Field-trial evaluation of the Q-factor penalty introduced by fiber four-wave mixing wavelength converters J.D. Marconi a, * , F.A. Callegari b , M.L.F. Abbade b , H.L. Fragnito a a Department of Quantum Electronics, Optics and Photonics Research Center, Unicamp – DEQ, IFGW, Campinas, SP 13083-970, Brazil b School of Electrical Engineering, PUC-Campinas, Rod. D. Pedro I- km 136, Campinas, SP 13086-900, Brazil article info Article history: Received 2 July 2008 Received in revised form 12 September 2008 Accepted 12 September 2008 Keywords: All-optical wavelength converters Wavelength division multiplexing All-optical networks Four-wave mixing abstract The performance of tunable all-optical wavelength converters based on four-wave mixing in optical fibers is experimentally tested in a field-trial network. Two converters were built with two different fibers. The first one was made with a small variation in the zero-dispersion wavelength (ZDW) dispersion shifted fiber and the second one with a highly nonlinear fiber that presents great ZDW variations. In order to compare the tuning ranges obtained in both cases we present an experimental spectral analysis. Numerical simulations that consider the influence of both the dispersion slope and the long-scale ZDW variations of the fiber complement the experiments. The tuning bandwidth was larger in the highly non- linear fiber case. For a set of different optical signal-to-noise ratios, the measurements of the Q-factor of the signal and those of the converted wave are our main results. These results show that the penalty imposed by the converters is different for each converted wavelength. The maximum penalty obtained for the Q-factor was 6 dB, but it was 63 dB for most cases. In all experiments we used a technique based on a dynamic polarization controller in order to avoid power fluctuations in the converted wave caused by polarization induced variations in the signal. Ó 2008 Elsevier B.V. All rights reserved. 1. Introduction All-optical wavelength converters (AOWC) are key elements for future wavelength division multiplexing (WDM) networks [1]. These devices are useful for several purposes such as dispersion compensation by optical phase conjugation [2,3], performing the wavelength compatibility of networks managed by enterprises that work with different channel grids [1,4], and reducing conges- tion by improving the capacity utilization of all-optical networks (AON) [5,6]. From the point of view of networking, this last feature is possibly the most challenging one and it is the motivation for the present work. There are three main techniques to implement all-optical wave- length conversions (WCs) [4,7]: optical gating, interferometry and wave mixing. In the first of them, a given device characteristic changes with the intensity of an input signal and transfers this change to a probe signal at a different wavelength. An example of this strategy is the cross-gain modulation AOWC observed in semiconductor optical amplifiers (SOAs) [8,9]. In the interferomet- ric approach, the variations of an input signal change the refractive index of a given device which in turn modulates the phase of the probe signal. This cross-phase modulation effect is further con- verted into intensity modulation by a Mach-Zehnder or other kind of interferometer and, in this way, the wavelength converted signal is generated [9–11]. Finally, in the wave mixing approach, an input signal with frequency f s is coupled to a cw pump with frequency f P in a nonlinear medium, which generates another signal in a new frequency f i . This nonlinear phenomenon can be generated by the second order susceptibility (v (2) ) as in the case of periodically poled lithium niobate (PPLN) waveguides [12,13] or by the third order susceptibility (v (3) ) as in the case of SOAs [14,15] and optical fibers [16,17]. In the PPLN case, when one pump is used, the new frequency is generated by the three wave mixing process at f i = f P  f s [12]. If two pumps are used it is possible to obtain a con- verted wave by two cascaded v (2) processes, at frequency f i = f s + f P1  f P2 (or f i = f P1 + f P2  f s ) [13]. In SOAs and optical fibers the process of four-wave mixing (FWM) takes place and creates a copy of the input signal at f i =2 f p  f s (f i =2 f s  f p ) [14–17]. The pros and cons of each of these approaches are found in [1,4,8]. Nevertheless, the wave mixing technique presents impor- tant advantages that should be enhanced here. First, the wave- length converted signal has approximately the same extinction ratio as the original one. Second, it does not depend on the modu- lation format; hence it may be applied to new modulation formats, like the differential phase-shift keying that improves the receiver sensitivity [18]. Finally, when accomplished in fibers or in PPLN waveguides, this technique is bit rate independent for any practical system because of the fast response of the virtual electronic tran- sitions [19,20]. However, the experimental efficiencies obtained 0030-4018/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2008.09.071 * Corresponding author. Tel.: +55 19 35215451; fax: +55 19 35215428. E-mail address: dmarconi@ifi.unicamp.com (J.D. Marconi). Optics Communications 282 (2009) 106–116 Contents lists available at ScienceDirect Optics Communications journal homepage: www.elsevier.com/locate/optcom