2328 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 16, NO. 12, DECEMBER 1998 Path Average Measurements of Optical Fiber Nonlinearity Using Solitons John K. Andersen, Student Member, OSA, Janet W. Lou, Student Member, IEEE, George A. Nowak, Student Member, OSA, Tiejun Xia, Member, OSA, Mohammed N. Islam, Fellow, OSA, Rance M. Fortenberry, Member, IEEE, and Steve A. Newton, Member, OSA, Abstract— This paper experimentally demonstrates a new method to determine the optical nonlinearity of single-mode optical ﬁber. The technique takes advantage of the well-known nonlinear response of optical ﬁbers and well-developed models for soliton pulse propagation to extract information about the ﬁber characteristics. Fiber nonlinearity can degrade the performance of communication systems by, for example, causing crosstalk and signal distortions. Measuring the ﬁber nonlinearity would greatly aid system designers in building and upgrading communication systems. The method is utilized to determine values for where is the nonlinearity of the glass and is effective area of the core, on various lengths of Corning SMF-28 ﬁber and Corning SMF-DS ﬁber. Experimentally measured propagation results for short ( 2 ps) optical pulses are compared to computer simulated models to determine the ﬁber nonlinearity. The method ﬁnds W values for short lengths ( 400 m) of Corning SMF-28 ﬁber and values of 2.7 10 10 W 1 for longer lengths ( 6.5 km and 20 km). The difference is expected due to the 8/9 polarization scrambling factor, and the values are in agreement with reported literature [1]. The method also determines W for a 12 km Corning dispersion shifted ﬁber. The method has two major regimes of operation based on the soliton period, a characteristic length for solitons. For few soliton periods the output phase is measured as a function of launched power; for many soliton periods the output pulsewidth is measured as a function of launched power. The method’s major advantage is its capability to measure long lengths of standard ﬁber, where it uses only standard diagnostic tools such as autocorrelation and optical power measurements. However, the method is only applicable in the soliton regime of ﬁbers. Index Terms— Dispersion, optical communications, nonlinear- ity, optical ﬁber. I. INTRODUCTION I NCREASING the information throughput of the currently deployed optical ﬁber transmission lines is of great interest. Either increasing the bit rates or adding additional wavelength channels represents the most cost-effective way of upgrading current systems. Also, longer distances between ampliﬁers are Manuscript received August 29, 1997; revised June 15, 1998. This work was supported by DARPA. J. K. Andersen, J. W. Lou, G. A. Nowak, T. Xia, and M. N. Islam are with The University of Michigan, Ann Arbor, MI 48109 USA. R. M. Fortenberry was with Hewlett-Packard Laboratories, Palo Alto, CA 94304 USA. He is now with Hewlett-Packard Lightwave Division, Santa Rosa, CA 95403 USA. S. A. Newton is with Hewlett-Packard Laboratories, Palo Alto, CA 94304 USA. Publisher Item Identiﬁer S 0733-8724(98)09302-5. important in lowering costs and increasing system reliability. However, these techniques to upgrade existing communica- tion systems increase the necessary launched power into the transmission ﬁber. Whereas traditionally the major limitation to ﬁber networks was ﬁber dispersion, today the increased launched optical powers are causing ﬁber nonlinearity to become a signiﬁcant source of bit error rate degradation. Fiber nonlinearity can generate crosstalk between channels through four wave mixing and cause signal distortion through self- phase modulation (SPM). Thus, in building and maintaining these communication networks, it is important to accurately characterize the nonlinearity in the optical ﬁber. Recently, a number of different techniques to measure ﬁber nonlinearity have been explored in the literature [2]–[7]. Each examines the result of a different nonlinear effect. One technique measures side band generation due to SPM on a pulsed source [5]. Another technique examines phase changes on one optical beam due to cross phase modulation from another optical beam [2]. Finally, a third technique measures the beat frequency components created through four wave mixing [7]. Each method has disadvantages either in complexity of the experimental setup or limits on the maximum ﬁber length or on the maximum ﬁber dispersion that can be tested. The limitations on the length of the ﬁber or dispersion of the ﬁber are based on the premise that dispersion does not play a signiﬁcant role in shaping the signal or cause walk-off between two signals. Consequently, these methods are limited to a fraction of a dispersive length. This limitation is on the order of hundreds of meters. For example, for a 10 ps pulse at 1550 nm, the dispersive length is 4.8 km in standard ﬁber, and to prevent signal distortion or walk-off between two signals, the propagation distance would need to be even less, perhaps one-tenth of a dispersive length or 480 m. The technique described here is of particular value in measuring long lengths (i.e., many dispersive lengths) of standard ﬁber or dispersion shifted ﬁber. In communication systems, a distributed measurement, one that measures the averaged ﬁber nonlinearity over the entire length of ﬁber, is preferred. This would be advantageous in characterizing already deployed ﬁbers. However, to measure the path av- eraged ﬁber nonlinearity, the signal should not undergo major distortions along the ﬁber length, due to either the ﬁber nonlinearity or the ﬁber dispersion. Optical solitons use the ﬁber nonlinearity to balance the ﬁber dispersion, thereby maintaining a constant signal over long ﬁber lengths. Solitons 0733–8724/98$10.00  1998 IEEE