All-Optical Nonlinear Pulse Processing Based on ... - Aston University

1912

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO. 8, AUGUST 2004

All-Optical Nonlinear Pulse Processing Based on Normal Dispersion Fiber-Enhanced Nonlinear Optical Loop Mirror Sonia Boscolo and Sergei K. Turitsyn, Associate Member, IEEE

Abstract—A novel simple all-optical nonlinear pulse processing technique using loop mirror intensity filtering and nonlinear broadening in normal dispersion fiber is described. The pulse processor offers reamplification and cleaning up of the optical signals and phase margin improvement. The efficiency of the technique is demonstrated by application to 40-Gb/s return-to-zero optical data streams. Index Terms—All-optical nonlinear pulse processing, loop mirror intensity filtering, nonlinear pulse broadening.

I. INTRODUCTION

A

LL-OPTICAL data processing is expected to play a major role in future optical communications. Recently, the use of nonlinear optical loop mirrors (NOLMs) has been shown to be a promising technique for passive 2R regeneration (reamplification, reshaping) that may achieve virtually unlimited transmission in high-speed return-to-zero (RZ) transmission systems dominated by pulse distortion and amplitude noise [1], [2]. But the NOLM is insensitive to pulse position in time, and thus, it does not aid problems of timing jitter in a system. Also recently, the use of Kerr nonlinearity in a normal dispersion fiber (NDF) has been addressed as a technique to reduce the effect of timing jitter at an optical RZ receiver [3], [4]. In this letter, we present a novel simple scheme of all-optical nonlinear pulse processing that combines the loop mirror intensity filtering action for reamplification and cleaning up of the optical signals with the nonlinear pulse broadening in a NDF for phase margin improvement. The proposed scheme is suitable for use at an optical RZ receiver to improve the signal quality before detection. The efficiency of the proposed pulse processor is numerically demonstrated when the technique is applied to 40-Gb/s RZ data streams. II. OPERATION PRINCIPLE AND CONFIGURATION The proposed pulse processor consists of an optical amplifier, a section of NDF, and an unbalanced NOLM (Fig. 1). Qualitatively, the idea of the method is as follows. An input pulse to the pulse processor is amplified to the preferred power level of the device by the optical amplifier. During transmission along

Manuscript received January 30, 2004; revised April 6, 2004. The authors are with the Photonics Research Group at the School of Engineering and Applied Science, Aston University, Birmingham B4 7ET, U.K. (e-mail: [email protected]). Digital Object Identifier 10.1109/LPT.2004.829764

Fig. 1.

Schematic diagram of the pulse processor.

the NDF, the temporal waveform of the pulse changes to a rectangular-like profile by the combined action of group-velocity dispersion and Kerr nonlinearity [5]. As a result, the pulsewidth is broadened and the center portion of the pulse changes to be flat. By utilizing this property, the phase margin of an RZ data signal can be improved [3], [4]. The phase margin improvement enables reduction of the influence of the displacement of pulse position at the receiver caused by timing jitter. Following the NDF, the pulse enters the NOLM. The unbalanced NOLM acts as a saturable absorber [6] and, hence, filters out low-intensity noise and dispersive waves from the higher power pulse. This allows for restoration of the pulse amplitude and cleaning up of the distorted pulse. Also, whenever the NOLM operates in the region just after the peak of its switching curve, it enables stabilization of amplitude fluctuations. In the case of a pulse train, the noise and radiative background in the zero timing slots is suppressed by the loop mirror saturable absorption action, and the amplitude jitter of ones is also reduced [1], [2]. In the sample system used for demonstration of the technique, the amplifier is an erbium-doped fiber amplifier (EDFA) with a noise figure of 4.5 dB. The NDF has a dispersion of 20 ps/(nm km), an effective area of 30 m , and an attenuation of 0.24 dB/km. The NOLM incorporates a 50 : 50 coupler, and a 1.5-km-long loop of dispersion-shifted fiber with zero dispersion, an effective area of 25 m , and an attenuation of 0.3 dB/km. Note that the requirement of zero dispersion for the NOLM fiber is not critical because the short loop fiber length makes the effect of chromatic dispersion on pulse propagation in the loop negligible, whereas, self-phase modulation is the dominant effect. Unbalancing of the NOLM is achieved with an attenuator asymmetrically placed in the loop, and the loss of the loop attenuator is 27.1 dB. III. MODELLING RESULTS To demonstrate the efficiency of the proposed pulse processor, without loss of generality, the following model simpseudorandom RZ ulations are run. Transmitted are

1041-1135/04$20.00 © 2004 IEEE

BOSCOLO AND TURITSYN: ALL-OPTICAL NONLINEAR PULSE PROCESSING BASED ON NDF-ENHANCED NOLM

Fig. 3.

Fig. 2. Eye diagrams at (a) the pulse processor input, (b) the output of the conventional NOLM-based regenerator, (c) the NDF output, and (d) the pulse processor output.

single-channel pulse trains at 40 Gb/s in a dispersion-managed system whose transmission performance is severely degraded by intrachannel nonlinear effects when regenerators are not used (see [7] for details). In the considered transmission system, each amplifier span consists of 20-km-long effective core area enlarged positive dispersion fiber (EE-PDF) and 20-km-long NDF. The EE-PDF has a dispersion of 20 ps/(nm km), an effective area of 110 m , and an attenuation of 0.2 dB/km. The NDF has identical parameters to those of the NDF incorporated in the proposed pulse processor. Each span also includes an EDFA that compensates for the energy losses, and a Gaussian filter that limits the bandwidth of the noise. An identical NOLM to the one used in the proposed pulse processing scheme is placed into the transmission line every ten amplifier spans. An extra gain is added to the amplifier prior to the NOLM to achieve power equalization. The periodical deployment in-line of NOLMs in the system effectively stabilizes the accumulation of pulse distortion and amplitude noise driven mainly by the intrachannel four-wave mixing, and the transmission distance is limited by intrachannel cross-phase modulation-induced timing jitter [7]. This presents a good model situation to demonstrate the action of the proposed device. The pulses after 20 000-km transmission are used as the input for the pulse processor (see Figs. 1 and 2). The input pulse energy and full-width at half-maximum (FWHM) pulsewidth are approximately 0.011 pJ and 7 ps, respectively. Such parameters are the width and energy that the pulses periodically recover in the transmission system after travelling the distance between two consecutive NOLMs. The signal quality is evaluated in terms of the standard (Gaussian-based) factor. Note that the factor is used here as a measure of the quality of the signal eye rather than a measure of the bit-error rate (BER). Indeed, in systems employing optical devices with an inherently nonlinear response, the error source usually has a substantially non-Gaussian statistics, and thus, predictions of the BER based on the factor method can yield incorrect results. A fifth-order

1913

Q values versus length of the NDF.

Bessel filter with a cutoff frequency of 30 GHz is used as a receiver low-pass filter. Fig. 2 shows an example of eye diagrams at the pulse processor input, the NDF output, and the pulse processor output. The output eye pattern for the conventional NOLM-based regenerator (without NDF) is also shown. In this example, the gain of the EDFA is 33.5 dB, and the length of the NDF is 0.5 km. It can be seen that the input eye is closed mainly due to a significant timing jitter of the optical pulses. The conventional NOLM-based regenerator does not ensure timing jitter reduction. On the other hand, when the proposed pulse processing scheme is employed, the pulse duration is broadened by utilizing dispersion and nonlinearity in the NDF (see also Fig. 1). In the example of Fig. 2, the FWHM pulsewidth is broadened to approximately 25 ps. Consequently, the eye opening after propagation in the NDF is wider than that at the pulse processor input. It can also be seen that the amplitude jitter of pulses at the center of the bit slot is slightly smaller. The additional widening of eye opening that can be observed at the pulse processor output is given by a significant reduction of the amplitude jitter provided by the NOLM. We now investigate the impact of intrinsic device parameters on the pulse processor performance. A key parameter to be tuned is the length of the NDF. In Fig. 3, the dependence of the factor is shown at the NDF output and the pulse on the NDF length. The factor at the processor output is also shown for reference. For each pulse processor input value of the NDF length, the gain of the EDFA is adjusted so as to provide both adequate power at the NOLM input and adequate enhancement of the nonlinearity in the NDF. It is seen that the pulse processor works with NDF lengths between 0.2 and 0.6 km, and for the optimum length of 0.5 km it improves the signal quality by 3.8 dB (corresponding to a linear improvement for lengths shorter factor of 2.4). The decrease of than the optimum one is due to less pulse broadening and flattening by nonlinearity and dispersion in the NDF. For lengths longer than the optimum one, the pulsewidth after propagation in the NDF is broadened appreciably beyond the timing slot, . The and therefore, this also results in a decrease of variation of with the NDF length indicates that the range of

1914

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 16, NO. 8, AUGUST 2004

corresponds to a pulse energy at the NOLM input of approximately 25.1 pJ. For gain deviations of 0.3 and 0.5 dB from the optimum value, which corresponds to variations of 6.4% and 10.4% in the pulse energy at the NOLM input, the factor at the pulse processor output is at least 3 dB higher than the factor at the pulse processor input. IV. CONCLUSION

Fig. 4.

Q values versus gain of the amplifier.

allowed NDF lengths is mainly determined by the NOLM. For NDF lengths shorter–longer than the optimum one, the power level of pulses at the NDF output is greater–smaller than the correct power level for the NOLM to operate in the region just after the switching peak. The correct power level at the NOLM input could be achieved by reduction–increase of the EDFA gain. But a lower–higher gain of the EDFA would diminish–enhance further the effect of nonlinearity in the NDF, and therefore, such an adjustment would result in a significant deterioration of the signal quality at the NDF output. Note that the nonlinearities in the NDF and in the NOLM could be easily decoupled by simply inserting an optical attenuator or amplifier between NDF and NOLM, which would permit us to adjust both power levels at the NDF input and at the NOLM input independently. This would widen the tolerance of the device parameters. It is also important to investigate the tolerance of the pulse processor to variations in the gain of the EDFA for a fixed NDF length. In Fig. 4, the factor is shown at the NDF output and the pulse processor output as a function of the amplifier gain, when the length of the NDF is set to its optimum value. It can factor at the NDF output varies very little be seen that the with the amplifier gain in the considered range of gains. But the factor at the pulse processor output does vary appreciably because of the sensitiveness of the NOLM to the power level at its input. The optimum value of the amplifier gain is 33.5 dB, which

We have described a simple all-optical nonlinear pulse processing technique for use at an optical RZ receiver that exploits the intensity filtering action of an NOLM for achieving reamplification and cleaning up of the optical signals and the Kerr effect in a NDF for improvement of the signal phase margin. The efficiency of the proposed pulse processor has been demonstrated by its application to timing jitter-degraded 40-Gb/s single-channel RZ pulse trains. We emphasize that the proposed design of enhanced NOLM does not depend on the particular transmission scheme to which the pulse processor is applied, and has a broad range of possible applications in amplitude noise and timing jitter-limited optical systems. In particular, it can be used in wavelength-division-multiplexed systems by applying the pulse processor after demultiplexing. REFERENCES [1] S. Boscolo, S. K. Turitsyn, and K. J. Blow, “Study of the operating regime for all-optical passive 2R regeneration of dispersion-managed RZ data at 40 Gb/s using in-line NOLMs,” IEEE Photon. Technol. Lett., vol. 14, pp. 30–32, Jan. 2002. [2] , “All-optical passive 2R regeneration for 40 Gbit/s WDM transmission using NOLM and novel filtering technique,” Opt. Commun., vol. 217, pp. 227–232, 2003. [3] M. Suzuki, H. Toda, A. H. Liang, and A. Hasegawa, “Improvement of amplitude and phase margins in an RZ optical receiver using Kerr nonlinearity in normal dispersion fiber,” IEEE Photon. Technol. Lett., vol. 13, pp. 1248–1250, Nov. 2001. [4] M. Suzuki and H. Toda, “ -factor improvement in a jitter limited optical RZ system using nonlinearity of normal dispersion fiber placed at receiver,” in Tech. Dig. Optical Fiber Communications (OFC), Anaheim, CA, 2001, Paper WH3. [5] H. Nakatsuka, D. Grischkowsky, and A. C. Balant, “Nonlinear picosecond-pulse propagating through optical fibers with positive group velocity dispersion,” Phys. Rev. Lett., vol. 47, pp. 910–913, 1981. [6] N. J. Smith and N. J. Doran, “Picosecond soliton transmission using concatenated nonlinear optical loop-mirror intensity filters,” J. Opt. Soc. Amer. B, vol. 12, pp. 1117–1125, 1995. [7] S. Boscolo, S. K. Turitsyn, and K. J. Blow, “All-optical passive quasi-regeneration in transoceanic 40 Gbit/s return-to-zero transmission systems with strong dispersion management,” Opt. Commun., vol. 205, pp. 277–280, 2002.

N 2

Q