We demonstrate reduction of four-wave mixing crosstalk using hybrid optical parametric amplifier with erbium-doped fiber amplifier. Crosstalk reduction of more than 13 dB has been achieved while providing 20-dB gain. Noise figures of different amplifier configurations are compared. Bit error rate measurements confirm the hybrid configuration introduces < 0.5 dB and < 1.5 dB power penalties in signal and idler wavelength, respectively, compared with a pure optical parametric amplifier.
© 2013 Optical Society of America
Fiber optical parametric amplifiers (OPAs) have some properties which can potentially be exploited for improving the performance of optical communication systems , including: low noise figure (NF), which can be as low as 0 dB; ability to perform wavelength conversion and phase conjugation; broad gain bandwidth and arbitrary pump wavelengths., Studies have been performed on OPA amplification of single-channel signals with different modulation formats (phase-shift keying, quadrature amplitude modulation) at different bit rates [2–4], which showed that OPAs can provide amplification with small power penalties. However, application to practical dense wavelength division multiplexing (DWDM) systems is limited by undesired nonlinear phenomena in the OPAs. In particular four-wave mixing (FWM) across channels generates new frequency components which can coincide with signals . This leads to optical signal-to-noise ratio (OSNR) degradation, which gets worse as the number of channels increases. As a result, amplification of WDM spectra has so far been limited to about 1 Tb/s in our previous experiment . Theoretical study  has shown that amplification of 11 Tb/s signals can be feasible, but still limited by the FWM crosstalk when the number of channels is large. To reduce the crosstalk, the nonlinear interaction among the channels should be reduced.
One approach for reducing FWM crosstalk is to use short fibers and high pump powers , but pump quality could be an issue. Other approaches include using different modulation formats  and special wavelength assignment , but they cannot easily adapt to different systems. Recently we proposed to follow a fiber OPA by a different type of booster amplifier, such as an erbium-doped fiber amplifier (EDFA) or a Raman amplifier, in order to push the high-power fields outside the OPA and into a medium less prone to FWM [11, 12]. In this manner, a 10-dB gain OPA followed by a 10-dB gain EDFA should reduce the power of the FWM product by over 10 dB, compared to a single OPA providing 20-dB gain. Of course such a combination presents trade-offs with some of the other OPA properties, but it may be worthwhile under some circumstances, as OPAs can provide high-efficiency wavelength conversion and other unique features. The phase-conjugated idlers can be useful for dispersion management and mitigation of nonlinear effects occurring in the transmission fibers.
Previously in [11, 12], we presented the detailed calculation of the quantum noise figure for a hybrid amplifier with phase-sensitive amplifier/phase-insensitive amplifier configuration. We forecasted that with such a hybrid amplifier, crosstalk reduction can be reduced as the noise figure is mainly dominated by the first-stage of the hybrid amplifier. Here we experimentally investigate the performance of the proposed hybrid OPA/EDFA. First we will study the reduction of signal-signal FWM in a fiber OPA by using this hybrid approach. Both simulation and experimental results show the reduction can exceed 13 dB for 20-dB gain. Second the operational NF of the proposed amplifier will be obtained experimentally by measuring the OSNR reduction after the amplifier. Compared with a single OPA, the hybrid amplifier can provide the same amount of gain (20 dB) while maintaining a signal NF of 5.9 dB and providing an idler NF of 6.8 dB. Furthermore, bit error rate (BER) measurements with a 40 Gb/s on-off keying (OOK) signal are used to compare the performances among the hybrid OPA/EDFA, OPA and EDFA, all providing 20-dB gain. For the signal wavelength, the power penalties are 0.3 dB and 2 dB when compared with OPA and EDFA, respectively. For the idler wavelength, the hybrid OPA/EDFA exhibits < 1.5 dB power penalty compared with OPA and ~2 dB power penalty compared with its signal wavelength.
2. Crosstalk reduction
The experimental setup for studying crosstalk reduction is shown in Fig. 1. A DFB laser is used to generate continuous-wave (CW) pump light at 1564.00 nm. The light is then phase-modulated by a 3 Gb/s pseudo-random binary sequence (PRBS) with a sequence length of 27-1 to increase the stimulated Brillouin scattering (SBS) threshold of the OPA. The modulated light is then amplified by an EDFA. Its amplified spontaneous emission (ASE) noise is filtered out by a 1-nm bandwidth tunable bandpass filter (TBPF1). Two CW signals at 1547.2 nm and 1547.5 nm are obtained from two tunable laser (TL) sources. Polarization controllers are inserted to ensure that their polarization states are aligned with that of the pump. They are coupled with a 3-dB coupler and connected to an optical isolator, which is used to block the back-scattered SBS light. The signals are combined with the pump using a 90/10 coupler. The total signal power after the coupler is −4 dBm. Another 99/1 coupler is used to monitor the back-scattered SBS light and the pump power before the highly nonlinear fiber (HNLF). Parametric amplification occurs inside the HNLF. Its length is 340 m, with nonlinearity coefficient γ = 15 W−1km−1, zero dispersion wavelength of 1560 nm, and dispersion slope of 0.023 ps nm−2 km−1. The output of the OPA is monitored using a 1% coupler. For hybrid OPA/EDFA, the signals (with FWM crosstalk) are further filtered using a 1.6-nm bandpass filter (TBPF2). They are then amplified by another EDFA. The ASE noise is removed using a 1.9-nm bandpass filter (TBPF3) and the filtered signals are analyzed on an OSA.
First we examine the FWM crosstalk in the OPA at different gain levels. Figure 2 shows the optical spectra when the OPA provides 10 dB and 20 dB gain, with pump powers 27 dBm and 29 dBm, respectively. Here we focus on the signal-signal crosstalk in the C-band. Although signal-pump crosstalk terms also exist in the spectrum, they can be kept away from the signals if the pump and signal wavelengths are properly arranged. However, the signal-signal FWM terms are more difficult to avoid, and thus constitute a more fundamental problem.
Figure 2 shows the spectra of the input signals and the output signals at different gain levels after the HNLF. They are captured after TBPF2 in Fig. 1. We define the left crosstalk ratio as the peak intensity ratio between the signal at 1547.2 nm and idler at 1546.9 nm, while the right crosstalk ratio is the peak intensity ratio between the signal at 1547.5 nm and the idler at 1547.8 nm. For the 20-dB gain OPA, the left and right crosstalk ratios are measured to be 30.1 dB and 31.2 dB, respectively. For the 10-dB gain OPA, the left and right crosstalk ratios are 43.5 dB and 44.2 dB, respectively. The crosstalk has been reduced by 13.4 dB and 13.0 dB, respectively. So the 10-dB gain OPA is used for the first stage of the hybrid amplifier due to its small-crosstalk performance. (The side peaks other than the signal-signal beats are caused by the residual side-modes of the DFB laser in the pump light.)
To implement the hybrid OPA/EDFA, the filtered output signals from the 10-dB gain OPA are directed to another EDFA which is used to provide 10 dB additional gain. At the hybrid amplifier output, a bandpass filter (TBPF3) is used to filter the out-of-band ASE noise. The spectrum of the output signals is shown in Fig. 3. The left and right crosstalk ratios are 44.1 dB and 44.7 dB respectively, resulting in crosstalk reduction of 14.0 dB and 13.5 dB while providing 20 dB overall gain. The crosstalk ratios are similar to those before EDFA amplification as the EDFA provides gain but not additional crosstalk (while a high-gain OPA would increase the crosstalk).
To model the crosstalk in the OPA, we also performed numerical simulations for 10-dB and 20-dB gain OPA. Figure 4 shows the simulation results. The signals are selected at 1547.2 and 1547.5 nm. The pump is at 1564 nm, with the same HNLF parameters as in the experiment. We use a split-step Fourier method to calculate the gain and the field intensities. For the 10-dB gain OPA, the left and right crosstalk ratios are 47.6 dB and 48.1 dB, while for the 20-dB gain OPA they are 33.6 dB and 34.5 dB, respectively. The crosstalk reductions are therefore 14 dB and 13.6 dB. The simulation results match well with the experimental values.
3. Noise figure
It is also important to study the signal NF of the hybrid OPA/EDFA, as it is expected that the additional amplifier should not introduce much noise in the whole system [11, 12]. Moreover, the idler NF will be studied as it is an advantage provided by OPA compared with EDFA. Here we focus on the comparison among OPA, hybrid OPA/EDFA and EDFA at gain level of 20 dB. 20-dB gain EDFA is achieved by directing the input signal to EDFA2 in Fig. 1 followed by TBPF3 for ASE noise filtering.
To measure the signal NF, a single signal channel at 1547.35 nm is used, instead of two. The DFB pump laser is replaced by the unused TL to remove the undesired laser side modes. Figure 5 shows the measured output signal spectra after different amplifiers. The spectrum of the output signal with OPA is captured after TBPF2 in Fig. 1, while output signals amplified by hybrid OPA/EDFA and EDFA are monitored after TBPF3.
The signal NF is defined as the ratio of optical signal-to-noise ratio (OSNR) between input signal and output signal :
where OSNRSignal,in and OSNRSignal,out are the power ratios between the peak intensity and noise power 0.5-nm away from the peak wavelength obtained from Fig. 5. The OSNR of the input laser is measured to be 53.8 dB. As a result, the NF defined here is an operational one as the input laser is not in coherent state. We focused on the OSNR reduction after the amplifier and compare them. The measured OSNRs and NFs of different amplifiers are shown in Table 1.
Although the laser input is not in a coherent state, the relative differences of NF among amplifiers give a brief idea of their performances in real applications. It is observed that the hybrid amplifier has a similar noise figure to that of the EDFA, showing that it may be practical in DWDM communication systems. Also the hybrid OPA/EDFA has a lower noise figure compared with the 20-dB gain OPA. The reason is that a 10-dB gain OPA always has a lower noise figure than a 20-dB gain OPA . After the 10-dB gain OPA, the noise effect from the EDFA is small , so the overall noise figure is smaller than that of a pure 20-dB gain OPA.
We also measured the OSNRs and NFs of the idler. As the idler is in L-band when the signal is at 1547.35 nm, devices necessary to measure its OSNR (filter and an identical EDFA) were not available. So we put the signal at the previous idler wavelength (1580.8 nm) and the new idler is generated at 1547.35 nm. In consequence all the components are the same as in the previous experiment, except the laser wavelengths. Figure 6 shows the spectra of the output idlers after different amplifier configurations. The linewidth of the idler is larger than that of the signal, as twice the phase modulation of the pump is transferred to it during the parametric amplification process.
The idler NF is defined as the ratio of OSNR between input signal and output idler:
where OSNRSignal,in and OSNRIdler,out are the power ratios between the peak intensity and noise power 0.5-nm away from the peak wavelength obtained from Fig. 6. The input signal at 1580.8 nm has the same OSNR as that of 1547.35 nm. The measured OSNRs and NFs are shown in Table 2.
The idler NFs are higher than those of the signals, as expected . Compared with the signal wavelength, the hybrid configuration shows 0.9-dB NF degradation, while for OPA the degradation is 0.6 dB. The difference is caused by the lower conversion efficiency in the first stage 10-dB gain OPA for the hybrid configuration. Nevertheless, the degradations are less than 1 dB for both cases, showing that the idler can be practical for different applications.
4. Bit-error rate measurements
To further examine the performance of the amplifiers, BER measurements of signals and idlers were performed using a 40 Gb/s OOK signal. Again three amplifiers will be tested: 20-dB gain OPA, 20-dB gain EDFA, and 20-dB gain hybrid OPA/EDFA. Idlers can only be generated with OPA and hybrid OPA/EDFA. The experimental setup for data signal generation is shown in Fig. 7(a). For signal measurements, a TL at 1547.35 nm is modulated by a 40 Gb/s PRBS data with sequence length of 231-1 using a Mach-Zehnder modulator, while for idler measurement the light is generated at 1580.8 nm. One polarization controller is used to maximize the modulation performance, while the other one is used to align the signal with the pump polarization for OPA and hybrid OPA/EDFA. The data signal then passes through a variable optical attenuator (VOA), which is used to vary the optical power for BER measurements. The signal is then split into two branches using a 50/50 optical coupler. One of the branches is used for power monitoring, while the other one is connected to an optical isolator to block the back-scattered SBS light.
Figure 7(b) describes the experimental setup for 20-dB gain OPA. It is similar to the one in Fig. 1. To reduce distortion of the data signal caused by phase-to-intensity modulation, four RF clock tones (100 MHz, 320 MHz, 980 MHz, and 2600 MHz) are used to modulate the phase modulator instead of 3 Gb/s PRBS data. Also two cascaded filters (TBPF2 and TBPF3) are used to block the intense pump. For 20-dB EDFA shown in Fig. 7(c), EDFA2 used in the previous experiment is utilized, followed by a 1-nn tunable bandpass filter (TBPF4) to remove the out-of-band ASE noise. For hybrid OPA/EDFA in Fig. 7(d), the experimental setup is a combination of Figs. 7(b) and 7(c) with 10-dB gain in each stage. At the receiver side shown in Fig. 7(e), different amplified signals are directed to another EDFA (EDFA3) to ensure optimal detection of the photodetector (PD). A 3-nm flat-top bandpass filter (TBPF5) follows to remove the noise from the receiver EDFA. The PD has a 3-dB bandwidth of 75 GHz. The detected signals are analyzed by an error analyzer (EA) and a sampling oscilloscope.
Figures 8(a)–8(e) show the eye diagrams of the output signals and idlers from different amplifiers. All eye diagrams are captured with the same input signal power to the amplifiers. It is observed that for OPA and hybrid OPA/EDFA, the noise in bit ‘1’ is more serious than that in EDFA. This could be caused by the relative intensity noise from the OPA pump [1, 14], which could potentially be reduced by simultaneous detection of pump and signal .
Finally we perform the BER measurements. The results are shown in Fig. 9. Error-free operations (BER < 10−9) have been achieved for all amplifiers. For the signal measurements, the hybrid approach introduced ~0.3 dB and ~2 dB power penalties compared with OPA and EDFA, respectively. For the idler measurements, the idler-to-signal power penalty of hybrid OPA/EDFA is ~2 dB, while for OPA it is 1 dB. Thus the idler from hybrid OPA/EDFA has ~1.3 dB power penalty compared with that of OPA. The higher power penalty at idler wavelength is caused by the lower conversion efficiency in the first stage 10-dB gain OPA. However, in a DWDM system, the 20-dB OPA would face a severe crosstalk problem, while our proposed hybrid configuration can provide more than 13 dB reduction of four-wave mixing induced crosstalk. With the substantial gain and the phase conjugation property, the idler could be useful for different applications including parametric tunable delay , dispersion compensation  and mitigation of nonlinear effects in long-haul transmission systems [18, 19].
Compared with the NF measurements, it is observed that for the cases of EDFA and OPA, the differences in NF could predict the power penalties well. However for hybrid amplifiers, the NF was underestimated. The reason is that OSNR cannot accurately predict the signal in-band noise transferred from the residual pump ASE noise in the first stage of hybrid OPA/EDFA . This residual ASE noise is even more severe after being amplified by second-stage EDFA. This noise would also be transferred to the idler, so both signal and idler OSNR cannot match with the BER measurements. The mismatch could be improved by applying cascaded filtering on the OPA pump. Electrical signal-to-noise ratio (ESNR) would be a more accurate technique to analyze the noise figure but it was not available at the time of experiment.
For further development, it would be of great interest to replace the OPA stage with a phase-sensitive amplifier (PSA). Very low noise Fig. (1.1 dB) has been recently demonstrated using PSA . With the hybrid configuration, the system requirement can be reduced as high-gain PSA is not required in our approach, and the noise performance can be potentially maintained at a low level. Another issue to be solved is that the bandwidth of the hybrid OPA/EDFA is currently limited by the EDFA stage. Using a C + L band EDFA could help to extend the operation bandwidth.
We have experimentally demonstrated a novel hybrid OPA/EDFA and analyzed its performance. With 20-dB gain divided equally between the two amplifiers, the FWM crosstalk has been reduced by more than 13 dB, compared to a 20-dB OPA. NFs of different amplifiers have been compared. BER measurements show that the hybrid approach introduced ~0.3 dB and ~1.3 dB power penalty at signal and idler wavelengths compared with a 20-dB OPA. This confirms the hybrid approach could be an interesting candidate when both high gain and unique OPA properties (i.e. phase-sensitive amplification, or wavelength conversion) are desired. The small power penalty and large crosstalk reduction could make the hybrid approach attractive in OPA-based DWDM systems.
This work was supported from the UK’s EPSRC in part by grant EP/J009709/2. The HNLF was provided by Sumitomo Electric Industries, Japan.
References and links
1. M. E. Marhic, Fiber Optical Parametric Amplifiers, Oscillators and Related Devices. (Cambridge University Press, 2007).
2. R. Elschner, T. Richter, M. Nölle, J. Hilt, and C. Schubert, “Parametric amplification of 28-GBd NRZ-16QAM signals,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2011, OSA Technical Digest (CD), paper OThC2 (Optical Society of America, 2011). [CrossRef]
3. Z. Lali-Dastjerdi, M. Galili, H. C. Hansen Mulvad, H. Hu, L. K. Oxenløwe, K. Rottwitt, and C. Peucheret, “Parametric Amplification of a 640 Gbit/s RZ-DPSK Signal,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2013, OSA Technical Digest (online), paper JW2A.21 (Optical Society of America, 2013). [CrossRef]
4. T. Richter, R. Elschner, A. Gandhe, K. Petermann, and C. Schubert, “Parametric ampliﬁcation and wavelength conversion of single- and dual-polarization DQPSK signals,” IEEE J. Sel. Top. Quantum Electron. 18(2), 988–995 (2012). [CrossRef]
5. T. Torounidis, H. Sunnerud, P. O. Hedekvist, and P. A. Andrekson, “Amplification of WDM signals in fiber-based optical parametric amplifiers,” IEEE Photon. Technol. Lett. 15(8), 1061–1063 (2003). [CrossRef]
6. N. El Dahdah, D. S. Govan, M. Jamshidifar, N. J. Doran, and M. E. Marhic, “Fiber optical parametric amplifier performance in a 1-Tb/s DWDM communication system,” IEEE J. Sel. Top. Quantum Electron. 18(2), 950–957 (2012). [CrossRef]
7. M. Jazayerifar, S. Warm, R. Elschner, D. Kroushkov, I. Sackey, C. Meuer, C. Schubert, and K. Petermann, “Performance evaluation of DWDM communication systems with fiber optical parametric ampliﬁers,” J. Lightwave Technol. 31(9), 1454–1461 (2013). [CrossRef]
8. M. Jamshidifar, A. Vedadi, and M. E. Marhic, “Reduction of four-wave mixing crosstalk in a short fiber-optical parametric amplifier,” IEEE Photon. Technol. Lett. 21(17), 1244–1246 (2009). [CrossRef]
9. B. P.-P. Kuo, P. C. Chui, and K. K.-Y. Wong, “A comprehensive study on crosstalk suppression techniques in fiber optical parametric amplifier by modulation format,” IEEE J. Sel. Top. Quantum Electron. 14(3), 659–665 (2008). [CrossRef]
10. F. Forghieri, R. W. Tkach, A. R. Chraplyvy, and D. Marcuse, “Reduction of four-wave mixing crosstalk in WDM systems using unequally spaced channels,” IEEE Photon. Technol. Lett. 6(6), 754–756 (1994). [CrossRef]
12. M. E. Marhic, “Quantum-limited noise figure of networks of linear optical elements,” J. Opt. Soc. Am. B 30(6), 1462–1472 (2013). [CrossRef]
13. H. A. Haus, “The noise figure of optical amplifiers,” IEEE Photon. Technol. Lett. 10(11), 1602–1604 (1998). [CrossRef]
14. M. E. Marhic, G. Kalogerakis, K. K.-Y. Wong, and L. G. Kazovsky, “Pump-to-signal transfer of low-frequency intensity modulation in fiber optical parametric amplifiers,” J. Lightwave Technol. 23(3), 1049–1055 (2005). [CrossRef]
15. V. Ataie, E. Myslivets, A. O. J. Wiberg, N. Alic, and S. Radic, “Pump-noise transfer mitigation in parametric sampling gates,” IEEE Photon. Technol. Lett. 24(17), 1469–1471 (2012). [CrossRef]
16. E. Myslivets, N. Alic, S. Moro, B. P. P. Kuo, R. M. Jopson, C. J. McKinstrie, M. Karlsson, and S. Radic, “1.56-micros continuously tunable parametric delay line for a 40-Gb/s signal,” Opt. Express 17(14), 11958–11964 (2009). [CrossRef] [PubMed]
17. S. L. Jansen, D. van den Borne, P. M. Krummrich, S. Spälter, G.-D. Khoe, and H. de Waardt, “Long-haul DWDM transmission systems employing optical phase conjugation,” IEEE J. Sel. Top. Quantum Electron. 12(4), 505–520 (2006). [CrossRef]
18. X. Liu, A. R. Chraplyvy, P. J. Winzer, R. W. Tkach, and S. Chandrasekhar, “Phase-conjugated twin waves for communication beyond the Kerr nonlinearity limit,” Nat. Photonics 7(7), 560–568 (2013). [CrossRef]
19. Y. Tian, Y.-K. Huang, S. Zhang, P. R. Prucnal, and T. Wang, “112-Gb/s DP-QPSK transmission over 7,860-km DMF using phase-conjugated copy and digital phase-sensitive boosting with enhanced noise and nonlinearity tolerance,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2013, OSA Technical Digest (online), paper OTu2B.5 (Optical Society of America, 2013). [CrossRef]
20. Z. Tong, A. Bogris, M. Karlsson, and P. A. Andrekson, “Full characterization of the signal and idler noise figure spectra in single-pumped fiber optical parametric amplifiers,” Opt. Express 18(3), 2884–2893 (2010). [CrossRef] [PubMed]
21. Z. Tong, C. Lundström, P. A. Andrekson, C. J. McKinstrie, M. Karlsson, D. J. Blessing, E. Tipsuwannakul, B. J. Puttnam, H. Toda, and L. Grüner-Nielsen, “Towards ultrasensitive optical links enabled by low-noise phase-sensitive amplifiers,” Nat. Photonics 5(7), 430–436 (2011). [CrossRef]