An experiment of all-optical regeneration of short-pulse differential phase-shift keying (DPSK) signals using fiber nonlinearity is reported. Bit error rate (BER) performance is measured for a two-span transmission system where the regenerator is inserted between the spans. Two cases are examined where the signal degradation before the regenerator is due mainly to nonlinearity of the transmission fiber, i.e., the nonlinear phase noise, and is due to addition of amplified spontaneous emission (ASE). The regenerator is shown to be more effective in recovering signal quality in the former case of degradation due to phase noise.
© 2009 Optical Society of America
All-optical signal regeneration/regularization is an efficient method to extend reach of highspeed optical signal transmission without relying on optical/electronic/optical conversion and signal processing in the electric domain. A number of studies on the all-optical regenerator have been reported aiming at higher-speed and robust operation with simpler composition . Currently efforts are being devoted to realization of multichannel all-optical signal regeneration using a single nonlinear device for cost effective use in wavelength-division multiplexed systems [2–4]. Another important issue is the application of all-optical regeneration to signals in advanced modulation formats. Differential phase-shift keying (DPSK) and differential quadrature phase-shift keying (DQPSK) formats, for example, have been shown to have advantageous features such as high receiver sensitivity, large tolerance to transmission-fiber nonlinearity, and, in the case of DQPSK, higher spectral efficiency. They are being introduced in commercial systems. All-optical regenerators that can remove noise on such signals will be highly desired in future high-speed large-scale photonic networks.
Recently several schemes of DPSK signal regeneration have been proposed and demonstrated [5–12]. In , we demonstrated a back-to-back short-pulse DPSK signal regeneration experiment using a scheme where the noise reduction is performed on the demodulated on-off keying (OOK) signals by a fiber-based all-optical amplitude regenerator and the amplitude information is transferred back to the phase by a subsequent all-optical phase modulator. The regenerator is an integrated version of 2R (reamplification and reshaping) regenerative conversion from DPSK to OOK formats [13,14] and all-optical format conversion from OOK to binary PSK (BPSK) formats using cross-phase modulation (XPM) in a nonlinear fiber .
In this paper we report an experiment where the regenerator is inserted in a two-span transmission system. Two cases are examined where the signal degradation before the regenerator is due mainly to nonlinearity of the transmission fiber and due to addition of amplified spontaneous emission (ASE) from inline optical amplifiers. The effectiveness of the DPSK regenerator is compared between the two different signal impairments before the regenerator.
2. All-optical DPSK signal regenerator
Figure 1 shows a setup of the DPSK signal regenerator. An incoming DPSK signal is demodulated to an OOK signal by a one-bit delay interferometer (DI), by which the phase variation including noise is transferred to amplitude variation. A 2R amplitude regenerator subsequently suppresses the amplitude noise both in space and mark bit slots of the OOK signal. In the setup shown in Fig. 1, a two-stage cascaded fiber-based 2R regenerator in bidirectional configuration [12, 16] is used. The amplitude-stabilized data pulses are then amplified and fed to an all-optical phase modulator where the data pulses modulate the phase of clock pulses generated by a mode-locked semiconductor laser diode (MLLD) acting as a local pulse source. The radio-frequency (RF) clock tone is extracted from the incoming signal. The all-optical phase modulation is based on XPM between the clock and control pulses.
Strength of amplitude noise suppression required for the regenerator can be estimated as follows: First we assume that the incoming pulses have a complex amplitude of the form Ein n = (As+ΔAn) exp[i(ϕn+Δϕn)] where As and ϕn ϕn-1 ϕn-1 = 0 or π) are an amplitude and phase of the pulse, respectively, and ΔAn and Δϕn are amplitude and phase fluctuations of the pulse. The complex amplitude of the pulse at the output port of the DI is given by EDI = (Ein n - Ein n-1) /2, and its power is calculated to be
in the first-order approximation under the conditions ∣ΔAn-1,n∣ << As and ∣Δϕn-1,n∣ << 1. Eq. (1) shows that the phase noise in the input signal is not transferred to the output signal power from the DI in the first-order approximation. This is due to the general behavior of interferometers that the output power is insensitive to the phase fluctuations when the phase difference is close to 0 or π. Larger phase fluctuations exceeding the range of the first-order approximation produce fluctuations in the output optical power from the DI, which causes decision errors. This is true irrespective of where the DI is located: in the receiver or in the regenerator. This indicates that the DPSK signal regenerator discussed in this paper is effective in suppressing accumulation of phase noise when it is inserted in the system before the phase noise of the signal grows significantly.
Then we consider the case of π phase difference between the pulses in Eq. (1). After the power fluctuation in ∣EDI∣2 is reduced with a factor of r (r<1) by the amplitude regenerator, the pulse is amplified and used as a control pulse in the subsequent all-optical phase modulator. When we assume that the phase modulation of the clock pulse is proportional to the power of the control pulse, the complex amplitude of the output pulse is expressed as
where Aclock is an amplitude of the clock pulse. For the output pulse to be in BPSK format, the gain of the amplification of the control pulse G should satisfy GA2 s = π. Then the phase fluctuation in Eq. (2) is given by Δϕout = rπ(ΔAn+ΔAn-1)/As and its variance is σ2 out = 2r2π2σ2 Ain/As, where σ2 Ain is the variance of the amplitude fluctuation of the input pulses. Here no correlation between amplitude fluctuations of neighboring input pulses is assumed. When the input signal is degraded by a circular Gaussian noise such as ASE, σ2 Ain = A2 sσ2 ϕin is satisfied. The phase noises in the output and input signals are then related by σ2 ϕout = 2r2π2ϕ2 ϕin. In this case, in order for the output phase noise to be smaller than the input phase noise, we need to use an amplitude regenerator with r smaller than (21/2π)-1 or the noise suppression factor 1/r larger than 10log10(21/2π) = 6.5dB. This factor is derived also in  with a different approach.
In the first-order analysis given above, no output appears from the DI when ϕn - ϕn-1 = 0 as shown in Eq. (1). In reality, input noise both in amplitude and phase produces small output even in the condition of destructive interference. The 2R amplitude regenerator after the DI should, therefore, have the function of noise suppression also on the space level.
It is noted that the DPSK regenerator studied in this paper maps the phase difference between adjacent pulses incoming to the regenerator to the absolute phase of the output pulses, which accompanies conversion of logic encoded on the signal phase. The logic conversion can be reversed either by an encoder in the transmitter or by a decoder in the receiver in the electronic domain [9, 10]. In reconfigurable and/or burst/packet networks, the number of regenerators that the signal passes may vary according to the route of the signal. In such a case, preservation of the data logic within the regenerator is desired. This will be achieved by inserting an all-optical exclusive OR gate such as one demonstrated in  with one-bit delay feedback in the regenerator where the signal stays in the OOK format after the DI.
3. Experimental setup
Figure 2 shows an arrangement of transmission fibers and the regenerator in the performance measurement. The pulse source in the transmitter consists of an actively mode-locked fiber ring laser (FRL) and a continuous-wave laser. XPM between them in a nonlinear fiber and subsequent narrowband filtering produce a phase-stable pulse train at 1548.5nm. The pulse width and the repetition rate are 6ps and 10GHz, respectively. The duty ratio of the pulses (6 percent) is much smaller than those (33, 50, or 67 percent) usually used in long-distance transmission. Short pulses are used in this experiment mainly because the fiber-based 2R amplitude regenerator works better for shorter pulses. By suitable scaling of the fiber parameters such as length and dispersion and pulse power, the regenerator will work at duty ratios up to ~ 30 percent . Still wider RZ-DPSK signals would be regenerated by placing a pulse compression stage prior to the 2R amplitude regenerator.
The pulses are then phase-modulated by a LiNbO3 phase modulator with a 256-bit random pattern. After amplification the pulses are launched to the first transmission fiber. The fiber is a densely dispersion-managed (DDM) fiber consisting of alternating normal- and anomalous-dispersion (~ ±3ps/nm/km) non-zero dispersion-shifted fiber sections with zero average dispersion around the signal wavelength. Length of each fiber section is 2km and the total length is 40km . In this fiber, dispersive pulse broadening is limited, which enhances the nonlinear phase noise that is caused by the translation from amplitude to phase noise via the effect of self-phase modulation (SPM) in the fiber. Similar transmission behavior is expected also when a dispersion-shifted fiber is used instead of the DDM fiber. The loss of the DDM fiber including splice loss is 13.7dB. After the transmission over the DDM fiber, an attenuator (ATT1) together with an erbium-doped fiber amplifier (EDFA) is inserted for the purpose of noise loading. The second fiber after the regenerator is a standard single-mode fiber (SMF) with 50km length that is fully dispersion compensated by a dispersion compensating fiber (DCF), total loss of which is 15.9dB. Again ASE is loaded by a combination of an attenuator (ATT2) and an EDFA. The receiver consists of a preamplifier, an optical bandpass filter (OBPF), a DI, a balanced detector, an RF amplifier, and a lowpass filter, followed by an error detector. Different programmed bit patterns are used for the error count when the regenerator is or is not inserted. No decoder is used. The first highly nonlinear fiber (HNLF1 in Fig. 1) in the regenerator for the 2R regeneration has a zero-dispersion wavelength λ0=1560nm, a dispersion slope dD/dλ = 0.03ps/nm2/km, a length L=1.8km, and a nonlinearity coefficient γ ~12/W/km. The filter offset is 2.5nm for both forward and backward directions. The direction of the wavelength shift is opposite so that the output wavelength of the bidirectional 2R amplitude regenerator is the same as that of the input signal. Bandwidth of the OBPFs is 1nm. The amplitude-regenerated data pulses at 1548.5nm together with the clock pulses derived from the MLLD at 1553nm are fed into the second HNLF (HNLF2) acting as an all-optical phase modulator. The clock pulses have a duration of 1.5ps before entering HNLF2, but are widened to 6ps after the OBPF with bandwidth of 0.8nm for the rejection of the control pulses. HNLF2 has dispersion D=2.2ps/nm/km and L=2.4km. Walk-off time between the data and clock pulses is 24ps and the timing between the two pulse trains is adjusted by a variable delay line so that complete walk-through between the control and probe pulses takes place in the fiber. XPM-based phase modulation without walk-off between control and probe pulses in a fiber with nearly-zero dispersion can be employed. This design, however, will give temporally non-uniform phase modulation, or chirp, to the probe pulse reflecting the temporal pulse shape of the control pulse. In order to avoid this impairment, we used the walk-through design for the XPM-based phase modulation .
In this experiment, the wavelength of the signal exiting from the DPSK regenerator is shifted by 4.5nm from that of the input signal. This is mainly caused by that the best performance of the bidirectional 2R amplitude regenerator is obtained when the signal wavelength is shifted back to the original signal wavelength at its second stage as mentioned above. Further optimization of the 2R amplitude regenerator, in terms of signal wavelength arrangement and parameters of the HNLF, will enable wavelength shifts in the same direction in the two-stage operation. Then after the XPM-based phase modulation, the output PSK signal can have the same wavelength as the input DPSK signal.
First we consider the case where the signal before the regenerator is degraded by nonlinearity in the preceding transmission. ATT1 in Fig. 2 is set at zero and the average signal power Ps launched to the DDM fiber is varied. At signal power levels larger than about 7dBm, degradation caused by the nonlinear phase noise appears. The optical signal to noise ratio at the entrance of the transmission fiber is 20dB/0.1nm noise bandwidth. Figure 3(a) shows the bit error rate (BER) performance measured after the DDM fiber with or without inserting the regenerator. The dotted curve is the reference back-to-back BER. When the regenerator is not used the BER degrades steadily as the launched signal power grows larger than about 8dBm as shown by dashed curves. BERs after regeneration are shown by solid curves in Fig. 3(a). The effect of regeneration is evident at low received power Prec < -36dBm, where BER behaviors are almost identical for different launched signal powers. Error floors, however, appear after the regenerator when the launched signal power is 9.5 and 11dBm.
The error floors appear even when the threshold of the regenerator, or the averaged input signal power to the regenerator, is optimally chosen. This is expected because the regenerator captures more noise than the detector at the receiver. Since the duration of the input pulses to the 2R amplitude regenerator should be narrow enough for proper operation of the regenerator, the noise bandwidth at the input of the amplitude regenerator is wider than that at the entrance of the detector in the receiver, which leads to enhanced error by the regenerator. Better design of the 2R amplitude regenerator that allows the use of wider pulse duration with narrower bandwidth will lower the error floors.
In spite of the error floor, the pulses are well reshaped by the regenerator. This gives rise to large reduction of power penalty after transmission over the subsequent span. Figure 3(b) shows the BER performance measured after the second fiber span consisting of an SMF and a DCF. The launched signal power to the first fiber span is fixed at 9.5dBm. ASE generated in the second span is enhanced by increasing the attenuation (ATT2). Figure 3(b) shows large benefits of the regenerator inserted before the second span especially when the noise added in the second span is large.
In the second measurement, the signal before the regenerator is degraded by ASE while the launched signal power to the DDM fiber is kept low. Figure 4(a) shows the BER performance measured after the first span with or without inserting the regenerator. The attenuation of ATT1 in Fig. 3 is varied between 8 and 16dB that is compensated for by the EDFA right after the attenuator. The ASE gives both amplitude and phase noise to the signal. As was discussed in Section 2, the amplitude noise on the DPSK input signal is transferred to the amplitude noise of the demodulated OOK signal after the DI. Suppression of the amplitude noise of the OOK signal by the 2R amplitude regenerator is more crucial in this case than in the previous case of degradation mainly due to phase noise. Reduction of penalty by the regenerator is weaker in the case of ASE degradation as shown in Fig. 4(a). Figure 4(b) shows the BER performance measured after the second fiber span. The amount of ATT1 in the first span is fixed at 12dB. Although the error floor originated in the first span remains, the reshaping effect gives rise to reduction of power penalty at BERs larger than about 10-8. The regenerator performance, however, will be improved by the use of the 2R amplitude regenerator having better noise suppression capability.
In order to assess the effectiveness of the DPSK signal regenerator more convincingly in different operation conditions, direct measurements of signal constellations before and after the regenerator by a method such as that described in  should be performed.
In this paper we reported an all-optical regeneration experiment of short-pulse DPSK signals. In the regenerator, a two-stage fiber-based 2R amplitude regenerator in bidirectional configuration was used to eliminate amplitude noise of the signal after DPSK-to-OOK format conversion. All-optical phase modulation based on XPM in another fiber was then performed for returning the OOK data back to BPSK data on clock pulses. BER performance of the system was studied for two cases. In one case, the signal before the regenerator was principally degraded by nonlinearity, or more specifically, the nonlinear phase noise. In the other case, the signal before the regenerator was degraded by ASE. The regenerator was more effective in the former case, that is, the regenerator was more effective in regenerating DPSK signals corrupted with phase noise.
Although the present experiment reported regeneration of signals at 10Gb/s, the data speed can be raised much higher beyond 100Gb/s if suitable clock pulse sources are used.
References and links
1. O. Leclerc, B. Lavigne, D. Chiaroni, and E. Desurvire, “All-optical regeneration: Principles and WDM implementation,” in Optical Fiber Telecommunications IV A, Components, I. Kaminow and T. Li, eds. (Academic Press, 2002), pp. 732–783.
2. T. Ohara, H. Takara, A. Hirano, K. Mori, and S. Kawanishi, “40-Gb/s x 4-channel all-optical multichannel limiter utilizing spectrally filtered optical solitons,” IEEE Photon. Technol. Lett. 15, 763–765 (2003). [CrossRef]
4. C. Kouloumentas, P. Vorreau, L. Provost, P. Petropoulos, W. Freude, J. Leuthold, and I. Tomkos, “All-Fiberized Dispersion-Managed Multichannel Regeneration at 43 Gb/s,” IEEE Photon. Technol. Lett. 20, 1854–1856 (2008). [CrossRef]
5. V. S. Grigoryan, M. Shin, P. Devgan, and P. Kumar, “Mechanism of SOA-based regenerative amplification of phase-noise degradaed DPSK signals,” Electron. Lett. 41, 1021–1022 (2005). [CrossRef]
6. I. Kang, C. Dorrer, L. Zhang, M. Dinu, M. Rasras, L. L. Buhl (Larry), S. Cabot, A. Bhardwaj, X. Liu, M. A. Cappuzzo, L. Gomez, A. W. -Foy, Y. F. Chen, N. K. Dutta, S. S. Patel, D. T. Neilson, C. R. Giles, A. Piccirilli, and J. Jaques, “Charcterization of dynamical processes in all-optical signal processing using semiconductor optical amplifiers,” IEEE J. Sel. Top. Quantum Electron. 14, 758–769 (2008). [CrossRef]
7. P. Vorreau, A. Marculescu, J. Wang, G. Böttger, B. Sartorius, C. Bornholdt, J. Slovak, M. Schlak, C. Schmidt, S. Tsadka, W. Freude, and J. Leuthold, “Cascadability and regenerative properties of SOA all-optical DPSK wavelength converters,” IEEE Photon. Technol. Lett. 18, 1970–1972 (2006). [CrossRef]
8. K. Croussore, I. Kim, C. Kim, Y. Han, and G. Li, “Phase-and-amplitude regeneration of differential phase-shift keyed signals using a phase-sensitive amplifier,” Opt. Express 14, 2085–2094 (2006). [CrossRef] [PubMed]
9. K. Cvecek, K. Sponsel, G. Onishchukov, B. Schmauss, and G. Leuchs, “2R-regeneration of an RZ-DPSK signal using a nonlinear amplifying loop mirror,” IEEE Photon. Technol. Lett. 19, 146–148 (2007). [CrossRef]
10. F. Futami, R. Okabe, S. Ono, S. Watanabe, R. Ludwig, C. Schmidt-Langhorst, and C. Schubert, “All-optical amplitude noise suppression of 160-Gb/s OOK and DPSK data signals using a parametric fiber switch,” 2007 Optical Fiber Communication Conference, paper OThB3 (2007).
13. S. Ferber, R. Ludwig, C. Boerner, C. Schmidt-Langhorst, L. Molle, K. Habel, M. Rohde, H. G. Weber, F. Futami, and S. Watanabe, “160Gb/s regenerating conversion node,” 2004 Optical Fiber Communication Conference, paper ThT2 (2004).
14. M. Yagi, S. Satomi, and S. Ryu, “All-optical 2R-regenerative interconnection node for DPSK polarization-division multiplexed systems,” 2007 Optical Fiber Communication Conference, paper OThR3 (2007).
15. C. Schmidt-Langhorst, R. Ludwig, M. Galili, B. Huettl, F. Futami, S. Watanabe, and C. Schubert, “160 Gbit/s all-optical OOK to DPSK in-line format conversion,” 2006 European Conference on Optical Communication, paper Th 4.3.5 (2006).
16. M. Matsumoto, Y. Shimada, and H. Sakaguchi, “Two-stage SPM-based all-optical 2R regeneration by bidirectional use of a highly nonlinear fiber,” IEEE J. Quantum Electron. 45, 51–58 (2009). [CrossRef]
17. M. Matsumoto, “A fiber-based all-optical 3R regenerator for DPSK signals,” IEEE Photon. Technol. Lett. 19, 273–275 (2007). [CrossRef]
18. R. P. Webb, R. J. Manning, G. D. Maxwell, and A. J. Poustie, “40 gbit/s all-optical XOR gate based on hybrid-integrated Mach-Zehnder interferometer,” Electron. Lett. 39, 79–81 (2003). [CrossRef]
19. T. -H. Her, G. Raybon, and C. Headley, “Optimization of pulse regeneration at 40 Gb/s based on spectral filtering of self-phase modulation in fiber,” IEEE Photon. Technol. Lett. 16, 200–202 (2004). [CrossRef]
20. M. Jinno, “All-optical signal regularizing/regeneration using a nonlinear fiber Sagnac interferometer switch with signal-clock walk-off,” J. Lightwave Technol. 12, 1648–1659 (1994). [CrossRef]
21. M. Sköld, J. Yang, H. Sunnerud, M. Karlsson, S. Oda, and P. A. Andrekson, “Constellation diagram analysis of DPSK signal regeneration in a saturated parametric amplifier,” Opt. Express 16, 5974–5982 (2008). [CrossRef] [PubMed]