Open Access
Add to BibSonomy
Abstract
The aggregation and de-aggregation between one QPSK and two BPSK signals are experimentally demonstrated based on cross-phase modulation (XPM) effects and phase-sensitive amplification (PSA), aiming to improve flexible spectral efficiency in the elastic optical network. Benefiting from the nonlinear-optical loop mirror (NOLM), the two BPSK signals can be extracted without any information redundancy for the de-aggregation scheme. The wavelength of the extracted signals stay the same as the input signal in the NOLM. Moreover, the aggregation from two BPSK signals to one QPSK signal is also successfully achieved. The feasibility of the two schemes can be confirmed by the transfer functions and the input-output constellation. The bit-error-rate (BER) and error vector magnitude(EVM) performance of the two schemes are also investigated and the corresponding OSNRs for error-free signal recovery are obtained respectively. The proposed schemes contribute to the realization of flexible optical networks and can be applied in the future gateway node between optical sub-networks.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
With the Internet-driven traffic such as Big Data, Internet of Things (IoT), and 4K/8K high-definite video, the growth rate of service traffic has far outpaced the growth rate of optical network capacity [1]. IP-based data services are beginning to replace traditional voice services as an important service type. Different from traditional services, IP services have the characteristics of burstiness and uncertainty. These characteristics require the optical network to have a smaller switching granularity and dynamically allocate bandwidth. The elastic optical networks (EON) with the advantage of improving flexible spectral efficiency and resource allocation are considered as an important architecture to fulfill the requirements of the ultra-large capacity, ultra-high speed, ultra-long distance and ultra-wideband flexibility [2]. The de-aggregation of the high-order signal into multiple lower-order signals and inverse conversions from multiple lower-order signals into a higher-order signal will be key technologies in EON.
Conventional aggregation and de-aggregation are performed in the electrical domain, where signal demodulation and re-modulation are necessary. It brings a higher cost to handle the high-speed traffic via photoelectric conversion equipment. In addition, some digital signal processing (DSP) algorithms during coherent demodulation require the signal sample rate to be twice the signal rate to ensure that the original high-speed traffic is correctly demodulated which severely limits the symbol rate of the signal due to the electronic rate bottleneck. The all-optical approach based on nonlinear effect can avoid inefficient optical-electronic-optical(OEO) conversion and operate at the line rate of the data with femtosecond response time, which will greatly improve the information processing rate and network transmission quality.
High-order modulation formats, e.g., multi-level phase-shift-keying (MPSK) or multi-level quadrature amplitude modulation(MQAM), have become one of the direct ways to alleviate capacity pressure due to their high spectrum efficiency [3,4]. Some schemes about the conversions from multiple low-order signals to a high-order signal have been investigated. For example, multiple OOK signals to MPSK signal via XPM [5], OOK and QPSK to 8 quadrature amplitude modulation(8QAM) using XPM and XGM in SOA [6], amplitude shift keying(ASK) and QPSK to 8QAM via FWM in highly nonlinear fiber [7,8], two BPSKs to PAM-4 [9] and three QPSK to 64QAM [10] via nonlinear wave mixing in periodically poled lithium niobate (PPLN) waveguide. However, the aggregation among the PSK and ASK signals are rarely used, which imposes a lot of restrictions on practical applications. Moreover, the schemes mentioned above only focus on the optical scheme and have no corresponding de-aggregation schemes.
As the opposite direction, the conversions from a high-order signal to multiple low-order signals have also been investigated. The QPSK is not only the standard format in 100G networks but also the candidate for the 400G networks along with 8QAM and 16QAM [3]. A DQPSK signal can be de-multiplexed to a DPSK signal via phase erasure [11]. But only the Q part of the QPSK signal can be extracted and the other I part of the QPSK cannot be recovered. QPSK signal also can be de-aggregated to two BPSK signals using four-wave mixing(FWM) [12] or PSA [13,14]. In addition, other high-order modulation signal de-aggregation schemes have also been studied. An 8PSK signal can be de-aggregated into two QPSK signals by PSA [15]. Based on the same PSA scheme, a 16QAM or 64QAM signal can be de-aggregated into two PAM-4 or PAM-8 signals with wavelength and polarization preservation [16,17]. These schemes normally require a large number of phase-locked pump signals or multiple non-linear processes for extracting the I and Q parts of the signal. And many of the solutions currently only give simulation results without experimental verification.
In this paper, we experimentally demonstrate the all-optical aggregation and inverse conversion of the QPSK signal. For the aggregation scheme, the two BPSK signals at different frequencies called BPSK1 and BPSK2 signal are employed as the input signals. The BPSK1 signal and intensity signal converted from the BPSK2 signal via delay interferometer (DI) can be aggregated to the QPSK signal by XPM in SOA. The input-output constellations, BER, and EVM of the signals before and after the aggregation are obtained to evaluate the performance. In the corresponding de-aggregation process, a bi-directional PSA-based optical de-aggregator scheme is applied to de-aggregate the 10Gbaud QPSK signal. The bi-directional PSA-based scheme can extract the I and Q components simultaneously with wavelength and polarization preservation. The transfer characteristics of the de-aggregation system and the input-output phase state demonstrate the phase squeezing effect. The performance of the scheme is evaluated in the same way as the aggregation scheme.
2. Operation principle
2.1 All-optical aggregator
To demonstrate the sub-signal aggregation, the scheme of a all-optical aggregator has been illustrated. Fig.1 shows the configuration of two BPSK signals with different frequencies aggregating into a QPSK signal. This process is achieved in two steps. First, the BPSK2 signal is converted to an intensity signal utilizing a DI. Then the BPSK1 and intensity signal are aggregated into a QPSK signal based on XPM.
As shown in Fig. 1, the BPSK2 is converted to an atypical OOK signal using a 1-bit delay interferometer [18,19]. Assume the ’1’ and ’0’ represent two bits of the BPSK signal respectively and the coordinates of the two constellation points are represented as (1,0) and (-1,0). After a 1-bit DI, the delayed signal is superimposed on the original signal and the coordinates after coherent addition are (2,0),(0,0), and (-2,0). Therefore, the BPSK signal is converted to an intensity signal with two intensities and the probability of each intensity is 1/2. Then the converted OOK signal is injected into SOA as a pump signal together with BPSK1 as a probe signal. The pump signal with two power values induces the counterclockwise phase shift on the probe signal due to the XPM effect in SOA. The OOK signal will modulated the gain in SOA by varying the carrier density. And the nonlinear phase change due to the change of the refractive index in the SOA is expressed by the following equation [20]:
Where $\bigtriangleup L$ is the section length, $\Gamma$ is the confinement factor, $\alpha$ is the material gain constant, $\lambda$ is the wavelength, $\ g_{m}$ is the material gain, i is the number of the section, $\bigtriangleup \overline {n}_{N}$ is the rate of change of the refractive index in the active region with carrier concentration. When the power of OOK is 0, the phase shift induced by XPM is 0. The BPSK1 signal and OOK signal can be aggregated into a QPSK signal, by controlling the non-zero power of the OOK signal to induce a $\pi / 2$ phase shift on the probe signal. The information of the aggregated QPSK entirely comes from the input BPSK signals without any information loss or redundancy. The specific bit mapping relationship are shown in Fig. 1. The first bit of the QPSK signal comes from the OOK signal after differential interference and the second bit comes from the BPSK1 signal. Similarly, the mapping and logical relationship between the input BPSK signals and the aggregated QPSK signal are summarized in Table 1
Fig. 1. (a) Scheme of the optical aggregator from Two BPSKs to QPSK;(b) Spectrum of the aggregation process.
Table 1. The logical and phase relationship in aggregation
2.2 All-optical de-aggregator
To demonstrate the sub-signal accommodation, an all-optical de-aggregation process will be shown. The optical vector signals like QPSK are generated by the coherent addition of optical vectors that respectively carrying the I and Q parts of the signal. The de-aggregator is designed to de-aggregate the QPSK signal into two BPSK signals simultaneously without involving a lot of non-linear processes. The BPSK1 signal carrying the I part of the signal can be recovered by squeezing the phase state of two adjacent points to $\pi / 4$ and $3 \pi / 4$ as shown in Fig. 2($b$). Likewise, the other part of the signal can be recovered by adjusting the gain axis as shown in Fig. 2($c$). The logical and phase relationships in the de-aggregation process are summarized in Table 2. According to Table 2, the extract binary data are the same as the input QPSK signal.
Fig. 2. Constellation diagrams of (a) input QPSK signal,(b) recovered BPSK1, and (c) recovered BPSK2 signal.
Table 2. Logical and phase relationship between Input and Recovered Signals
The schematic of the optical de-aggregator applicable for de-aggregating QPSK into I and Q parts is depicted in Fig. 3. It consists of a nonlinear-optical loop mirror (NOLM) based on a SOA and a 90$^{\circ }$ optical hybrid. The NOLM is used to extract the original signal(S) and the idler signal(I) by four-wave mixing (FWM)in SOA. The input signal(S) and pump1 (P1) are combined to inject into port A and pump2(P2) is incident to port B in the opposite port. The dual-pumps FWM between the signal and pumps will occur in SOA. Then the signal and the generated idler output from different ports. Fig. 3($b$) shows the spectrum of the degenerate FWM process in the SOA. After the SOA, the converted signal at $\omega _{s}$ can be given by:
Fig. 3. (a) Scheme of the optical de-aggregator from QPSK into two BPSK signals;(b) Spectrum of the dual-pump PSA-based de-aggregator process.
As shown in Fig. 3($a$), the electrical field of the launched signals at the port A and B can be expressed as:
Fig. 4. Phase-to-phase and phase-to-amplitude transfer functions for QPSK de-aggregation to two BPSKs.
3. Experiments setup and results
3.1 Aggregation from two BPSKs to QPSK
The experimental setup of two BPSK signals to a QPSK signal is illustrated in Fig. 5. The BPSK1 signal with frequency 193.375THz is generated by modulating a CW using an MZM modulator with automatic bias control. Likewise, the BPSK2 signal with frequency 193.4THz is generated by the other MZM modulator, by applying a pseudorandom bit sequence (PRBS) of length $2^{15}-1$ from an arbitrary waveform generator(AWG, Tektronix). The BPSK2 signal is converted to an atypical OOK signal after a 1-bit delay interferometer. The atypical OOK signal is the three constellation points with two intensities and the probability of each intensity is 1/2. The power of the BPSK1 and the other OOK signal are set 5.3 dBm and 12.5 dBm respectively by EDFAs. The two amplified signals are combined by a 3-dB coupler and launched into the SOA with a signal gain of 30 dB and a saturation output signal power of 6 dBm work at a bias current of 200mA. The aggregated QPSK signal with a power of -2.7dBm and OSNR exceeds 30 dB is obtained at the output of the SOA as shown in the insertion diagram in Fig. 5. A tunable band-pass filter (OTF-970, santec) is inserted after the SOA to filter the aggregated QPSK signal. Coherent detection and offline DSP process are applied in our scheme to recover the constellation map and calculate the bit error rate. In the DSP process, we adopt frequency estimation and phase recovery based on the Viterbi-Viterbi algorithm and do not use any equalizer.
The corresponding input-output constellation maps of the original BPSK1 signal, converted OOK signal and converted QPSK signal after the offline DSP process are illustrated in Fig. 6. The error vector magnitude(EVM) of the input BPSK signal is 14. $33 \%$ and is dropped by 5. $33 \%$ for the aggregated QPSK signal, which mainly comes from the noise conversion from amplitude noise to phase noise in XPM. Some dislocation can be observed in constellation points of the converted QPSK signal compared with the standard QPSK constellation map. The reasons can be attributed to imperfect modulation and the delay jitter of the delay interferometer. As shown in Fig. 6, the modulated BPSK signal is not a circular constellation distribution but a long strip compared with the ideal BPSK signal which results in the signal after aggregation is not so standard. Moreover, with the reduction of the OSNR, the performance of the converted OOK signal due to the decreasing euclidean distance is getting worse than the BPSK1 signal, which will also affect the conversion efficiency. To further evaluate the performance of the XPM effect, the constellation maps of the aggregated QPSK signal under different pump powers are investigated. The phase offset of the XPM effect depends on the power of the pump signal, an appropriate pump power is preferred in the scheme to inducing a $\pi / 4$ phase offset. We investigate the performance of the aggregation by adjusting the output power of the EDFA2. The EVMs of the QPSK signals increase 2. $86 \%$ and 5. $62 \%$ respectively when the pump power is boosted to 14 dBm and 16 dBm compared with the pump power at 12.5 dBm. According to the the EVM performance of constellation maps in Fig. 6 and Fig. 7, better performance can be obtained when the pump is boosted to 12.5 dBm. Moreover, it is worth noting that high pump power will induce a large phase shift result in the distortion of constellation points [7]. So we choose the relatively low pump power for better performance.
For evaluating the noise tolerance of the generated signal, the BER performance of the aggregator is illustrated in Fig. 8 and the OSNR is degraded by the addition of ASE noise for measuring BER curves. The two BPSK signals represent the original input signals, the B2B-QPSK signal is the QPSK signal under back-to-back and the C-QPSK signal is the QPSK signal after the aggregation process. According to the forward error correction (FEC) threshold of $3.8 \times 10^{-3}$, the signal cannot be corrected when the log(BER) is higher than -2.4. As shown in Fig. 8, the BERs of the two input signals up to FEC are 7.54dB and 8dB respectively, which shows better BER performance due to the larger euclidean distance. But the BER performance of the BPSK signal shows a penalty of almost 5dB compared with the theoretical value [23]. The reason can be the imperfect modulation and limited extinction ratio. The BER performance of the converted QPSK signal is below the FEC when OSNR exceeds 13.2 dB. Compared with the B2B QPSK signal, there is almost a 2.62 dB OSNR penalty at the FEC threshold for aggregated signal, which attributes to a combination of ASE noise accumulation and phase noise. The XPM process is insensitive to the phase and only determined by the amplitude. So the intensity noise of the pump signal will transfer to the phase noise. The converted OOK signal by DI behaves worse noise tolerance than the BPSK signal which in turn affects the quality of the aggregated QPSK signal.
3.2 De-aggregation from QPSK to two BPSKs
The scheme of the QPSK signal to two BPSK signals format conversion is shown in Fig. 9. A CW signal from a tunable narrow-linewidth external cavity laser at 193.4THz is employed to generate an optical frequency comb with a spacing of 25GHz by phase modulator. Three phase-locked pilot tones including a signal(S) and two pumps (P1, P2) positioned -25GHz and +25GHz relative to S are extracted using a wavelength selective switch(WSS, finisar). The signal (S) with the frequency of 193.4THz is modulated with a 10 Gbaud pseudo-random binary sequence (PRBS) to generate a QPSK signal. Data is generated in a high-speed arbitrary waveform generator(AWG). To keep the modulator in a safe working area, the amplitude of the AWG output is set at 300mV peak-to-peak. Then the modulated signal and pump P1 are amplified respectively by EDFA and combined by a 3-dB coupler to launch into the lower path of the NOLM. Likewise, the P2 at 193.425THz is amplified and injected into the upper path of the NOLM. Their state of polarization is controlled by PC for the best performance. The NOLM consists of a 3-dB coupler, two circulators, and an SOA. The SOA with a signal gain of 30 dB and a saturation output signal power of 6 dBm work at a bias current of 200mA. Optical band-pass filters are used to remove the disturbance signals. Moreover, a variable optical attenuator (VOA) is employed for controlling the relative power between the signal and the conjugated signal. Finally, the signal and the generated idler signal are launched into the 90$^{\circ }$ optical hybrid respectively for interference in two different paths. Two BPSK signals carrying the I and Q parts of the signal can be obtained simultaneously from ports 90$^{\circ }$ and 270$^{\circ }$ and received by respective coherent receivers. In addition, to lock the waves in phase, an optical phase-locked loop is applied in the scheme configuration. 10 $\%$ of the idler signal is detected by the PD and used as a reference signal for the feedback loop based on the PZT to stabilize the relative phase relationship.
Fig. 9. Experimental setup for QPSK de-aggregation. WSS, wavelength-selective switch; PM, phase modulator; EDFA, erbium-doped fiber amplifier; PC, polarization controller; SOA, semiconductor optical amplifier; BPF, band-pass filter; VOA, variable optical attenuator.
The spectrum at the lower node a when the signal is combined with the pump P1 is shown in Fig. 10($a$). and the spectrum depicting the pump P2 is shown in Fig. 10($b$). The modulated signal and pump P1 are boosted up to 9.75dBm and 16.14 dBm respectively by EDFAs. Likewise, the P2 at 193.425THz is boosted up to 16.5 dBm. The pumps had an extinction ratio of about 40 dB(limited by the noise figure of the EDFA). The power level of the generated S* can be measured around 24dB less than the original input QPSK signal, indicating the conversion efficiency of the dual-pump FWM in Fig. 10($c$). In addition, the other frequency pilot tones are the FWM products that are generated by the pump interaction.
The constellations of the input QPSK signals and corresponding I and Q de-aggregated parts are shown in Fig. 11. The input OSNR was degraded by combining with the output of an amplified spontaneous emission (ASE) source to the QPSK signal. The results exhibit the feasibility of de-aggregating the QPSK signal into two BPSK signals with the use of the NOLM-based de-aggregator. For the high-quality QPSK signal without additional noise, the input QPSK signal with EVM of 13. $98 \%$ is de-aggregated into two BPSK signals with EVMs of 17. $27\%$ and 18. $23\%$. For the low-quality QPSK signal with OSNR at 16 dB, the input QPSK signal with EVM of 26. $59\%$ is de-aggregated into two BPSK signals with EVMs of 32. $15\%$ and 33. $1\%$. The constellation of the BPSK2 reflects a bit worse conversion efficiency than the I-component. The reason can be imperfect power controlling in the de-aggregation and unoptimized DSP calculation. Limited by the equipment in hand, imperfect modulation of the input QPSK signal leads to the EVMs of the constellations worse than normal. The efficiency can be greatly increased if the modulation of the signal can be further optimized.
Fig. 11. The constellation of the original QPSK, the de-aggregated I and Q parts after the DSP process.
Fig. 12($a$) shows the EVM of the recovered BPSK signals under different input QPSK signal OSNR values from 14 dB to 24dB. After the de-aggregation, the BPSK signals have about less than a $10 \%$ penalty. The EVMs gap between the recovered BPSK signals and the input QPSK signal is getting smaller as the input OSNR increases because of limited phase squeezing effect. Fig. 12($b$) shows the phase noise(PN) and the amplitude standard deviation(ASD) of the recovered BPSK signals with varying OSNR. The phase noise of the recovered BPSK signals is squeezed to a level less than 20 because of the function of the phase squeezing according to the transfer curve shown in Fig. 4. However, the ASDs of the recovered BPSK signals demonstrate an obvious degradation, which means more amplitude noise will generate as the OSNR of the input signal decreases. The reason can be the conversion from phase noise to amplitude noise due to the phase squeezing characteristics in the PSA as shown by the amplitude gain curve in Fig. 4.
Fig. 12. (a)EVM of the recovered BPSK signals versus input OSNR;(b)The PN and the ASD of the recovered BPSK signals versus input OSNR.
To further evaluate the performance of the de-aggregation setup, the BER measurement of the de-aggregator is illustrated in Fig. 13. and comparing with back-to-back performance for both a BPSK signal and QPSK signal. According to the forward error correction (FEC) threshold of $3.8 \times 10^{-3}$, the signal cannot be corrected when the log(BER) is higher than -2.4. The reference BPSK signals were demodulated respectively and the obvious BER sensitivity improvement at FEC can be observed. At the FEC threshold, the OSNR penalty between the recovered BPSK signals and the reference BPSK signals is 1.32 dB and 1.56 dB. The penalty is mainly attributed to the in-band ASE noise of the employed amplifiers and the unstable polarization state in the experiment. Comparing with the B2B QPSK signal, the required OSNRs for BPSK signals reduce to 8.56 dB and 8.93 dB at the FEC. The result benefits from the improved noise tolerance in the de-aggregation process due to the phase squeezing effect. For better performance, we can adopt superior optical amplifiers with stable output power and low noise figure for making up for losses in the line. In addition, the precise control of polarization state, the appropriate wavelength spacing, and the power of the pumps should also be noted.
Fig. 13. BER performance of the de-aggregator for the recovered signals together with back-to-back QPSK and BPSK as references.
4. Conclusion
In this article, we have carried out experiments on all-optical aggregation and de-aggregation. To evaluate the performance of the aggregation from two BPSK signals to QPSK signal, the input-output constellations, the EVM of the input-output signal, and the BER performance have been taken into consideration. For error-free recovery of the aggregated signal, the OSNR of the QPSK signal exceeds 13.2 dB. Comparing with the B2B QPSK signal, there is almost a 2.62 dB OSNR penalty at the FEC threshold for aggregated signal. Moreover, this scheme can also be applied to the optical modulator of the MPSK/MQAM. For example, the 8PSK signal can be generated by aggregating the BPSK and QPSK signal based on this scheme.
In addition, de-aggregation from QPSK to two BPSK signals was also achieved. By employing the NOLM, we can extract the I and Q parts of the input QPSK signal. To evaluate the performance of the proposed scheme, the input-output constellations and the BER performance of the system have been analyzed respectively. According to the results of the experiment, the OSNR penalty between the recovered BPSK signals and the reference BPSK signals is 1.32 dB and 1.56 dB. Furthermore, this architecture can be applied to more de-aggregation schemes. As for the QPSK signal instead of being modulated but all-optical aggregation based on XPM, this scheme is also applicable. We can recover the OOK signal and BPSK signal by changing the output port of the 90$^{\circ }$ optical hybrid. Moreover, this can also be applied to orthogonal de-aggregation of the 16QAM into two PAM-4 signals. In future work, we will consider experimental verifications of all-optical de-aggregation schemes for WDM signals and simultaneous all-optical aggregation and de-aggregation processes.
Funding
National Key Research and Development Program of China (No. 2019YFB1803601); Beijing Natural Science Foundation (Z210004).
Disclosures
The authors declare no conflicts of interest.
Data Availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. Y. Ji, R. Gu, Z. Yang, J. Li, H. Li, and M. Zhang, “Artificial intelligence-driven autonomous optical networks: 3S architecture and key technologies,” Sci. China Inf. Sci. 63(6), 160301 (2020). [CrossRef]
2. Y. Ji, H. Wang, J. Cui, M. Yu, Z. Yang, and L. Bai, “All-optical signal processing technologies in flexible optical networks,” Photonic Netw. Commun. 38(1), 14–36 (2019). [CrossRef]
3. P. J. Winzer, “Beyond 100G ethernet,” IEEE Commun. Mag. 48(7), 26–30 (2010). [CrossRef]
4. M. Mazur, A. Lorences-Riesgo, J. Schroeder, P. A. Andrekson, and M. Karlsson, “High spectral efficiency PM-128QAM comb-based superchannel transmission enabled by a single shared optical pilot tone,” J. Lightwave Technol. 36(6), 1318–1325 (2018). [CrossRef]
5. 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]
6. H. Kishikawa, M. Uetai, and N. Goto, “All-Optical Modulation Format Conversion Between OOK, QPSK, and 8QAM,” J. Lightwave Technol. 37(16), 3925–3931 (2019). [CrossRef]
7. B. Zhang, H. Zhang, C. Yu, X. Cheng, Y. K. Yeo, P.-K. Kam, J. Yang, H. Zhang, Y.-H. Wen, and K.-M. Feng, “An all-optical modulation format conversion for 8QAM based on FWM in HNLF,” IEEE Photonics Technol. Lett. 25(4), 327–330 (2012). [CrossRef]
8. F. Wan, B.-J. Wu, F. Wen, and K. Qiu, “All-optical modulation format conversion from star-QAM to PSK and ASK signals,” Opt. Commun. 451, 23–27 (2019). [CrossRef]
9. A. Fallahpour, F. Alishahi, K. Zou, Y. Cao, A. Almaiman, A. Kordts, M. Karpov, M. H. P. Pfeiffer, K. Manukyan, H. Zhou, P. Liao, M. Tur, T. J. Kippenberg, and A. E. Willner, “Demonstration of tunable and reconfigurable optical Nyquist channel aggregation of QPSK-to-16QAM and BPSK-to-4PAM using nonlinear wave mixing and a Kerr frequency comb,” in 2019 Optical Fiber Communications Conference and Exhibition (OFC), (IEEE, 2019), pp. 1–3.
10. A. Fallahpour, M. Ziyadi, A. Kordts, C. Bao, P. Liao, A. Mohajerin-Ariaei, M. Karpov, M. H. P. Pfeiffer, Y. Cao, A. Almaiman, F. Alishahi, B. Shamee, L. Paraschis, M. Tur, C. Langrock, M. M. Fejer, J. Touch, T. J. Kippenberg, and A. E. Willner, “Experimental generation of a 64-QAM by optically aggregating three independent QPSK channels using nonlinear wave mixing of multiple Kerr comb lines,” in CLEO: Applications and Technology, (Optical Society of America, 2017), pp. JTh2A–59.
11. G.-W. Lu and T. Miyazaki, “Optical phase erasure based on FWM in HNLF enabling format conversion from 320-Gb/s RZ-DQPSK to 160-Gb/s RZ-DPSK,” Opt. Express 17(16), 13346–13353 (2009). [CrossRef]
12. Y.-H. Wen, J.-W. Ho, and K.-M. Feng, “Phase transparent demultiplexer for QPSK signal based on nonlinear response in a double-pass architecture,” IEEE Photonics J. 8, 1–7 (2016). [CrossRef]
13. M. Gao, T. Kurosu, T. Inoue, and S. Namiki, “Efficient quadrature squeezing of QPSK signals by sideband-assisted dual-pump phase sensitive amplifier,” in Optical Fiber Communication Conference, (Optical Society of America, 2013), pp. OW3C–1.
14. A. Lorences-Riesgo, L. Liu, S. L. Olsson, R. Malik, A. Kumpera, C. Lundström, S. Radic, M. Karlsson, and P. A. Andrekson, “Quadrature demultiplexing using a degenerate vector parametric amplifier,” Opt. Express 22(24), 29424–29434 (2014). [CrossRef]
15. H. Liu, H. Wang, J. Cui, and Y. Ji, “Design of all-optical modulation format converter from one 8PSK to two QPSK signals based on phase sensitive amplification in elastic optical network,” IEEE Access 7, 51379–51385 (2019). [CrossRef]
16. J. Cui, G.-W. Lu, H. Wang, and Y. Ji, “On-chip optical vector quadrature de-multiplexer proposal for QAM de-aggregation by single bi-directional SOA-based phase-sensitive amplifier,” IEEE Access 7, 763–772 (2018). [CrossRef]
17. J. Cui, G.-W. Lu, H. Wang, L. Bai, and Y. Ji, “Single-stage phase-sensitive amplifier based quadrature de-multiplexer for de-aggregating qam signals into in-phase and quadrature components,” in CLEO: Applications and Technology, (Optical Society of America, 2018), pp. JTu2A–36.
18. M. Matsumoto, “A fiber-based all-optical 3r regenerator for dpsk signals,” IEEE Photonics Technol. Lett. 19(5), 273–275 (2007). [CrossRef]
19. H. Wang, L. Pan, and Y. Ji, “All-optical aggregation and de-aggregation of 4× bpsk-16qam using nonlinear wave mixing for flexible optical network,” IEEE J. Sel. Top. Quantum Electron. 27, 7700408 (2019). [CrossRef]
20. K. Mishina, S. M. Nissanka, A. Maruta, S. Mitani, K. Ishida, K. Shimizu, T. Hatta, and K.-i. Kitayama, “All-optical modulation format conversion from NRZ-OOK to RZ-QPSK using parallel SOA-MZI OOK/BPSK converters,” Opt. Express 15(12), 7774–7785 (2007). [CrossRef]
21. H. Liu, H. Wang, Z. Xing, and Y. Ji, “Simultaneous all-optical channel aggregation and de-aggregation based on nonlinear effects for OOK and MPSK formats in elastic optical networking,” Opt. Express 27(21), 30158–30171 (2019). [CrossRef]
22. J. Cui, Y. Ji, G.-W. Lu, H. Wang, and M. Zhang, “Phase-sensitive amplifier-based optical conversion for direct detection of complex modulation format to bridge long-haul transmissions and short-reach interconnects,” Opt. Express 28(2), 2349–2365 (2020). [CrossRef]
23. R.-J. Essiambre, G. Kramer, P. J. Winzer, G. J. Foschini, and B. Goebel, “Capacity limits of optical fiber networks,” J. Lightwave Technol. 28(4), 662–701 (2010). [CrossRef]
