A method for determining most probable errors in nonlinear lightwave systems

We present a method for computing bit-error ratios in soliton-based lightwave systems. The method uses soliton perturbation theory and calculus of variations to find approximate versions of the most probable paths through sample space leading to errors, combined with importance-sampled Monte Carlo simulations of the full set of equations around these approximate paths to compute the actual error rates. As a specific example, the method is applied to a differential phase-shift-keyed lightwave system. For this example the method not only computes the bit-error ratio but also predicts the set of failure modes leading to large pulse distortions, thus illuminating the specific manner in which errors occur.