1. Introduction

Digital signal processing (DSP) is playing an increasingly important role in coherent detection for reconstructing the complex field of signal and compensating the transmission impairments. It dramatically simples the reception of multi-level and multi-dimensional modulation formats such as high-order quadrature amplitude modulation (QAM), thus making high-order QAM become a promising and practical approach for achieving higher bit rate and higher spectral efficiency. However, optical signal processing is still highly desirable in order to overcome the electronics bottlenecks, support the transparency and ultra-fast processing in future optical networks. Among various optical signal processing techniques, all-optical wavelength conversion (AOWC), or fiber frequency conversion, is one of essential network functionalities to enhance the re-configurability, non-blocking capacity, and wavelength management in future transparent optical networks.

Since lots of advanced modulation formats like single-carrier high-order QAM like 64QAM [1–3] or multi-carrier optical orthogonal frequency-division multiplexing (OFDM) have been introduced and realized in optical communications for enabling spectrally-efficient and ultra-fast optical transmissions, research efforts are desired to exploit AOWC schemes suitable for these advanced optical modulation formats. Several schemes have been reported to demonstrate the AOWC functions of OFDM, 8ary phase-shift keying (8PSK), 16QAM and 64QAM by using the second-order nonlinear effect in periodically-poled Lithium Niobate (PPLN) waveguide [4], four-wave mixing (FWM) in highly-nonlinear fibers (HNLF) [5,6], semiconductor optical amplifier (SOA) [7–9], or silicon waveguide. For high-order QAM signals, the increasing number of states in the constellation makes the signal more sensitive to the intensity and phase noise. In the previously reported AOWC schemes for 64QAM, power penalties of around 4-dB at 5Gbaud [9], and 2-dB at 21Gbaud [6] have been demonstrated for the converted 64QAM at bit-error rate (BER) of 10⁻³. In order to achieve a high-quality AOWC for enabling the transmission through multiple AOWCs in the future optical networks, it is essential to suppress the distortions introduced in the AOWC process. The distortions introduced in the AOWC mainly originate from: i) the phase noise from the pumps due to the finite laser linewidth; and ii) other undesired nonlinear distortions or crosstalk co-existed in the nonlinear process. To suppress the linear phase noise from pumps, narrow-linewidth lasers, such as external-cavity laser (ECL) or fiber laser (FL), are suggested to be used as pump sources. On the other hand, since the nonlinear media in the sub-system is operated in the nonlinear operation region, expect the dominant nonlinear effect utilized for implementing AOWC, it is highly possible that other undesired nonlinear effects co-occur in this process, thus deteriorating the quality of the converted signal. For example, for the AOWC utilizing the FWM in SOA, extra distortion from cross-gain modulation, cross-phase modulation (XPM), and self-phase modulation (SPM) may deteriorate the converted signal, while in the AOWC based on FWM in HNLF, additional undesirable distortions are mainly from stimulated Brillouin scattering (SBS), SPM or XPM. High-order QAMs, especially going up to 32QAM, 64QAM or beyond, exhibit more sensitive to nonlinear phase noise like SPM or XPM. Therefore, in order to realize a high-quality AOWC sub-system for high-order QAMs, it is essential to optimize the system performance of AOWC through effective monitoring approach to suppress the distortion introduced by extra undesired nonlinear distortions. To monitor the performance of the converted high-order QAM, BER is usually deployed as an effective performance indicator. However, the co-exited SPM effect in the AOWC process usually introduces a spiral rotation in the high-order QAM constellation [10], which sometimes happens to enlarge the symbol distance between symbols, thus decreasing the BER. Therefore, instead of using BER, it is suggested to optimize the AOWC performance of high-order QAM by monitoring the constellations of the converted signal using a coherent receiver. The undesired extra nonlinear distortion like SPM or SBS in the process can be simply visualized by monitoring the constellations. As a quantitative metric for the constellation monitoring, error vector magnitude (EVM) is measured when varying the pump of probe and pump to optimize the performance of the high-order QAM AOWC. In this paper, by optimizing the operation conditions in AOWC through constellation monitoring, we experimentally demonstrate the AOWC of optical 10-Gbaud (60-Gbps) 64QAM through FWM effect in HNLF with power penalty of less than 0.3-dB at BER of 10⁻³.

2. Experiment and results

Figure 1 depicts the experimental setup to achieve the AOWC of 64QAM through FWM in HNLF. The AOWC is implemented through a simple degenerate FWM in HNLF. The phase of the converted signal follows the phase relationship:${\theta }_{idler}=2{\theta }_{pump}-{\theta }_{prob}{}_e$, where ${\theta }_{idler}$,${\theta }_{pump}$ and ${\theta }_{prob}{}_e$ are the phase of the idler, pump and probe, respectively. As high-order QAM signals like 32QAM or 64QAM are sensitive to the phase noise in the system, it is preferred to deploy narrow linewidth light sources, especially for the pump source, in the experiment. Due to the equipment availability, a tunable external cavity laser (ECL) with a linewidth of around 100 kHz serves as a light source of the input 64QAM (probe) in the experiment, while two fiber lasers (FLs) with a linewidth of around 10 kHz are used as light sources for pump and local oscillator (LO) at the coherent receiver. Since a narrow-linewidth FL was deployed as pump source, the phase noise from pump is relatively negligible. In order to synthesize optical 64QAM, the light from ECL at 1551.38 nm is modulated by a single in-phase/quadrature (IQ) modulator, which has a 3-dB bandwidth of around 25 GHz, and a 3.5-V half-wave voltage. Two de-correlated 8-level driving electronics originating from PRBS streams with length of 2¹⁵-1 are generated from an arbitrary waveform generator (AWG) with a peak-to-peak swing of round 1V to drive the IQ modulator. After power amplification, the 64QAM signal is combined with an amplified CW light from 1551.95 nm, and then fed into a piece of HNLF, which has a length of 150 meter, an attenuation coefficient of 0.9 dB/km, a nonlinear coefficient of 18 /W/km, a zero-dispersion wavelength of 1548 nm, and a dispersion slope of around 0.02 ps/nm2/km. Notice that, due to the wavelength non-tunability of the deployed FLs in the experiment, the wavelength arrangement of probe and pump could not be arranged to maximize the FWM conversion efficiency. However, thanks to the high nonlinearity and flat dispersion profile of the deployed HNLF, the experimental results show that the conversion efficiency is high enough to ensure the good performance of the converted signal. After the HNLF, the generated idle signal at 1552.52nm is filtered and sent to a phase-diversity intradyne coherent receiver for detection and BER measurement. The receiver includes a LO, an optical 90-degree hybrid and two balanced photo-detectors (PDs). After detection by balanced PDs, the data is digitized at 50 GSamples/s by using a digital storage oscilloscope with a 12.5-GHz analog bandwidth (Tektronix DPO71254). The captured data is then off-line processed through digital signal processing (DSP), which includes compensation of skew, power and IQ imbalance, data resampling, linear equalization by finite impulse response (FIR) filtering, carrier phase recovery and final hard-decision circuits. 89,285 symbols were deployed for BER measurement.

 

Fig. 1 Experimental setup.

Download Full Size | PDF

The main undesired nonlinear components in the AOWC are from SPM of the probe (input 64QAM) and SBS of the pump (CW). These undesired phase and intensity noise originating from probe and pump will be transparently transferred together with the desired data modulation to the converted signal through FWM, thus degrading the performance of the AOWC. In order to eliminate these deleterious components in the converted signal, the launched pump and probe power should be well managed. Figure 2(a) depicts the measured EVMs and BERs at the received optical signal-to-noise ratio (OSNR) of around 25 dB when tuning the probe power from 7 to 15 dBm and fixing the pump power at around 20 dBm, while Fig. 2(b) shows the corresponding conversion efficiency. Here the conversion efficiency is defined as the power ratio of the converted signal to the launched probe. It is obvious that the increase in the launched probe power does not make contribution in improving the conversion efficiency. However, it does help decrease the BER of the converted signal till around 12.2dBm. When further increasing the probe power, BER starts to increase, which is probably attributed to the introduced phase noise originating from SPM in the probe. According to this BER characterization with different probe power, it is suggested to operate the probe with the launched power at the range of 9~14 dBm. On the other hand, the measured constellations offer more intuitive and accurate approach to optimize the performance. EVMs with different probe power are measured and plot in Fig. 2 (a) as well. Although similar trends with the increase of probe power are obtained for both BER and EVM results, the different optimal probe power of ~9.2dBm and ~12.4dBm were obtained according to the EVM and BER results, respectively. The measured corresponding constellations provide an intuitive approach to interpret the difference. Figure 3 shows the measured corresponding constellations with different probe power. When launching the probe with power of around 7 dBm, the constellation shown in Fig. 3(a) is quite noisy due to the limited OSNR of the converted idler. The OSNR of the converted idler is improved when the probe power is increased to 9.2 dBm, shown in Fig. 3(b). This is consistent with both the measured BER and EVM results in Fig. 2. However, with the launched probe power increasing to around 12.4 dBm, although we could achieve the lowest BER, the distortion from SPM starts to become visible in the measured constellation, shown in Fig. 3(c). Each symbol is phase-rotated in accordance to its own amplitude level. The symbols with higher amplitude have larger spreading in the radial direction. Consequently, the measured constellation starts to obtain a spiral-like shape. As shown in Fig. 3(d), obviously, this distortion becomes more severe when the probe power increases to 14.8 dBm. Although the SPM-induced distortion becomes visible in constellation when the launched probe power is 12.4 dBm, the slightly-distorted constellation somehow enlarges the symbol distance, thus resulting in the decrease of BER. This probably leads to the different reflections of SPM distortion on BERs and EVMs at the probe power of around 12 dBm. Therefore, it is suggested to optimize the performance of AOWC by monitoring constellation or EVM, which gives more intuitive and proper means to optimize the AOWC performance, in order to eliminate the extra nonlinear phase noise introduced in the process.

 

Fig. 2 (a) Measured BERs and EVMs at the received OSNR of around 25 dB and (b) corresponding conversion efficiencies when tuning the probe power from 7 to 15 dBm and fixing the pump power at 20 dBm.

Download Full Size | PDF

 

Fig. 3 Measured corresponding constellations of the converted signals with different launched probe power: (a) 7 dBm, (b) 9.2 dBm, (c) 12.4 dBm, and (d) 14.8 dBm.

Download Full Size | PDF

Another main nonlinear distortion originates from the SBS of the pump. It is well known that the phase dithering is helpful to increase the SBS threshold. However, the phase dithering in pump is not applicable to the AOWC of high-order QAMs since it will deteriorate the phase of the converted signal. Thanks to the short length of the deployed HNLF (150 meters), the measured SBS threshold is around 24 dBm, allowing a high launching power. However, the optimization of the pump power is required in order to avoid the SBS distortion in the pump. To optimize the pump power, as shown in Fig. 4(a), the EVM and BER values are measured with the increase of the pump power from 17 to 22 dBm when the probe power is set at around 9 dBm. Similar behaviors are observed for both EVM and BER when increasing the pump power. To avoid the distortion from SBS, it is suggested to operate the pump power from 17.5 to 20.5 dBm. On the other hand, Fig. 4(b) shows the measured conversion efficiency when tuning the pump power from 17 to 22 dBm. It is obvious that high pump power is helpful to increase the conversion efficiency and offer a sufficient OSNR for the converted signal. Therefore, the optimal pump power is set at 20 dBm in the experiment, which offers a conversion efficiency of around −15 dB and ensures the converted signal free of the SBS distortion. As shown in Fig. 5, the measured constellations also confirm the behaviors of EVM and BER when increasing the pump power. The distortion from SBS works mainly as amplitude noise in the constellations, and becomes severe when the pump power is increased.

 

Fig. 4 (a) Measured EVMs (triangles) and BERs (squares) and (b) corresponding conversion efficiencies with different launched pump power.

Download Full Size | PDF

 

Fig. 5 Measured constellations with pump power of (a) 21.6dBm and (b) 22dBm and (c) 22.3dBm.

Download Full Size | PDF

As discussed in [11–13], with limited OSNR of pump source, the pump may introduce nonlinear phase noise to the converted signal through XPM in AOWC. In our experiment, thanks to the low relative intensity noise (RIN) of the FL, the OSNR of pump source was measured as around 57dB, which effectively avoids the pump-induced nonlinear phase noise. The dominant distortion from the pump is mainly from SBS when operating the AOWC with high pump power.

Figure 6 shows the measured optical spectrum after the HNLF with the optimal launched pump and probe power at 20 and 9 dBm, respectively. The conversion efficiency of around −15 dB with respect to the probe power is observed. As shown in Fig. 7, BER performance as a function of OSNR at 0.1 nm is measured for both input and converted 64QAMs under the optimized operation conditions. The implementation penalty compared with the theoretical BER curve is around 2.8 dB at BER of 10⁻³, which is much better than those of the reported 64QAM transmitters in [1,2]. The implementation penalty is mainly attributed to the imperfectness in the transmitter. A negligible power penalty (<0.3dB) is observed for the converted signal at BER of 10⁻³ with respect to the input signal. The measured constellations of input and converted 64QAMs at the received OSNR of around 35 dB are shown in the insets of Fig. 7.

 

Fig. 6 Measured optical spectrum after HNLF.

Download Full Size | PDF

 

Fig. 7 Measured BER results of input (blue squares) and converted (red triangles) 64QAM signals as a function of OSNR (at 0.1nm) with the theoretical BER curve of 64QAM provided as reference (solid line). Insets: the constellations of input (left) and converted (right) signals 64QAMs.

Download Full Size | PDF

3. Conclusions

We have experimentally demonstrated the AOWC of 60-Gbps 64QAM through FWM in HNLF with negligible power penalty (<0.3dB at BER of 10⁻³). To eliminate the extra nonlinear distortions such as SPM and SBS, it is suggested to optimize the operation condition by monitoring the constellations (EVM), rather than BER. The constellation monitoring provides an intuitive and accurate approach to monitor the extra distortion in the nonlinear process, especially in the presence of nonlinear phase distortion like SPM.