All-optical technique for modulation format conversion from 4 channels non-return-to-zero on-off-keying (OOK) to return-to-zero 16 quadrature amplitude modulation (16QAM) employing nonlinear optical loop mirror with 1:2 coupler configuration is proposed and experimentally demonstrated at 10Gs/s. The experimentally converted 16QAM signal was distorted in its phase by cross-phase modulation induced amplitude-to-phase noise conversion. The effect of amplified spontaneous emission noise in the amplified OOK signals to the converted 16QAM’s phase was theoretically discussed.
© 2012 OSA
Due to the rapidly worldwide growing data and internet traffic in telecommunication networks, the next-generation core optical networks will require significant improvements in transmission capacity . Multi-level modulation, preferable in combination with coherent detection, as the key technique for increasing the spectral efficiency (SE) while simultane-ously maintaining long transmission distance , is part of the innovative solutions to satisfy its requirement. On the other hand, in metro area network (MAN), the conventional non-return-to-zero on-off-keying (NRZ-OOK) formats might keep be employed due to the simple and cost-effective transceiver configuration. Consequently, disparate modulation formats may incorporate together in the future optical networks . The function of NRZ-OOK to Multi-level modulation format conversion will be an important interface technology [4, 5] for fitting the optical network flexibility in choosing optimized modulation format.
All-optical modulation format conversions from NRZ-OOK to phase shift keying (PSK) and amplitude phase shift keying (APSK) using highly nonlinear fiber (HNLF) [4, 5] have already been proposed. Among multi-level modulation formats, 16 quadrature amplitude modulation (16QAM), which carries four bits per symbol, is an attractive candidate owning to its outstanding performance transmission capacity, as high SE [6, 7] and data capacity per wavelength [8, 9]. All-optical OOK to 16QAM modulation format conversion employing nonlinear optical loop mirror (NOLM) based on parametric amplification and cross-phase modulation (XPM) in optical fibers has been proposed and theoretically analyzed .
In this paper, the first experimental demonstration of an all-optical modulation format con-version from NRZ-OOK to return-to-zero (RZ)-16QAM in NOLM with 1:2 coupler configur-ation is presented at 10Gs/s . The simulation and experimental results confirm the feasibility of this proposed converter. However, the generated 16QAM signals are affected in its phase by the amplitude-to-phase-noise conversion  due to the amplitude-dependent property of XPM. The OOK signals should be amplified to high power before launching into the HNLF for intensity-to-phase conversion, and so, the amplified spontaneous emission (ASE) noise from the amplifiers degrades the quality of the converted phase. The effect of ASE noise will be theoretically discussed to improve the quality of conversion.
2. Scheme of OOK to 16QAM modulation format conversion
The proposed OOK to 16QAM conversion scheme employs HNLF for intensity-to-phase conversion by XPM [4, 10]. Figure 1 shows the configuration of the proposed modulation format converter which consists of a NOLM with 1:2 coupler. We define four ports of the 1:2 couplers 1, 2, 3, and 4 as shown in Fig. 1. It has the 1:2 power-splitting ratio and the phases of the coupled output signal and the through output signal are 90° apart. Port 3 is the input port of this converter. When the probe pulse incidents on the NOLM from the coupler, it will be split into two pulses from port 1 and 2. If we define the electrical field of the input probe pulse as Ein, the relation between the input and the output electrical field at port 1 and 2 can be described asFig. 1. On the other hand, the counter-clockwise OOK signals 3 and 4 are for the modulation of QPSK2. For simplicity we presume here the use of polarization-maintaining (PM)-HNLF, or the equivalent, so that only one polarization state of the optical field needs to be considered . The nonlinear phase changes of the clockwise and counter-clockwise probe pulses which are induced by co-propagating OOK signals due to XPM can be described asEquation (2) shows that the phase change of the probe pulse is proportional to the peak powers of OOK signals acting as control pulses. So by properly adjusting the peak powers of the control pulses, both and can be set to 0, π/2, π, or 3π/2. Namely, the clockwise and counter-clockwise probe pulses can achieve 4-level phase change after it transmits through the HNLF. After one round transmission, the electrical field of the probe pulses at port 1 and 2 can be described as10]. From Eq. (2), the non-ideal phase difference depends on the peak-power difference asEq. (5) gives the variance of the phase fluctuations of the clockwise and counter-clockwise QPSK signals which is proportional to the power fluctuations of the OOK signals asEq. (7), even reasonable power fluctuation occurs in the OOK signals probable converts into considerable phase noise in the probe pulse because of the large nonlinear coefficient and long interaction length of HNLF.
3. Experimental setup and results
The experimental setup of the converter is shown in Fig. 2 . The generation unit of NRZ-OOK signals is shown in the top left dotted box. The continuous wave (CW) lights from the tunable laser source 1 (TLS 1) and 2 with the wavelengths of 1546.2nm and 1550.3nm, respectively, were modulated together into NRZ-OOK signals with 10 Gb/s PRBS of length 215-1 by the intensity modulator (IM). A phase modulator (PM) was also used after the IM in order to suppress the stimulated Brillouin scattering (SBS)  which may emerge when high power OOK signals enter the PM-HNLF. Then two OBPFs with the pass width of 0.4 nm follow with a 3dB coupler were used to separate the generated NRZ-OOK signals that took different wavelengths into two branches. We define the separated OOK signals with different wavelength as OOK1, 3 and OOK2, 4. They will be amplified by two EDFAs into two fixed peak powers, separately, for inducing π/2 and π phase shift to the probe pulse. Two OBPF with the pass width of 1nm followed in order to filter out the ASE noise generated in the EDFA as much as possible since that ASE will induce fluctuations to the peak power of OOK signals, then turn out to be a phase fluctuation of the generated QPSK signal by XPM in the PM-HNLF. A PM-3dB coupler was used to separate the NRZ-OOK signals generated in the upper and lower branches with different wavelengths and peak powers into two groups OOK1, 2 and OOK 3, 4 for the phase modulations to the clockwise probe pulse and the counter-clockwise probe pulse. Polarization controllers (PCs) were used to maximize the output powers from the PM-3dB coupler. The variable optical delay lines (VODL) were also used to skew the time delay between the OOK signals and the probe pulses. The PM-NOLM as the NRZ-OOK to RZ-16QAM modulation format converter is shown in the top right dotted box. In the PM-NOLM, a variable optical attenuator (VOA) was used to balance the peak powers of clockwise OOK1, 2 and the counter clockwise OOK3, 4. Two PM-WDM couplers were used to combine the clockwise and the counter-clockwise OOK signals with the probe pulses into the both sides of the PM-HNLF. The optical isolators (ISOs) were used before the PM-WDM coupler for removing the returned OOK signals. The probe pulse is generated from the TLS 3 which takes another wavelength, λp = 1565nm. And the PC is used to maximize the input power of the probe pulses. Then the 16QAM signal is generated at the output of the PM-NOLM. The sampling oscilloscope and the optical coherent receiver follow with the real-time sampling oscilloscope which sampling rate is 10Gs/s were used to observe the waveform and the constellation map of the output signal, respectively. The fast mode of the PM-HNLF was used. The zero dispersion wavelength, the dispersion slope, the loss coefficient, the nonlinear coefficient, and the length of the HNLF are λ0 = 1590 nm, Dλ = 0.024 ps/ nm2/km, α = 2.7 dB/km, γ = 18.0 W−1km−1, and L = 0.5 km, respectively. The wavelengths of two OOK signals were set to be 1550.3nm and 1546.2nm, respectively.
The experimental results without digital signal processing are shown in Fig. 3 . Figure 3(b) is the eye diagram of the output 16QAM signal. The waveform is shown in Fig. 3(a). The 16QAM signals have three power levels in a power ratio of 1:5:9. The three power levels of the 16QAM signal were measured as 1.5mW, 7mW, 13mW. The constellation map of 16QAM signal is shown in Fig. 3(c). The ideal position is also shown as black circles. We can found that the output signals were phase noisy for the ASE noise in the amplified OOK signals, in addition of a bit parametric gain and Raman gain between the control and probe pulses. The effect of the co-generated ASE noise from the EDFAs will be theoretical discussed in the next section. In Fig. 3(c), the upper half of the constellation map gets so noisy that hard to identify each point. So amplitude phase shift keying 1(APSK1) and APSK2 as part of the 16QAM signal’s constellation map were also measured in order to confirm the feasibility of OOK to 16QAM modulation format conversion by this converter. If all the OOK1, 3 signals are fixed as 0, the probe pulse will only get phase shift 0 or π from the control pulses. The amplitude of the output signal will only take two levels. It is defined as APSK1 signal, which is shown in Fig. 4(a) . If all the OOK2, 4 signals are fixed as 0, the probe pulse will only get π/2 phase shift from the control pulses. The eye diagram and the constellation map of APSK 2 are shown in Fig. 4(b). This was achieved by turning off one of the amplifiers in the upper and lower branch, separately. The principle constellation maps of the clockwise and counter-clockwise probe pulses are summarized on the left hand side of the experimental results. Compare the constellation map of APSK1 and APSK2, we can found that the right half of APSK1 got some phase noises since that the EDFA which were used to amplify OOK signals for π phase modulation is not stable enough, so that the upper half of the 16QAM signals become so noisy that hard to identify each point in Fig. 3(c).
In this section, the split-step Fourier method will be used to elaborate the effect of the ASE noise which was generated from the EDFAs on OOK to QPSK/16QAM modulation format converter. The simulation model is shown in Fig. 5 . In Fig. 5, the electric fields of the divided probe pulses in the upper and lower branches of the Mach-Zehnder interferometer illustrate the clockwise and counter-clockwise probe pulses in the NOLM. In this simulation, the phase changes of the clockwise probe pulse induced from counter-propagated OOK signals were not taken into account since the phase change is small enough to be ignored for long trains of input OOK signals . The parameters of the HNLF and wavelengths of the control and probe pulses were as same as the experimental demonstration. Figure 6 shows the simulation results of the eye diagram and the constellation map of the output 16QAM signal when the input OOK signals are 10Gb/s PRBS with the length of 27-1 without ASE noise accompanied. The red circles in the constellation map indicate the sampling point of each pulse. The clear eye opening and undistorted constellation points could be obtained on this ideal condition. The principle peak power of the OOK signals for inducing π/2 and π phase shift should be 20dBm and 23dBm, separately.
Then, the AWGN as a model of ASE noise was added to the OOK signals. The Gaussian-shape OBPF with the pass width of 1nm was followed to cut the infinite flat power spectrum into narrow range. The optical signal to noise ratios (OSNRs) of the OOK signals which used to induce π/2 (ΟΟΚ1, 3) and π (ΟΟΚ2, 4) phase shift to the probe pulse were OSNROOK_π/2 = 34.2 dB, OSNROOK_π = 22 dB, respectively. The numerically obtained it as is probability density function of the peak power fluctuation of the OOK PRBS with the length of 216-1 is shown with red lines in Fig. 7(a) . Then the variance of the numerically obtained power fluctuation of the OOK signals could be calculated by the discrete probability density function asEq. (8) into Eq. (6), the fitting Gaussian distribution could be obtained which was shown with black line in Fig. 7(a). The peak power to fluctuation ratio (POOKi/σOOKi) of the ASE noise added OOK signals is also shown in Fig. 7(a). The simulation results of the constellation maps of QPSK1 and QPSK2 signals generated in the upper and lower branch of the Mach-Zehnder interfero-meter are shown in Fig. 7(b), (c). Even the OSNROOK_π was low as 22 dB, the phase fluctuati-on didn’t exceed the range of ± π/4. The phase fluctuation of the constellation point of QPSK2 was as the same as QPSK1 except the π phase offset. The numerically obtained phase probability distribution functions of each constellation point of the QPSK1 signal were shown inside Fig. 7(b). The numerically obtained variances of the peak power and phase fluctuation, which are shown in Fig. 7(a) and (b), are in agreement well with the analytical relation shown by Eq. (7).
Figure 8 shows simulation results of the constellation map and the eye diagram observed at the output of the converter as the superposition of the two noisy QPSK signals. Each cell was marked by red letters which was divided by the boundry lines of each constelation point. From Fig. 8(a), we can find that the 16 constellation points got a rotation with different degrees, and errors obviously arose in the upper half of the constellation map for low OSNR of OOK_π signals. Compared Fig. 8(b), (c) with the experimental results which were shown in Fig. 4, we can found that the experimental results are in agreement well with the simulated results. However, the constellation map of the experimental results seems even worse for the non-ideal optical coherent receiver and time mismatching of the real-time sampling oscilloscope. It implies that the constellation map of 16QAM and APSK1 can be improved obviously, if the OOK_π signals were amplified with limited ASE noise accompanied.
Different from the QPSK signal, the amplitude-to-phase-noise conversion has evolved to crucial problem to the 16QAM signal for that the ASE noise in two OOK signals accumulated in addition with the compact signal point alignment of the 16QAM signal. Moreover, the constellation point didn’t absolutely rotate relative to the base point. In order to evaluate the bit error rates (BER) of the 16QAM signal. The method of least squares was used for finding the fitting circle of each curved constenlation point. In Fig. 8(a), the constelation point of the output 16QAM signal which take the data of (1001) was marked as “C”. For instance, the fitting circle of the constelation point “C” were shown in Fig. 9(a) . According to this method, the phase boundry points of each constelation point relative to the fitting circle center are labeled as ΔΦL and ΔΦS which is shown in Fig. 9. The numerically obtained phase probability angles function of each constelation point of the converted 16QAM signal relative to the fitting circle center could also be achieved which were still follows as Gaussion distribution just like point “C” as shown in Fig. 9 (b). The bit error rates (BER) of the converted 16QAM signal could be wroted as
The simulated result of the BER of the 16QAM signal versus peak power fluctuation caused by ASE noise in the OOK_π and OOK_π/2 signals, separately, were shown in Fig. 10 . The BER gets better as the POOK_π/σOOK_π and POOK_π/2/σOOK_π/2 gets larger from 12.2 dB to 18 dB, which are shown by the black and red lines in Fig. 10, respectively. The constelation map of the 16 QAM signal was shown in the upper right side of Fig. 10 when OSNR of all of the OOK signals is 34.2 dB, namely P/σ = 18 dB. Compare to Fig. 8(a), signal points on the upper half of the constellation map is condensed obviously for small peak power fluctuation of OOK_π signals. Also, the constelation map of the 16QAM signal when the OSNR of OOK_π/2 signal got to be 22 dB is shown in the lower left side of Fig. 10. We can find that it is different from the constelation map in Fig. 8(a), that the APSK2 signal will get phase noisy if the OSNR of OOK_π/2 signals were low. According to Fig. 10, the peak power to fluctuation ratio of all the OOK signals should be larger than 18 dB. Although the simulated result does not tell us precisely how the phase noise will limit the transmission capability of the converted 16QAM modulation format, it can help us to surmise the errors induced from amplitude-to-phase-noise conversion.
We have proposed and experimentally demonstrated an all-optical OOK to 16QAM modulation format conversion scheme using NOLM with 1:2 coupler for the first time. The feasibility of this converter can be proved by the experimental results. However, the output signals is phase noisy for the ASE noise in the amplified OOK signals, in addition of a bit parametric gain and Raman gain between the control and probe pulses. The effect of ASE noise from EDFAs which used to amplify the control pulses for phase modulation to the probe pulses was also theoretically discussed. According to the simulation results, the experimental results of the constellation map of 16QAM and 4APSK could be improved obviously, if the OOK_π signals were also amplified with limited ASE noise accompanied. The BER of the 16QAM signal versus to the peak power to fluctuation ratio of the OOK signals is numerically calculated. The errors induced by amplitude-to-phase-noise conversion as is the XPM employed all-optical OOK to 16QAM modulation format conversion system could be surmised by the simulation result.
References and links
1. A. A. M. Saleh and J. M. Simmons, “Evolution toward the next-generation core optical network,” J. Lightwave Technol. 24(9), 3303–3321 (2006). [CrossRef]
2. P. J. Winzer and R.-J. Essiambre, “Advanced optical modulation formats,” Proc. IEEE 94(5), 952–985 (2006). [CrossRef]
3. D. Norte and A. E. Willner, “All-optical data format conversions and reconversions between the wavelength and time domains for dynamically reconfigurable WDM networks,” IEEE/OSA J. Lightwave Technol. 14(6), 1170–1182 (1996). [CrossRef]
4. K. Mishina, S. Kitagawa, and A. Maruta, “All-optical modulation format conversion from on-off-keying to multiple-level phase-shift-keying based on nonlinearity in optical fiber,” Opt. Express 15(13), 8444–8453 (2007). [CrossRef] [PubMed]
5. A. Maruta and N. Hashimoto, “Experimental demonstration of all-optical modulation format conversion from nrz-ook to rz-8apsk based on fiber nonlinearity,” Optical Fiber Communication Conf. Expo. (OFC2012), LA, USA, Paper OM3B.
6. G. Charlet, “Progress in optical modulation formats for high-bit rate WDM transmissions,” IEEE J. Sel. Top. Quantum Electron. 12(4), 469–483 (2006). [CrossRef]
7. J. Wang and J. Yang, I. M. Fazal1, N. Ahmed, Y. Yan, B. Shamee, A. E. Willner, K. Birnbaum, J. Choi, B. Erkmen, S. Dolinar, and M. Tur, “Demonstration of 12.8-bit/s/Hz spectral efficiency using 16-QAM signal over multiple orbital-angular-momentum modes,” The 37th European Conference and Exhibition on Optical Communication (ECOC2011), Geneva, Switzerland, Paper We.10.P1.76.
8. S. Yamanaka, T. Kobayashi, A. Sano, H. Masuda, E. Yoshida, Y. Miyamoto, T. Nakagawa, M. Nagatani, and H. Nosaka, “11 x 171 Gb/s PDM 16-QAM Transmission over 1440 km with a Spectral Efficiency of 6.4 b/s/Hz using High-Speed DAC,” The 36th European Conference and Exhibition on Optical Communication (ECOC2010), Turin, Itary, Paper We.8.C.1.
9. A. Sano, H. Masuda, T. Kobayashi, M. Fujiwara, K. Horikoshi, E. Yoshida, Y. Miyamoto, M. Matsui, M. Mizoguchi, H. Yamazaki, Y. Sakamaki, and H. Ishii, “Ultra-High capacity WDM transmission using spectrally-efficient PDM 16-QAM modulation and C- and extended L-Band wideband optical amplification,” IEEE/OSA J. Lightwave Technol. 29(4), 578–586 (2011). [CrossRef]
10. G. Huang, Y. Miyoshi, A. Maruta, Y. Yoshida, and K. Kitayama, “All-optical OOK to 16QAM modulation format conversion employing nonlinear optical fiber loop mirror,” IEEE/OSA J. Lightwave Technol. 30(9), 1342–1350 (2012). [CrossRef]
11. G. Huang, Y. Miyoshi, Y. Yoshida, A. Maruta, and K. Kitayama, “All-optical OOK to 16QAM modulation conversion based on nonlinear optical loop mirror with 1:2 coupler,” Optical Fiber Communication Conf. Expo. (OFC2012), LA, USA, Paper OTh4H.
13. S. Kumar, A. Selvarajan, and G. Anand, “Nonlinear copropagation of two optical pulses of different frequencies in birefringent fibers,” J. Opt. Soc. Am. B 11(5), 810–817 (1994). [CrossRef]
14. M. C. Ho, M. E. Marhic, K. Y. K. Wong, and L. G. Kazovsky, “Narrow-linewidth idler generation in fiber four-wave mixing and parametric amplification by dithering two pumps in opposition of phase,” IEEE/OSA J. Lightwave Technol. 20(3), 469–476 (2002). [CrossRef]