Theoretical and experimental investigation of the impact of four wave mixing on dense wavelength division multiplexing at 2.5 Gbit/s

Wavelength division multiplexing techniques have proven in the last years to be the most attractive means to design ultra-high capacity networks through optical fibers. The immediate availability of optical amplifiers in the C and L bands will give rise to systems with high data rates. However, the increasing demand in capacity tends to decrease the channel spacings rates together with high power allocations and tends to use new types of optical fibers in order to reduce the nonlinear effects. Four wave mixing (FWM) is one of the major limitations for low channel spacings and hence, the low difference in the local chromatic dispersion between these channels. For a 50 GHz spacing and a high number of channels, the practical means to evaluate the FWM crosstalk is to drop the channel under test. In this paper, an analytical method is used, firstly to determine the minimum number of channels necessary to be representative of a multiplex with a large number of channels while keeping a manageable laboratory experiment. Secondly, it is used to identify the major influencing parameters of the link. An experimental approach is carried out to evaluate the impact of FWM on the system performance of a 16-channel transmission at 2.5 Gbit/s with a 50 GHz channel spacings over non-zero-dispersion-shifted optical fibre (NZDSF) with a zero dispersion wavelength at 1505 nm