to date, both analytical and computational, of the bit error rate typically rely on the suspect assumption that the quantities of interest are Gaussian-distributed. In fact, fiber nonlinearity due to the Kerr effect, coupled with spontaneous emission noise from erbium-doped amplifiers, can lead to nonGaussian distributions of the quantities of interest. The consequences for determining the bit error rate in communication systems is profound. With the assumption that the distribution function is Gaussian, one can immediately determine the bit error rate if the variances and threshold levels of the quantities of interest are known.