We experimentally demonstrate fiber nonlinearity compensation in dual polarization coherent optical OFDM (DP CO-OFDM) systems using mid-span spectral inversion (MSSI). We use third-order nonlinearity between a pump and the signal in a highly nonlinear fiber (HNLF) for MSSI. Maximum launch powers at FEC threshold for two 10 × 80-km 16-QAM OFDM systems were increased by 6.4 dB at a 121-Gb/s data rate and 2.8 dB at 1.2 Tb/s. The experimental results are the first demonstration of using MSSI for nonlinearity compensation in any dual polarization coherent system. Simulations show that these increases could support a 22% increase in total transmission distance at 1.2-Tb/s system without increasing the number of inline amplifiers, by extending the fiber spans from 90 to 110 km. When spans of 80 km are used, simulations reveal that MSSI system performance shows less degradation with increasing transmission distance, and an overall transmission distance increase of more than 70% is expected using MSSI.
© 2014 Optical Society of America
Coherent optical orthogonal frequency division multiplexing (CO-OFDM) has emerged as a promising candidate for next generation optical communication systems . It inherently offers virtually unlimited chromatic dispersion (CD) and polarization mode dispersion (PMD) tolerance . Therefore, the capacity for long-haul optical systems is presently limited by fiber nonlinearity [3–5]. For best system performance, the launch power into a fiber span should be set below the nonlinear threshold (NLT), which restricts the signal to noise ratio at the receiver and hence the size of the constellation that can be used [3–5]. Mitigating the effects of fiber nonlinearity allows higher launch powers, which increases reach and/or spectral efficiency.
Previously, mid-span spectral inversion (MSSI) has been shown to be effective for nonlinearity compensation in intensity modulated systems [6, 7] and differential phase-shift-keyed (DPSK) systems . MSSI is scalable for WDM systems because multiple wavelengths can be spectrally inverted simultaneously using a single optical phase conjugation (OPC) module . Recently, both theoretical and simulation results for MSSI with CO-OFDM have been reported [10, 11]. Our group has previously reported the first experimental results for MSSI in single-polarization CO-OFDM systems [12, 13], which to the best of our knowledge represents the first use of MSSI for coherent communication systems. At the Opto-Electronics and Communications Conference 2013 (OECC 2013) , we experimentally demonstrated MSSI for fiber nonlinearity compensation in a dual-polarization (DP) CO-OFDM system for the first time, at a rate of 1.2 Tb/s. This demonstration is, to the best of our knowledge, the first for dual-polarization MSSI for any coherent communication system. Subsequently, there have been demonstrations of Raman-enhanced MSSI [15, 16] and multiple phase-conjugation based coherent systems  at the 2014 conference on Optical Fiber Communications (OFC).
In this paper, we elaborate on the results in  and the MSSI module for DP signals, which is the key enabling system in this experiment . We experimentally demonstrate the benefit of MSSI for 121-Gb/s and 1.2-Tb/s DP CO-OFDM MSSI systems for fiber nonlinearity compensation, and give an overview of important design considerations. These results are the first for DP coherent systems. Additionally, we have simulated the system to define the benefits gained from the increase in maximum allowable power using MSSI. We also show the simulation results for improving the performance the OPC module itself.
2. System carrying 121 Gb/s
2.1 Experimental setup
Figure 1(a) shows the transmitter configuration for 121-Gb/s system. An external cavity laser (ECL, 193.1 THz) provides the optical carrier. The external cavity lasers used throughout this paper are specified to have a linewidth less than 100 kHz. A 30.3-Gb/s OFDM signal band is created using 194-point inverse fast Fourier transforms (IFFT) with 154 16-QAM modulated subcarriers and a 9-point cyclic prefix (CP). A 10-GS/s arbitrary waveform generator (AWG) is used for digital to analog conversion, producing 20.3-ns long, 7.94-GHz bandwidth OFDM symbols.
A 20-GHz optical bandwidth complex Mach-Zehnder modulator (C-MZM) modulates the electrical OFDM signal onto the output of the ECL. The optical signal is then divided into two paths; one path is frequency shifted by 8 GHz using another C-MZM as a serrodyne modulator, then the shifted and un-shifted signals are combined to form a continuous 16-GHz wide channel, as shown in inset (i) of Fig. 1(a). The spectrum also shows the optical carrier at the center of each band. In order to maintain the optical carrier at the center, the two central subcarriers of the electrical OFDM signal were nulled. A delay of exactly 5 OFDM symbols (101.5 ns) produces an integer-OFDM-symbol delay between the shifted and un-shifted signals. The channel is then passed through a POLMUX emulator (Kylia PDME-00019), with a 19.8-ns delay between the two orthogonal polarizations to de-correlate them, to create a 121-Gb/s DP signal.
Figure 1(b) shows the receiver configuration, which is the same for the 121-Gb/s and 1.21-Tb/s systems. A wavelength selective switch (WSS, Finisar Waveshaper) is used to remove the out-of-band noise. The filtered signal is fed into the signal port of a coherent receiver, consisting of an optical hybrid (Kylia mint-2 × 8) and 25-GHz bandwidth balanced photodiodes. A second ECL, tuned to match the frequency of either the original signal for reference system without MSSI (193.1 THz) or the conjugate idler for the MSSI system (193.5 THz), is used as the local oscillator. A 40-GSample/s, 16-GHz bandwidth real-time sampling oscilloscope (Agilent DSO-X 92804A) digitizes the photo-detected signals. The signal is processed offline, with the equalizer comprising : a resampler; a frequency offset compensator; a butterfly OFDM 1-tap equalizer ; and a blind symbol phase estimator [18, 19]. In the system without MSSI, digital chromatic dispersion (CD) compensation precedes the 1-tap equalizer.
Figure 1(c) shows the details of the optical link with MSSI module. The link comprises 10 × 80-km standard single-mode fiber (S-SMF) spans with variable output power erbium doped fiber amplifiers (EDFA) to compensate for the loss of each span. The MSSI module is placed after the fifth span. The launch power into each span is set by adjusting the output power of the EDFAs. The EDFAs have specified maximum noise figures of 7 dB. A dispersion compensation fiber (DCF, Siemens BDCM-60, 1011 ps/nm, 12 dB loss), placed after the fifth span, is used to improve the fiber nonlinearity compensation . After the DCF, another EDFA amplifies the signal to 18 dBm. Then the signal is filtered by a band pass filter (BPF, 5 nm bandwidth) before being fed into the 10%-port of a 90%/10% coupler. The pump from a third ECL (193.3 THz), set to 16 dBm output power, is input to the 90% port of the coupler. The pump power is kept at 16 dBm, because when the MSSI unit was pumped with higher powers the system’s stability was found to degrade, indicating significant stimulated Brillouin scattering (SBS). To achieve efficient generation of the OPC signal, the wavelength of the pump is tuned slightly above the zero dispersion wavelength (ZDW) of the HNLF, to minimize the phase mismatch between the interacting waves .
The combined signal and pump wave is passed to a dual-polarization MSSI module , as recently used to demonstrate pre-compensation of fiber nonlinear effect in 80-Gb/s RZ-DPSK dual polarization signals . Note that this polarization diverse configuration is insensitive to the polarization state of the input signal [22, 23]. The combined waves feed Port 1 of an optical circulator. Port 2 of the circulator connects to the common port of a polarization beam splitter (PBS). The two polarized (X Pol. and Y Pol.) ports of the PBS are interconnected via a polarization controller (PC), 1 km of highly nonlinear fiber (HNLF) and a 99%/1% coupler for power monitoring. After the insertion losses of the circulator, PBS and polarization controller, the pump and signal powers launched into the HNLF module from each direction are 12 dBm and 4 dBm respectively. The HNLF has a nonlinear coefficient, γ, of 11.5 W−1km−1, CD of 0.01 ps/nm/km at 1550 nm, CD slope of 0.02 ps/nm2/km, zero-dispersion-wavelength (ZDW) of 1549.120 nm and loss coefficient of 0.81 dB/km. Port 3 of the circulator connects to a 90%/10% coupler; the 10% port connects to an optical spectrum analyzer (OSA): the 90% port connects to a 200-GHz channel spacing demultiplexer (Siemens TransXpress), which selects the conjugated signal and removes the original signal, the pump and the out of band amplified spontaneous emission (ASE). The output of the demultiplexer is transmitted through the second half of the link.
Polarization controllers can be avoided entirely by using polarization maintaining fiber in the MSSI module, to ensure alignment of the pump wave to be 45° to the reference polarization axis of the polarization beam splitter and linear polarization states at input to the HNLF. However, our demonstration used non-PM HNLF, so required polarization controllers. PC3 is used to adjust the polarization of the travelling waves to be linearly polarized upon launch into the HNLF to maximize conversion efficiency (CE), which is continuously monitored at the OSA connected at Port 3 of the circulator, using a 90%/10% coupler. PC1 and PC2 can then be adjusted to reduce the pump power as measured at the 1% coupler output within the polarization diversity loop by 3 dB, which corresponds to an equal power split of the pump to both the clockwise (‘x’) and anti-clockwise (‘y’) arms of the polarization diverse MSSI module. Once set, the polarization of the pump wave in the polarization diversity loop was stable enough to gather consistent performance measurements.
By splitting the pump wave equally into the two branches of the loop, the CE of the orthogonally polarized counter-clock- and clockwise travelling signals should be similar . This is because CE is given by, where is pump power, is nonlinear coefficient of HNLF, and is its effective length. Therefore, in a counter propagating fiber loop scheme where both signals experience very similar loss and dispersion, we in fact expect almost equal CE between the two counter propagating signals.
Inset (i) of Fig. 1(c) shows an optical spectrum analyzer (OSA) trace (sensitivity: −75 dBm, resolution bandwidth: 0.1 nm) containing the original signal, pump and the OPC signal. This trace is taken after the circulator and before the demultiplexer filter in an 800-km MSSI system. The CE, defined here as the ratio of conjugate power to signal input power at circulator port 3, was about −20 dB.
Even if the CEs of the counter propagating signals are slightly different in practice, the true performance of the MSSI module is best revealed through rigorous testing through measuring the quality of the received signal at the end of the link.
2.2 Results for the 121-Gb/s system
Figure 2 shows the Q in a back-to-back system with our fiber spans for both X polarization (●) and Y polarization (○) versus signal input power measured at the output of the EDFA placed after the DCF. The Q was calculated from the counted bit error ratio (BER), using [25, 26].
Figure 2 shows that the X and Y polarizations have very similar performance, demonstrating that the performance of MSSI in our system is polarization independent, able to effectively handle dual-polarization, coherent signals. It also shows that the optimum signal input power is 18 dBm, which gives the maximum back-to-back Q of 12.5 dB. For the transmission results shown in Fig. 3, the output power of the EDFA after the DCF was fixed at 18 dBm to maximize the transmission performance. The Q in a back-to-back configuration without MSSI is 15.2 dB. The 2.7-dB decrease in measured Q after the MSSI module, shown in Fig. 2, is likely to be due to nonlinear mixing products generated within the MSSI and ASE from the EDFA that is used to compensate for the low four-wave mixing conversion efficiency within the HNLF [27, 28].
Figure 3 shows the transmission performance in a 10 × 80-km link against the launch power into the S-SMF spans for two systems: one with MSSI (X Pol.: ●, Y Pol.: ○) and one without MSSI (X Pol.:▲, Y Pol.:∆). Although the MSSI system has 2.7-dB less back-to-back Q than the system without MSSI, MSSI increases the maximum launch power measured at 7%-overhead hard-decision FEC limit by about 6.4 dB, from 3 dBm to 9.4 dBm. This shows that MSSI is effective in reducing the impact of fiber nonlinearity in DP CO-OFDM systems. The peak Q is not improved with MSSI. This is due to back-to-back performance penalty  introduced by the MSSI module, which we investigate in Section 4.
3. System carrying 1.2-Tb/s
3.1 Experimental setup
The majority of the experimental setup for the 1.21-Tb/s system is the same as for the 121-Gb/s system. Only the differences are discussed below.
The output of a 193.7-THz ECL is fed into the frequency comb generator module. Two 40-GHz bandwidth phase modulators are used to generate 10 comb lines. The modulators haveV, and are driven by a 16-GHz RF signal, that is split into two paths before amplification by SHF amplifiers to give peak-to-peak voltages of 7.10 V. These RF signals are fed into the modulators, to generate peak phase shifts of about. In order to flatten the output comb lines from the second modulator, the phase difference between the two RF paths is adjusted by a tunable RF delay. The 10 selected comb lines used for the OFDM super-channel have a 6-dB variation between the edge and center lines. These lines are selected and equalized in power by a WSS, as shown in inset (i) of Fig. 4(a).The tones are then amplified and fed through the C-MZM to modulate OFDM signal onto these tones. The modulated signal is then divided into two paths; one path is frequency shifted by 8 GHz and then combined with through paths to form a 20-band 160-GHz wide super-channel as shown in Fig. 4(c).
Some modifications were made to MSSI module also for the 1.2-Tb/s system, to increase the back-to-back performance with MSSI. Figure 4(b) shows the module that combines the signal and pump. The pump (ECL3, 193.1 THz) is amplified from 16 dBm to 33 dBm using a high power EDFA (Amonics AEDFA ـ33ـBـFA) run at maximum power. The signal is amplified to 19 dBm using another variable gain EDFA. A multiplexer with a 200-GHz channel spacing simultaneously combines the signal and pump, while reducing out-of-band ASE in each of these wavebands. The HNLF in the MSSI module is shortened to 45 m, to reduce the long-scale fluctuations of the ZDW which become significant in longer HNLFs. Such variations along the fiber occur in an unpredictable manner, which degrades the OPC gain and bandwidth . Thus, a short HNLF is more suitable for broad band OPC . However, the pump power needs to be increased in order to get reasonable conversion efficiency, and this is possible due to the higher SBS threshold of shorter HNLFs .
The HNLF has a nonlinear coefficient, γ, of 11.5 W−1km−1, CD of −0.05 ps/nm/km at 1550 nm, CD slope of 0.02 ps/nm2/km, zero-dispersion-wavelength (ZDW) of 1552.82 nm and loss coefficient of 0.97 dB/km. Note that, the wavelength of the pump is adjusted to match with the ZDW of this HNLF. The pump and signal powers launched into the HNLF from each direction are 29 dBm and 15 dBm respectively.
Figure 4(c) shows the spectrum of 160-GHz OFDM super-channel comprising 20 OFDM bands each 8-GHz wide, measured with a high-resolution spectrophotometer (resolution 20 MHz) after the 50%/50% coupler and before the POLMUX emulator. Figure 4(d) is the spectrum after the circulator, measured with an optical spectrum analyzer (OSA) with a sensitivity of −80 dBm and resolution bandwidth of 0.1 nm.
The conversion efficiency was −24 dB. Although the conversion efficiency was 4-dB lower than the previous MSSI module, due to the shorter HNLF, the OPC signals from the 121-Gb/s and 1.21-Tb/s systems had similar optical signal-to-noise ratio (OSNR), as shown by the OSA in Fig. 1(c) and Fig. 4(d). This is because of higher signal power and higher pump to signal ratio (121-Gb/s system: 8 dB, 1.21-Tb/s system: 14 dB) fed into the MSSI module in the 1.21-Tb/s system. Again, the OSA is placed after the circulator and before the demultiplexer filter in the 800-km link.
3.1 Results for the 1.21-Tb/s system
Figure 5 shows the Q, derived from the average BER of the center OFDM band at different launch powers. The back-to-back penalty for MSSI is 2.5 dB, comparable to the penalty measured for the 121-Gb/s system. The performance with MSSI for two orthogonal polarization (X Pol: ●, Y Pol: ○) and without MSSI (X Pol:▲, Y Pol:∆) are very similar at low powers, where ASE from the link dominates. However, at higher powers, the MSSI mitigates fiber nonlinearity; the maximum launch power (measured at the FEC limit) is increased by 2.8 dB, to 10.0 dBm. This benefit is reduced compared with the 121-Gb/s system, partially due to the lower back-to-back performance for the 1.21-Tb/s system with MSSI (Q = 10.3 dB for the 1.21-Tb/s signal versus 12.6 dB for the 121-Gb/s system). This performance could be improved by using lower-loss filters (to reduce the added ASE in the MSSI stage) as discussed in section 4, and by using two-stage OPC (to reduce spurious nonlinear mixing products) as discussed in . Additionally, the benefit of MSSI decreases with higher bandwidth signals .This is because MSSI decreases the benefit of the phased-array effect , which reduces inter-channel nonlinear cross-talk in high bandwidth systems. This phased array effect is decreased as MSSI effectively reverses the effect of dispersion in the second half of the link.
Figure 6(a) shows the BERs for all of the OFDM bands at a 5-dBm launch power, which is the optimal power for the system without MSSI.
The channels in the middle of the band have similar BERs; however, the roll-off of the 200-GHz multiplexers attenuates the edges of the MSSI signal, increasing the error rates. This could be avoided by using an optical filter with a wider passband.
Figure 6(b) shows the BER of the OFDM bands at a power of 8 dBm, where nonlinearity dominates the system’s performance. All but three of the edge subcarriers of the MSSI system have BERs better than the 7%-overhead hard-decision FEC limit of 3.8 × 10−3. When the BER is averaged over all of the subcarriers, the MSSI beneficially decreases the BER from 1.0 × 10−2 to 3.0 × 10−3. Pairwise coding or similar techniques could be used to balance the error rates across the channels .
4. Simulation results and discussion
In the previous section, we have shown experimentally that although MSSI did not improve the maximum Q, it improved the maximum launch power of 1.21-Tb/s dual polarization 80 × 10-km transmission system by 2.8 dB. The system would be able to benefit from this increase in permissible launch power to overall transmission distance by having longer spans. In order to demonstrate this benefit with MSSI, we simulated a 20-band 160-GHz wide OFDM super-channel using VPItransmissionMaker, for different span lengths with 10 spans.
For sake of simplicity, the simulated system is a single polarization system. However, in a system that is limited by fiber nonlinear impairment rather than cross polarization nonlinear effects, this is able to demonstrate the practical benefit of using MSSI with a 160-GHz wide dual polarization 16-QAM OFDM super channel. In our simulations, we have used the combined insertion loss inside MSSI module as a simulation parameter, which comprises losses of a Siemens TransXpress Mux, Circulator, 99%/1% coupler, 90%/10% coupler and Siemens TransXpress Demux. Peak Q in the simulation and experiment was made similar by tuning the total combined insertion loss (IL) between 9 and 12 dB. Figure 7(a) shows the simulation results for reference system (▲) and system with MSSI (■) with 12-dB IL after 10 × 80-km transmission. Compared with the results shown in Fig. 5, a similar relation in peak Q between experiment and simulation was obtained. All other parameters in simulation were set as described in the experimental setup in Section 3.1.
Figure 7(a) also shows the simulation results of Q versus launch power with increased span lengths (10 × 90-km: ▲; and 10 × 110-km: ▲) in a 160-GHz OFDM super-channel for the reference system without MSSI. The purple curve with squares (■) shows the 10 × 110-km results with MSSI. The gray dashed line (–) shows the FEC limit for BER ˂ 3.8 × 10−3. The results show that maximum span length of the reference system is about 90 km, beyond which performance degrades below the FEC limit. On the other hand, the MSSI system sustains its performance over the FEC limit up to span lengths of 110 km. At launch powers of 8 dBm, the MSSI system clears the FEC limit, while the 10 × 110-km system without MSSI is well below this limit for all launch powers. Therefore, the overall transmission distance could be increased from 10 × 90 km to 10 × 110 km; a reach improvement of nearly 22% without increasing the number of inline amplifiers. Figure 7(b) shows the maximum Q value for different span lengths from 80 km to 110 km for both systems (Reference system: ▬; MSSI system: ▬) and the corresponding improvement with MSSI (▲). This shows that effect of nonlinearity mitigation due to MSSI increases with longer spans. This is due to the increased maximum launch power with MSSI. With span lengths of 90 km, MSSI shows about 1.2 dB improvement compared with the reference system.
We next investigate the performance improvement due to MSSI for longer transmission distances when the span length is kept fixed at 80 km. Figure 8(a) shows Q versus launch power for three different transmission distances (10 × 80km, 12 × 80km, 14 × 80km), for both the reference system and MSSI. While the optimum Q of the reference system falls rather sharply, the system with MSSI shows less degradation in performance with increasing transmission distance. Nonlinear power asymmetry with respect to accumulated dispersion between the first half and second half of the link is largely caused by the nearest span either side of the MSSI module. This asymmetry becomes less and less significant with increasing number of spans, which makes MSSI more effective for longer transmission distances. Figure 8(b) shows the Q at optimum input power versus transmission distance for both systems. The reference system peak Q falls below the FEC limit with links longer than 15 × 80 km. On the other hand, the MSSI system performs above the FEC limit up to 26 × 80 km. This shows that MSSI has the potential to increase the transmission distance by more than 70% when 80-km spans are used. The red curve in Fig. 8(b) shows about 1-dB peak performance improvement at 1240 km when MSSI is used.
We next investigate improving the maximum performance of the MSSI module itself by changing the parameters of components within the MSSI module. Three parameters were considered here: the total combined insertion loss (IL) inside the MSSI module, the OPC conversion efficiency (CE) and HNLF length (L). Figure 9(a) shows the Q versus launch power for the reference system and MSSI system with different total insertion losses. The transmission distance is 10 × 80 km. Figure 9(b) shows the improvement in maximum Q performance, comparing systems with and without MSSI, versus total insertion loss. These results show that reducing total insertion loss inside MSSI is critical in increasing maximum Q. With the ideal case of no insertion loss, the MSSI system could improve the maximum Q by about 1 dB.
Figure 10 shows Q improvement for two different lengths of HNLF, 45 m (∆) and 22.5 m, (■) versus CE. The insertion loss in both of these cases is 12 dB. The HNLF nonlinear coefficient was kept as in Section 3.1. The result shows that maximum Q performance could be improved by about 0.7 dB by improving the CE from −24 dB to −18 dB using a shorter HNLF. Numerical results shows better performance with shorter HNLF at the same conversion efficiency, which agrees with the analytical prediction of . Shorter HNLF is also preferable to minimize the effect of the variation in zero dispersion wavelength along the HNLF. This helps maintain a uniform gain over a wide OPC bandwidth and a higher threshold for SBS.
We have experimentally demonstrated the first use of MSSI for fiber nonlinearity compensation for DP CO-OFDM systems. Polarization independence of the MSSI module was confirmed by observing very similar performance between counter-propagating X and Y polarization signals. MSSI increases the maximum permissible launch power by 2.8 dB for a 1.21-Tb/s 16-QAM DP CO-OFDM system and 6.4 dB at 121 Gb/s. Our simulations show that this increase in maximum viable launch power could support about a 22% increase in total transmission distance without increasing the number of inline amplifiers, by extending each fiber span from 90 km to 110 km, for 1.21-Tb/s system. For a fixed length of span, the MSSI system shows less degradation in optimum performance with increasing number of spans (i.e. increased transmission distances), and supports more than 70% increase in overall transmission distance when 80-km spans are used. As for optimization of the MSSI module itself, simulations show that reducing total insertion loss inside MSSI module is very important in achieving higher maximum Q performance as is using shorter HNLF and increasing the OPC conversion efficiency.
This research was conducted by the Australian Research Council Centre of Excellence for Ultrahigh bandwidth Devices for Optical Systems, CUDOS (Project number CE110001018). We should like to thank VPIphotonics.com for the use of VPItransmissionMakerTM.
References and links
2. S. L. Jansen, I. Morita, T. C. Schenk, and H. Tanaka, “Long-haul transmission of16×52.5 Gbits/s polarization-division-multiplexed OFDM enabled by MIMO processing (Invited),” J. Opt. Netw. 7(2), 173–182 (2008). [CrossRef]
3. R.-J. Essiambre, “Exploring capacity limits of fibre-optic communication systems,” in 34th European Conference on Optical Communication, ECOC(2008), p. We.1.E.1. [CrossRef]
4. A. D. Ellis, Z. Jian, and D. Cotter, “Approaching the non-linear Shannon limit,” J. Lightwave Technol. 28(4), 423–433 (2010). [CrossRef]
6. S. L. Jansen, D. van den Borne, P. M. Krummrich, S. Spalter, 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]
7. P. Minzioni, F. Alberti, and A. Schiffini, “Optimized link design for nonlinearity cancellation by optical phase conjugation,” IEEE Photon. Technol. Lett. 16(3), 813–815 (2004). [CrossRef]
8. S. L. Jansen, D. van den Borne, B. Spinnler, S. Calabrò, H. Suche, P. M. Krummrich, W. Sohler, G. D. Khoe, and H. de Waardt, “Optical phase conjugation for ultra long-haul phase-shift-keyed transmission,” J. Lightwave Technol. 24(1), 54–64 (2006). [CrossRef]
9. S. L. Jansen, D. van den Borne, C. Climent, M. Serbay, C. J. Weiske, H. Suche, P. M. Krummrich, S. Spalter, S. Calabro, N. Hecker-Denschlag, P. Leisching, W. Rosenkranz, W. Sohler, G. D. Khoe, T. Koonen, and H. de Waardt, “10,200 km 22×2×10 Gbit/s RZ-DQPSK dense WDM transmission without inline dispersion compensation through optical phase conjugation,” in Optical Fiber Communication Conference (OFC/NFOEC)(2005), p. PDP 28.
10. V. Pechenkin and I. J. Fair, “Analysis of four-wave mixing suppression in fiber-optic OFDM transmission systems with an optical phase conjugation module,” Optical Communications and Networking, IEEE OSA Journal of 2(9), 701–710 (2010). [CrossRef]
11. X. Liu, F. Buchali, and R. W. Tkach, “Improving the nonlinear tolerance of polarization-division-multiplexed CO-OFDM in long-haul fiber transmission,” J. Lightwave Technol. 27(16), 3632–3640 (2009). [CrossRef]
12. L. B. Du, M. M. Morshed, and A. J. Lowery, “604-Gb/s coherent optical OFDM over 800 km of S-SMF with mid-span spectral inversion,” in Opto-Electronics and Communications Conference (Busan, 2012), pp. 3B2–3. [CrossRef]
14. M. Morshed, L. B. Du, B. Foo, M. D. Pelusi, and A. J. Lowery, “Optical phase conjugation for nonlinearity compensation of 1.21-Tb/s pol-mux coherent optical OFDM,” in 18th Opto-Electronics and Communications Conference(Kyoto, Japan, 2013), pp. PD 3–4.
15. K. Solis-Trapala, T. Inoue, and S. Namiki, “Nearly-Ideal Optical Phase Conjugation based Nonlinear Compensation System,” in Optical Fiber Communication Conference(Optical Society of America, San Francisco, California, 2014), p. W3F.8. [CrossRef]
16. I. Phillips, M. Tan, M. F. Stephens, M. McCarthy, E. Giacoumidis, S. Sygletos, P. Rosa, S. Fabbri, S. T. Le, T. Kanesan, S. K. Turitsyn, N. J. Doran, P. Harper, and A. D. Ellis, “Exceeding the Nonlinear-Shannon Limit using Raman Laser Based Amplification and Optical Phase Conjugation,” in Optical Fiber Communication Conference(Optical Society of America, San Francisco, California, 2014), p. M3C.1. [CrossRef]
17. H. Hu, R. M. Jopson, A. Gnauck, M. Dinu, S. Chandrasekhar, X. Liu, C. Xie, M. Montoliu, S. Randel, and C. McKinstrie, “Fiber Nonlinearity Compensation of an 8-channel WDM PDM-QPSK Signal using Multiple Phase Conjugations,” in Optical Fiber Communication Conference(Optical Society of America, San Francisco, California, 2014), p. M3C.2. [CrossRef]
18. I. Fatadin, D. Ives, and S. J. Savory, “Blind equalization and carrier phase recovery in a 16-QAM optical coherent system,” J. Lightwave Technol. 27(15), 3042–3049 (2009). [CrossRef]
19. Y. Xingwen, W. Shieh, and T. Yan, “Phase estimation for coherent optical OFDM,” IEEE Photon. Technol. Lett. 19(12), 919–921 (2007). [CrossRef]
20. P. Minzioni, I. Cristiani, V. Degiorgio, L. Marazzi, M. Martinelli, C. Langrock, and M. M. Fejer, “Experimental demonstration of nonlinearity and dispersion compensation in an embedded link by optical phase conjugation,” IEEE Photon. Technol. Lett. 18(9), 995–997 (2006). [CrossRef]
21. M. Karlsson, “Four-wave mixing in fibers with randomly varying zero-dispersion wavelength,” J. Opt. Soc. Am. B 15(8), 2269–2275 (1998). [CrossRef]
22. T. Hasegawa, K. Inoue, and K. Oda, “Polarization independent frequency conversion by fiber four-wave mixing with a polarization diversity technique,” IEEE Photon. Technol. Lett. 5(8), 947–949 (1993). [CrossRef]
24. M. E. Marhic, G. Kalogerakis, and L. G. Kazovsky, “Gain reciprocity in fibre optical parametric amplifiers,” Electron. Lett. 42(9), 519–520 (2006). [CrossRef]
25. P. Wei-Ren, T. Tsuritani, and I. Morita, “Transmission of high-baud PDM-64QAM signals,” J. Lightwave Technol. 31(13), 2146–2162 (2013). [CrossRef]
26. G. P. Agrawal, Fiber-optic Communication Systems (Wiley and Sons, 2010).
27. M. Morshed, L. B. Du, and A. J. Lowery, “Performance limitation of coherent optical OFDM systems with non-ideal optical phase conjugation,” in IEEE Photonics Conference (IPC)(Burlingame, CA, 2012), pp. 394–395, TuU. [CrossRef]
28. M. Morshed, L. B. Du, and A. J. Lowery, “Mid-span spectral inversion for coherent optical OFDM systems: Fundamental limits to performance,” J. Lightwave Technol. 31(1), 58–66 (2013). [CrossRef]
29. S. Le Floch and P. Cambon, “Theoretical evaluation of the Brillouin threshold and the steady-state Brillouin equations in standard single-mode optical fibers,” J. Opt. Soc. Am. A 20(6), 1132–1137 (2003). [CrossRef] [PubMed]
31. V. Pechenkin and I. J. Fair, “On four-wave mixing suppression in dispersion-managed fiber-optic OFDM systems with an optical phase conjugation module,” J. Lightwave Technol. 29(11), 1678–1690 (2011). [CrossRef]
32. M. Nazarathy, J. Khurgin, R. Weidenfeld, Y. Meiman, P. Cho, R. Noe, I. Shpantzer, and V. Karagodsky, “Phased-array cancellation of nonlinear FWM in coherent OFDM dispersive multi-span links,” Opt. Express 16(20), 15777–15810 (2008). [CrossRef] [PubMed]
33. Y. Hong, A. J. Lowery, and E. Viterbo, “Sensitivity improvement and carrier power reduction in direct-detection optical OFDM systems by subcarrier pairing,” Opt. Express 20(2), 1635–1648 (2012). [CrossRef] [PubMed]