Performance of nonlinear amplitude regenerators in optical networks

We investigate theoretically and by simulations the performance of chains of amplitude regenerators. The combination of a transmission fiber followed by a regenerator is modelled by a two-step process: (i) the output amplitude y of a signal is obtained from its input value x by means of the relation y = g(x); (ii) noise is added as a random variable. We first investigate the performance of a single and multiple regenerator in the absence of noise. It is then possible to regenerate signals indefinitely without any degradation. However, with added noise, we find that signal quality must necessarily drop as the number of stages increases. Yet the increase is slow, and the BER values relatively low; hence such a chain of regenerators can outperform a chain of linear amplifiers under certain circumstances. An advantage of this approach is that it provides a clear visualization of the evolution of the noise pdf along the chain, which yields insights into the design of effective amplitude regenerators.