ANALYSIS OF FOUR-WAVE MIXING IN OPTICAL FIBER LINKS WITH NON- UNIFORM CHROMATIC DISPERSION J. R. Souza 1 and P. B. Harboe 2 1 Center for Telecommunication Studies Pontifical Catholic University of Rio de Janeiro Rua Marque ˆ s de Sa ˜ o Vicente 225, Rio de Janeiro–RJ, CEP 22451-041, Brazil 2 Telecommunication Engineering Department Federal Fluminense University Rua Passo da Pa ´ tria 156, Nitero ´ i–RJ, CEP 24210-240, Brazil Received 24 March 2003 ABSTRACT: The four-wave mixing (FWM) process in single-mode optical fibers is investigated by considering that a zero-dispersion wavelength var- ies along the length of the fiber link, which is represented as a series of segments with different lengths, propagation constants, and zero-dispersion wavelengths. The results indicate that new optical frequencies are efficiently generated for particular combinations of the input-signal wavelengths, even when the phase-matching condition is not fully met. © 2003 Wiley Periodi- cals, Inc. Microwave Opt Technol Lett 39: 102–105, 2003; Published on- line in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ mop.11139 Key words: nonlinear fiber optics; four-wave mixing; chromatic disper- sion; WDM systems 1. INTRODUCTION Very high-capacity, long-haul optical communication systems are made possible by the extremely wide bandwidth of optical fibers, which is best exploited by wavelength-division multiplexing (WDM) [1]. The performance of long-distance optical communi- cation systems is limited, however, by chromatic dispersion and nonlinear effects of fiber, which interact and accumulate along the length of the optical link. Chromatic dispersion, which broadens the pulses, can be reduced by using dispersion-shifted fibers at the 1550-nm wavelength range. But low chromatic dispersion en- hances some nonlinear effects of fiber, especially four-wave mix- ing (FWM), as the phase mismatch due to dispersion is also reduced [2]. Four-wave mixing is a nonlinear process associated with the third-order electric susceptibility (3) of the optical fiber, where any three frequencies f i , f j , and f k interact through (3) to generate a new optical wave of frequency f ijk = f i + f j - f k [2, 3]. In WDM systems, every three channels will originate, nine new optical frequencies, due to FWM. Whenever these new frequencies are generated within the system bandwidth and coincide with one or more channel frequencies, the system performance is degraded by the resulting cross talk. Also, the power of the signal carriers is reduced in the FWM process, contributing to further performance degradation. FWM is then one of the major limiting factors in long-haul WDM systems. The effects of FWM on WDM systems can be reduced either with the introduction of dispersion into the system, or by maximizing the channel separation. This paper addresses the effect of the FWM phenomenon on two-channel systems operating close to the zero-dispersion wave- length of the optical fiber. We investigate the influence of both the variation of the dispersion along the fiber link, which can be seen as a form of dispersion management, and the channel separation. The analysis is based on the undepleted pump model, according to which the signal power is not diminished by the generation of FWM frequencies [2]. The paper is organized as follows: the next section introduces the basic formulation for the analysis of the four-wave mixing process in segmented optical-fiber links. In section 3, results are presented for the efficiency of FWM gener- ation, and also for the ratio between the FWM and signal powers at the end of the fiber link. Conclusions are drawn in section 4. 2. FORMULATION In a real optical-fiber link, the zero-dispersion wavelength may vary along the link length, due to the perturbations introduced during the fabrication process. To circumvent the difficulty of treating this situation in an exact form, the fiber link is represented by a series of segments of length L i , zero-dispersion wavelength 0i , and propagation constant i ( i = 1, 2, . . . , N), as illustrated in Figure 1. This same figure can also illustrate a dispersion- managed link, where spans of fiber with different chromatic dis- persion are used to introduce a low net dispersion in the system, and thus reduce the efficiency of four-wave mixing. The FWM analysis is carried out for each of the segments in Figure 1, subjected to the appropriate boundary conditions [4]. The amplitudes of the optical carriers are usually much higher than those of the new optical waves generated by FWM. So, in a first-order approximation, it is assumed that the former are not affected by the FWM process. The fundamental equation describing the FWM process in each of the N fiber segments in Figure 1 is written as [4 – 6]: d dz E F i z + 2 E F i z = j n 2 F cA eff D 3 E p i z i-1 E q i z i-1 E r i *z i-1 exp  - 3 2 + j i z - z i-1 (1) where i = 1, 2, . . . , N; z is the distance along the fiber link, E F ( z ) is the FWM field component generated at the frequency f F = f p + f q - f r = F /2, E s (0) ( s = p, q, r ) is the signal amplitude at frequency f s at z = 0, is the loss coefficient of the fiber, D is the FWM degeneracy factor ( D = 3 for p = q r , and D = 6 for p q r ), n 2 is the fiber nonlinear refractive index, c is the speed of light in vacuum, A eff is the modal effective area of the fiber, i = i ( f p ) + i ( f q ) - i ( f r ) - i ( f F ) is the phase mismatch. In addition, z i = ¥ k=1 i L k , and L k is the length of the k th fiber segment, z 0 = 0. The boundary conditions at the interface between two fiber segments, at z = z i , are described as [4]: E F i z i expj i F L i = E F i+1 z i , E s i z i-1 exp  - 2 + j i s L i = E s i+1 z i , s = p, q, r, (2) with i = 1, 2, . . . , N - 1, and E F 1 (0) = 0. For each segment, the solution of Eq. (1) at z = z i is calculated as Figure 1 Optical-fiber link composed by segments of different zero- dispersion wavelengths 102 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 39, No. 2, October 20 2003