In this paper, polarization-insensitive wavelength conversion based on orthogonal-pumps and four-wave mixing in HNLF is experimentally demonstrated for high-speed 112-Gb/s PolMux-RZ-QPSK transmission with digital coherent detection. The conversion performance of the proposed scheme is investigated for both single- and four-channel input signals, with the achieved post-conversion OSNR for the two cases shown to be 30 and 20 dB, respectively. Moreover, it is shown that the OSNR of the converted single-channel signal can be maintained above 25 dB even if the wavelength spacing between the original and converted signals is larger than 25 nm. Finally, the BER of 4×112-Gb/s PolMux-RZ-QPSK converted signals after 1 km HNLF transmission is measured to be below 1×10-4. The optimum OSNR and launched HNLF power are also investigated.
©2008 Optical Society of America
Wavelength conversion (WC) is expected to play an important role in future WDM based high-speed optical networks, because it provides simultaneous conversion of a single data channel into different channels without the necessity of multiple optical-electronic-optical (O-E-O) transponders [1–18]. Consequently, the optical converters, as well as optical frequency addressing or routing systems would be greatly simplified and rendered significantly more flexible with this approach. Several methods for optical wavelength conversion based on self-phase modulation (SPM), cross-phase modulation (XPM), and cross-gain modulation (XGM) have been proposed in the literature [1–3]. However, most of the hereto proposed WC schemes are also sensitive to the polarization state of the optical signal and are therefore not well-suited to many applications. Wavelength conversion based on four-wave mixing (FWM), on the other hand, is one of the most promising WC techniques, because it is fully transparent to the signal bit rate and modulation format [4–7], and can be accomplished in semiconductor optical amplifiers (SOAs) or fiber. Since transmission speed in the SOA is limited due to carrier response time , high-speed all-optical WC can be realized in highly-nonlinear fiber (HNLF) by using one or more optical pumps . In single-pump schemes, conversion is usually polarization sensitive, so to bypass this difficulty, we have demonstrated a co-polarized pumps WC scheme in . However, co-polarized pump WC suffers from a relatively narrow frequency detuning range. In this investigation, we build on previous work by utilizing orthogonal pumps in 1-km HNLF, a method previously proposed for SOAs and fiber only [5–8], to achieve polarization insensitive WC with a larger frequency detuning range. A primary advantage of this method is that it preserves the data integrity of a polarization multiplexed (PolMux) signal even after wavelength conversion, such that the PolMux signal may be coherently detected. With digital coherent detection, both data-bearing polarizations can be recovered, resulting in notably higher spectral efficiency, while optical transmission impairments such as chromatic dispersion (CD) and polarization-mode dispersion (PMD)  may be readily compensated in the electrical domain. For example, due to these features, PolMux-RZ-QPSK with digital coherent detection has recently been shown to be a potential candidate for 100-Gb/s high-speed fiber transmission. In this paper, we propose polarization-insensitive WC employing orthogonal-pumps and experimentally demonstrate the new approach on 112-Gb/s PolMux-RZ-QPSK transmission with digital coherent detection. We report superior performance over 1-km HNLF for both single- and four-channel input signals.
2. Principle of wavelength conversion using two-orthogonal-pumps
Polarization-independent wavelength conversion is an important feature for future all-optical networks. Figure 1 shows the principle of polarization-insensitive wavelength conversion using two orthogonal pumps for single-channel (Fig. 1(a)) and two-channel (Fig. 1(b)) conversion, respectively. In both cases, it is assumed that the two pumps (P1 and P2) have orthogonal polarizations (P1⊥P2) and longer wavelengths than the input signal(s). For a single input channel (S1 in Fig. 1a), two polarization sensitive signals (C2 and C3) as well as one polarization-insensitive signal (C1) are generated after WC, such that the differential power between C2 (C3) and C1 is about 3 dB. For two input wavelengths (S1 and S2 in Fig. 1b), the converted signals (S1’, S2’) are spectrally inverted, and are spectrally interleaved between three polarization-sensitive signals. Denoting the angle frequencies of signals S1/S2, P1, P2, C1, C2, and C3 by ωs, ωp1, ωp2, ωc1, ωc2, ωc3, respectively, and assuming that ωp1>ωp2, we obtain the following relationships:
As shown by (4), the frequency spacing between the converted signals C2 and C3 is double that between the two orthogonal pumps. Likewise, it can be seen from (5) and (6) that the frequency spacing between C1 and C3 (or C1 and C2) is equal to the spacing between the optical pumps.
3. Experimental setup and results
Figure 2 shows the experimental setup for WC of a single-channel input signal. Three tunable external cavity lasers (ECLs) with linewidth <100 kHz are utilized, with each ECL terminated by polarization maintaining fiber pigtails. Two ECLs at 1561.82 and 1562.22 nm, respectively, denoted by ECL2, ECL3 in Fig. 2, are used as pumps (P1, P2). Since P1 and P2 are coupled by a polarization beam combiner (PBC), their polarizations are made orthogonal to each other. The 1557.77 nm input signal provided by ECL1 is used to drive the 112-Gb/s PolMux-RZ-QPSK transmitter, which is composed of two Mach-Zehnder modulators (MZM1, MZM2), a phase modulator (PM), a polarization-maintaining EDFA (PM-EDFA), and a polarization-multiplexing unit. The first two modulators, MZM1 and PM, are each driven by a 28-Gb/s data stream to provide π and ½π phase modulation, respectively, while MZM2 is driven by a 28-GHz clock to carve out 50%-duty-cycle return-to-zero (RZ) pulses. The 28-Gb/s data stream is obtained by time- multiplexing four 7-Gb/s PRBS signals, each having a pattern length of 211-1. We note that the two 28-Gb/s data signals modulating MZM1 and PM are de-correlated by introducing a differential bit delay between them, such that after electrical multiplexing, the pattern length of the resulting 28Gigabaud/s signal is 213–4). Polarization-multiplexing (PolMux) is achieved by splitting the RZ-QPSK signal and recombining the two branches with a PBC after the insertion of a 322-symbol differential delay. A polarization controller (PC) is used to generate a random polarization direction for the PolMux-RZ-QPSK signal that is entirely independent of the polarization state of the pumps . The launched optical power of pumps P1 and P2 is 13 dBm, while that of the PolMux-RZ-QPSK is 12 dBm. The highly-nonlinear fiber (HNLF) used in this experiment has 0.4 dB/km loss, a nonlinear coefficient of 10 W-1/km, zero dispersion wavelength (ZDW) of 1561 nm, and a 0.02 ps/nm2/km dispersion slope.
The optical spectra with 0.1-nm resolution for combined S1, P1, P2 signal before and after WC are shown in insets (i) and (ii) in Fig. 2, where WC occurs via FWM in 1km HNLF transmission. As shown in Fig. 2, three converted signals are generated after WC, of which two are polarization-sensitive (two outer peaks in Fig. 2ii) and one is polarization-insensitive (middle peak, S1’ in Fig. 2ii). We note that the measured optical signal-to-noise ratio (OSNR) of S1’ is 31 dB, despite a 17 dB power differential compared to the original input signal S1. To select the polarization-independent converted signal S1’ at the receiver, a tunable optical filter (TOF) with a 3-dB bandwidth of 0.35 nm is used. The optical receiver next performs polarization-and phase-diverse coherent detection using a polarization-diversity 90-degree hybrid, a tunable local oscillator (ECL4 with approximate 100 kHz linewidth), and four single-ended photo detectors (PD). The sampling and digitization (A/D) function is carried out by a 4-channel Tektronix digital storage scope (DSA72004) with a 50Gs/s sample rate and 16 GHz electrical bandwidth. The captured data is then post-processed using a desktop PC; the complete details of the digital signal processing (DSP) procedure can be found in . For this experiment, errors were counted over 20×60,000 symbols (20 data sets of 60,000 symbols each), such that the average BER values for the PolMux-RZ-QPSK signal are based on 4.8×10 6 bits.
Fig. 3 illustrates the measured BER of the PolMux-RZ-QPSK signal as a function of OSNR (in 0.1-nm reference bandwidth) before (Back-to-back) and after wavelength conversion, along with the corresponding post-conversion constellation figure. As seen in Fig. 3, to achieve a BER = 2×10-3 after wavelength conversion, an OSNR = 16.5 dB is needed, revealing that the WC-induced OSNR penalty is less than 0.5 dB at the FEC limit. Fig. 4 plots the optical power and OSNR of the converted signals versus the wavelength spacing between them. From Fig. 4, we see that with a wavelength conversion spacing greater than 25 nm, optical power variations in the converted signal are smaller than 3 dB, while the achieved OSNR remains above 25 dB. Moreover, the conversion spacing could be further extended if the dispersion slope of the HNLF were reduced. In the second experiment (Fig. 5), we add three more ECLs to the experimental setup of Fig. 2 to generate 4 × 112-Gb/s PolMux-RZ-QPSK signals at 1555.75 nm (S1), 1556.55nm (S2), 1557.36 nm (S3) and 1558.17 nm (S4), respectively.
In this case, the four channels are first combined by an array-waveguide grating (AWG) and following PolMux-RZ-QPSK modulation, are transmitted over 40-km SMF-28 fiber to decorrelate the resulting WDM signal. Next, a 50/100GHz optical interleaver (IL) is used to remove any overlap between adjacent WDM channels and reserve space for converted polarization-sensitive signals. Two identical WDM filters are employed to combine and separate the pumps and signals within the fiber link. The input power into 1-km of HNLF is 12 dBm per optical pump and -2.3 dBm per PolMux-RZ-QPSK channel so as to also reduce crosstalk effects. The remaining experimental parameters are unchanged with respect to the setup of Fig. 2. The inset (i) in Fig. 5 illustrates the measured optical spectrum after the second WDM filter in Fig. 5, while the received optical spectra after WC for all four input signals are plotted in Fig. 6. As shown by Fig. 6, the OSNR of the converted signals is higher than 25 dB, while the conversion efficiency is -20dB. Fig. 7 presents the measured BER for the 4×112 Gb/s PolMux-RZ-QPSK signal after 1 km HNLF transmission, along with the corresponding constellation figures. From Fig. 7, we note that the BER for all converted signals (S1–S4 in Fig. 6) is around 1×10-4, and could be further reduced by optimizing the pump and signal input powers, as shown in Fig. 8 for the converted channel S2’. In Fig. 8(a), only the total signal power of the original WDM signal S2 is increased from 2.6 dBm to 14.2 dBm, while the total pump power is fixed at 17 dBm. As illustrated by Fig. 8(a), in the optimal input power range of 6–11 dBm, the BER of the converted signal S2’ can be lowered to less than 1×10-5. For S2 input power less than 6 dBm, the OSNR of the converted signal becomes too low and leads to higher BER, while an input power greater than 11 dBm causes strong crosstalk, which also raises BER. Fig. 8(b) plots the measured BER versus variable total pump power at a fixed total signal power of 10.6 dBm. In this case, when the total pump power is between 16 dBm and 19 dBm, BER is less than 2×10-5. Outside of this range, the BER will increase either due to low OSNR when the pump power is too low, or due to prominent SBS effects for overly high pump power. We note that if we were to broaden the pump spectrum, however, we could increase the SBS threshold of the pump in HNLF by up to 17 dBm .
Figure 9 exhibits the OSNR as a function of the input power into the HNLF for converted channel S2’, showing that the OSNR is increased while the total signal power is between 14 and 17 dBm. However, as input power increases beyond 17 dBm, the accumulated ASE noise becomes, causing a saturation in the achievable OSNR. This is illustrated in the insets in Fig. 9 that show the optical spectra for input powers of 14 dBm and 18 dBm, and explains why measured BER increases for fiber input powers greater than 17 dBm.
We have proposed a novel FWM-based wavelength conversion scheme using two orthogonal pumps and have successfully demonstrated polarization-independent WC of 4×112-Gb/s PolMux-RZ-QPSK signals in 1-km HNLF with digital coherent detection. Experimental results have shown that, by using the proposed technique, post-conversion OSNR >30 dB for single-channel WC and OSNR >25 dB for four-channel WC can be achieved. Moreover, for the single-channel case, the wavelength conversion OSNR penalty was shown to be smaller than 0.5 dB, with the converted signal shown to maintain OSNR >25 dB even for conversion wavelength spacing larger than 25 nm. In the case of four PolMux-RZ-QPSK signals, BER <1×10-4 was achieved for all four converted channels. To our best knowledge, this is the first experimental realization of wavelength conversion of 100-Gb/s polarization-multiplexed phase-modulated WDM signals based on polarization-insensitive orthogonal-pump FWM that also uses coherent detection and digital signal processing.
We’d like to thank Dr. D. Qian, Dr. T. Wang and Dr. N. Cvijetic from NEC-Labs, and Dr. X. Zhou from AT&T Lab for their great support on this work.
References and links
1. K. Uesaka, K. Wang, M. E. Marhic, and L. G. Kazovsky, “Wavelength exchange in a highly nonlinear dispersion-shifted fiber: theory and experiments,” IEEE J. Quantum Electron. 18, 560–568 (2002).
2. Y. Liu, E. Tangdiongga, Z. Li, H. de Waardt, A. M. J. Koonen, G. D. Khoe, H. J. S. Dorren, X. Shu, and I. Bennion, “Error-free 320 Gb/s SOA-based wavelength conversion using optical filtering,” in Proc. OFC, paper PDP28 (2006).
3. W. Wang, H. N. Poulsen, L. Rau, H. F. Chou, J. E. Bowers, and D. J. Blumenthal, “Raman-enhanced regenerative ultrafast all-optical fiber XPM wavelength converter,” J. Lightwave Technol. 23, 1105–1115 (2005). [CrossRef]
4. R. M. Jopson and R. E. Tench, “Polarisation-independent phase conjugation of lightwave signals,” Electron. Lett. 29, 2216–2217, (1993). [CrossRef]
5. K. Inoue, “Polarization independent wavelength conversion using fiber four-wave mixing with two orthogonal pump lights of different frequency,” J. Lightwave Technol. 12, 1916–1920 (1994). [CrossRef]
6. L. Y. Lin, J. M. Wiesenfeld, J. S. Perino, and A. H. Gnauck, “Polarization-insensitive wavelength conversion up to 10 Gb/s based on four-wave mixing is a semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 10, 955–957 (1998). [CrossRef]
7. M. Matsuura, N. Kishi, and T. Miki, “Broadband regenerative wavelength conversion and multicasting using triple-stage semiconductor-based wavelength converters,” Opt. Lett. 23, 1026–1028 (2007). [CrossRef]
8. S. H. Wang, L. Xu, P.K.A. Wai, and H. Y. Tam, “All-optical wavelength conversion using multi-pump Raman-assisted four-wave mixing,” in Proc OFC, paper OWQ1 (2007).
9. J. Ma, J. Yu, C. Yu, Z. Jia, X. Sang, Z. Zhou, T. Wang, and G. K. Chang, “Wavelength conversion based on four-wave mixing in high-nonlinear dispersion shifted fiber using a dual-pump configuration,” J. Lightwave Technol. 24, 2851–2858 (2006). [CrossRef]
10. W. Mao, P. A. Andrekson, and J. Toulouse, “All-optical wavelength conversion based on sinusoidal cross-phase modulation in optical fiber,” IEEE Photon. Technol. Lett. 17, 420–422 (2005). [CrossRef]
11. S. Radic, C. J. McKinstrie, R. M. Jopson, A. H. Gnauck, J. C. Centanni, and R. Chraplyvy, “Bit-level switching in a fiber parametric processor with inherent wavelength conversion and optical conjugation,” in Proc. OFC , 23–27 (2004).
12. A. Hamié, A. Sharaiha, M. Guégan, and J. L. Bihan, “All-optical inverted and noninverted wavelength conversion using two-cascaded semiconductor optical amplifiers,” IEEE Photon. Technol. Lett. 17, 1229–1231 (2005). [CrossRef]
13. J. Yu, Z. Jia, Y. K. Yeo, and G. K. Chang, “Spectrally non-inverting wavelength conversion based on FWM in HNL-DSF and its application in label switching optical network,” in Proc. 25th ECOC , 32–35 (2005).
14. C. Porzi, A. Bogoni, L. Poti, and G. Contestabile, “Polarization and wavelength-independent time-division demultiplexing based on copolarized-pumps FWM in an SOA,” IEEE Photon. Technol. Lett. 17, 633–635 (2005). [CrossRef]
15. J. P. R. Lacey, M. A. Summerfield, and S. J. Madden, “Tunability of polarization-insensitive wavelength converters based on four-wave mixing in semiconductor optical amplifiers,” J. Lightwave Technol. 16, 2419–2427 (1998). [CrossRef]
16. K. Uesaka, K. K. Y. Wong, M. E. Marhic, and L. G. Kazovsky“Polarization-insensitive wavelength exchange in highly-nonlinear dispersion-shifted fiber,” in Proc. OFC, paper ThY3 (2006).
17. T. Tanemura, K. Katoh, and K. Kikuchi, “Polarization-insensitive asymmetric four-wave mixing using circularly polarized pumps in a twisted fiber,” Opt. Express 13, 7497–7505 (2005). [CrossRef] [PubMed]
19. D. van den Borne, T. Duthel, C. R. S. Fludger, E. D. Schmidt, T. Wuth, C. Schulien, E. Gottwald, G. D. Khoe, and H. de Waardt, “Coherent equalization versus direct detection for 111-Gb/s Ethernet transport,” in Proc ECOC, paper MA2.4 (2007).
20. X. Zhou, J. Yu, D. Qian, T. wang, P. Zhang, and Magil, “8×114Gb/s, 25-GHz-spaced, polmux-RZ-8PSK transmission over 640km of SSMF employing digital coherent detection and EDFA-only amplification,” in Proc OFC, paper PDP1 (2008).
21. S. K. Korotky, P. B. Hansen, L. Eskildsen, and J. J. Veselka, “Efficient phase modulation scheme for suppressing stimulated Brillouin scattering,” in Proc IOOC, paper WD2-1 (1995).