A multiple-channel multiple-function optical signal processor (MCMF-OSP) including wavelength-waveform conversions, pulsewidth tunability, and signal regeneration is realized through AND logic gate based on optical parametric processing with a pulsewidth-tunable RZ clock pump. The proposed scheme simultaneously offers four signal processing functions which are useful in wavelength-division multiplexing (WDM) transmission systems, and at network nodes with the necessity for multiple-channel data processing. After the discussions on the concept of MCMF-OSP, a proof-of concept experiment is demonstrated on four 10 Gb/s nonreturn-to-zero (NRZ) data format channels using nonlinearities in semiconductor optical amplifier (SOA) and highly nonlinear fiber (HNLF). A wavelength and waveform conversions to return-to-zero (RZ) modulation format are obtained together with pulsewidth-tunable range from 20% to 80% duty cycles for all input signals. The converted signals inherit the timing and waveform of the RZ clock pump, thus resulting in a time regeneration and large tolerance to narrow-band optical filtering (NAOF) and fiber accumulated chromatic dispersion (CD).
© 2009 Optical Society of America
Properly utilizing nonlinear effects in highly nonlinear medium such as semiconductor optical amplifier (SOA) and optical fiber has been recognized as the common techniques in optical signal processing due to the ultrafast response and the commercial availability of the devices [1, 2]. The focus of interest in the field has been well on how to boost the line rate as well as improve the signal quality of a single channel on a separate wavelength [3–5]. Applications on wavelength-division multiplexing (WDM) communication systems and networks, however, need the processing with multiple functions for multiple data channels. Straightforward solutions [6, 7] by processing the channels individually and separately adding further functions afterwards are not practical due to the accumulation of network complexity and cost. As a result, including many functions in a processing scheme to realize a multiple-channel multiple-function optical signal processor (MCMF-OSP) is apparently a desirable approach for future photonic networks.
Multiple-channel but single-function optical processors such as wavelength/waveform converters or regenerators have been extensively studied for applications on optical switching, transmultiplexing between WDM and optical time-division multiplexing (OTDM), and signal regeneration. Grouped-wavelength or multiple-wavelength conversions have been demonstrated mostly using optical parametric processing (OPP) either in highly nonlinear fiber (HNLF) [8, 9] or periodically-poled lithium-niobate (PPLN) [10, 11]. Whereas, multiple-channel waveform conversions between nonreturn-to-zero (NRZ) and return-to-zero (RZ) formats have been developed along with the progress of single-wavelength converters based on cross-phase modulation (XPM) in SOAs or fiber in integration to interferometric arrangements [12–14]. For researches on regenerations inWDM systems [15–16], the widely-used technique has been developed originally from spectrum broadening by self-phase modulation (SPM) and optical filtering proposed for the first time in Ref. . Although capable of processing for multiple channels, above schemes are limited to single-function demonstrations.
Reference  has reported WDM-channel signal regeneration while performing an NRZ-to-RZ format conversion using a phase modulator (PM) and a fiber delay-interferometer (DI). Modulating the phase of the input signals in PMdriven by a local RF clock allowed a portion of mark level passed through the DI, but a small misalignment of one input NRZ signal with the RF clock would cause degradation on the converted RZ signal. This restriction is the common feature in most of multiple-channel optical processing with necessity of a control signal. It is particularly difficult to overcome due to the need of simultaneous alignment of all input channels with the control signal. In addition, including pulsewidth tunability in a multiple-channel NRZ-to-RZ format converter has also been proposed by using XPM in a SOA followed by a DI . The pulsewidth tunability is considered to be one of the important functions for performance improvement in RZ transmission systems over the pulsewidth management [20–23]. The pulsewidth tunability in Ref. , however, is restricted due to the characteristic of detuned filtering which determines the output waveform. On top of these, the coupling of wavelength conversion and waveform conversion with signal regeneration or pulsewidth tunability to form multiple-channel optical signal processing with more than two functions remains a challenge in proposed schemes so far.
In this paper, we propose and experimentally demonstrate a multiple-channel multiple-function optical signal processor (MCMF-OSP) including wavelength and NRZ-to-RZ waveform conversions, pulsewidth tunability, and signal regeneration. Four 10 Gb/s wavelength-division multiplexed NRZ signals are converted in both wavelength and waveform by optical parametric processing using a pulsewidth-tuned RZ clock pump. The converted signals follow the waveform and timing of the RZ clock pump, resulting in pulsewidth tunability and signal regeneration for output RZ signals. Tuning the pulsewidth of the RZ clock pump allows wide pulsewidth tunability from 20% to 80% duty cycles. The converted RZ signal at different pulsewidths shows different characteristics to timing jitter. A large tolerance up to ± 30 ps timing jitter (± 30% of 10 GHz bit interval) of the input NRZs is obtained as pulsewidth is tuned at 20 ps and 40 ps. Last but not least, the proposed MCMF-OSP also offers a further signal generations as inputWDM-NRZ signals are distorted by narrow-band optical filtering (NAOF) or fiber accumulated chromatic dispersion (CD).
Following in the content, we first introduce the concept of theMCMF-OSP in Section 2. The setup of proof-of-concept experiment is presented in Section 3. Experimental results in terms of wavelength-waveform conversions with pulsewidth tunability, and signal regenerations are then shown successively in Section 4.
2. Concept ofmultiple-channelmultiple-function optical signal processor (MCMF-OSP)
Figure 1 shows the basic concept of the proposed MCMF-OSP: wavelength-waveform conversions (function 1, 2), pulsewidth tunability (function 3), and signal regeneration (function 4). The MCMF-OSP consists of an AND logic gate based on optical parametric processing (OPP), a pulsewidth tunable RZ clock generator, and a clock recovery. In the AND logic gate, WDM-NRZ signals, which are aligned in time with each other, are OPP interacted with a RZ clock pump created from the pulsewidth tunable RZ clock generator. This clock generator is synchronized by an optical clock control whose timing is extracted from the incomingWDM-NRZ signals by the clock recovery scheme. Over the OPP with the RZ clock pump, which can be realized by using either four-wavemixing (FWM) in HNLF , or cascaded second-harmonic generation and difference frequency generation (SHG+DFG) in PPLN , the multiple-channel wavelength and NRZ-to-RZ waveform conversions (functions 1, 2) are performed simultaneously. The OPP gives the proposed scheme two important advantages. First, it is a data-format and bit-rate transparent process. Second, as shown in Fig. 2, over the AND logic function between the input WDM-NRZ signals and the RZ clock pump, the converted signals will inherit the timing and pulse shape of the pump. This in turn leads to the perfect synchronization and the large tolerance to timing-jitter for all channels after the conversion as shown in Fig. 2(a). The perfect synchronization is particularly useful in applications where bit-level synchronization among WDM channels is required, for example, in bit-parallel WDM data transmission systems or in WDM optical packet switching [24, 25]. The timing-jitter tolerance also facilitates the strict time alignment among input WDM-NRZ signals and with the control clock, which has been a disadvantage in Ref. . Furthermore, it can be seen from Fig. 2(b) that aligning the RZ clock pump to the centre region of NRZ signals makes it possible to eliminate and overcome the signal distortions resulting from narrow-band optical filtering and/or fiber accumulated chromatic dispersion. All of these introduce to the proposed system the signal regenerative function (function 4). As mentioned above, since the converted signals follow the shape of the RZ clock pump, the pulsewidth tunability (function 3) is realized by tuning the pulsewidth of the RZ clock pump. The width-tunable RZ clock can be created by delayed interferometric switches, such as Sagnac- or Mach-Zehnder-type interferometer, which uses cross-phase modulation (XPM) in SOAs [26, 27]. Adjusting the switching window results in pulsewidth tunability for the output signals.
Various schemes for the clock recovery can be applied for the proposed processor such as all-optical schemes [28–30]. A widely-used electrical configuration consisting of a photodetector and a phase-lock loop (PLL) is also a suitable solution. In this paper, fourWDM-NRZ channels are input to form the four-channel four-function optical processor. It should be noted that the processing of a larger number of input channels is also possible dependently on the conversion efficiency of the OPP. Interband multiple-wavelength conversions with over 30 channels have been successfully demonstrated in Ref. [8–11]. Polarization diversity can be introduced to the OPP-based AND logic gate to realize the polarization-independent optical signal processing by using counter-directional or parallel scheme [8, 31].
3. Proof-of-concept experimental setup
As discussed in Section 2, different schemes can be employed to realize the concept of the proposed MCMF-OSP that is dependent on how to implement the modules in Fig. 1. Figure 3 shows the setup of our proof-of-concept experiment which utilizes optical nonlinear effects in semiconductor optical amplifier (SOA) and highly nonlinear fiber (HNLF).
The pulsewidth-tunable RZ clock generator is a delayed Sagnac interferometer using XPM in SOA . The interferometer consists of a polarizer at the output of a nonlinear fiber loop including a polarization beam splitter (PBS), a variable delay line (DL), and a SOA placed at an arbitrary position in the loop. A continuouswave (CW) probe at 1552.1 nmfromexternal-cavity laser-diode (ECL) through a circulator and the PBS are divided into two orthogonally-polarized waves propagating in two arms of the loop. The two waves experience the XPMin SOA, which is excited by a 10 Gb/s local optical RZ clock, at different times which are determined by the delay line (DL) in the loop. A polarization controller (PC) in the loop is to direct the waves to the output. The interference of the differential phase modulated waves at the polarizer after the loop results in the formation of RZ pulse train with the pulsewidth Δt tuned by the DL, and synchronized with the local clock. Operating principle of the pulsewidth-tunable RZ clock generator is similar to SOA-based wavelength converters , but instead of the incoming data signal, an optical clock is used as a control signal. The SOA with carrier recovery time of 100 ps is biased by a 200 mA current. The saturated power of SOA is about 10 dBm. The local RZ clock is generated by modulating a CW at 1542.0 nm from ECL in an electroabsorption modulator (EAM) driven by a 10 GHz synthesizer. Input powers of the CW probe and the local RZ clock are set at 0 dBm and 2 dBm, respectively.
The OPP-based AND logic gate uses four-wave mixing (FWM) in a highly nonlinear fiber (HNLF). The HNLF is 500 m long, with nonlinear coefficient (γ) of 12.6 W-1·km-1, zero dispersion wavelength at 1552 nm, and dispersion slope of 0.032 ps/nm2/km. Input WDM-NRZ channels are FWM interacted and overlapped in time with the pulsewidth-tuned RZ clock as a pump signal. Over the AND logic function of FWM effect, beside converted to different wavelengths, the FWM products inherit the pulsewidth of the pulsewidth-tuned clock and the data of the input NRZ signals. Therefore, changing the delay time Δt in the width-tunable RZ clock generator simultaneously results in the pulsewidth tunability of the converted signals at the output. To generateWDM-NRZ signals, four 100 GHz channel spacingWDM signals with wavelength from 1554.1 nm (ch1) to 1556.5 nm (ch4) are modulated by a LiNbO3 modulator (LNM) with a 10 Gb/s data (PRBS: 231-1). For a maximum FWM process, the WDM-NRZ signals and the RZ clock pump are controlled to have the same state of polarization by PCs. The power levels prior to the AND logic gate are set to 3 dBm for the total WDM signal and 14 dBm for the width-tuned clock pump by erbium-doped fiber amplifiers (EDFAs). At the receiver, the converted channels are filtered by a 100 GHz channel spacing arrayed waveguide grating (AWG), and detected by 10 Gb/s non-preamplifier receiving scheme.
To investigate the timing regeneration, the signal performance is monitored while the width-tunable RZ clock pump is shifted away from the center of the input NRZ signals. To do that, a phase shifter is used on the direct synchronization link between the NRZ signals and the local RZ clock. The demonstration of clock recovery is beyond the scope of this paper.
4. Experimental results and discussions
4.1. Wavelength-waveform conversions and pulsewidth tunability (function 1, 2 and 3)
Due to the close relation among the functions, the wavelength-waveform conversions and the pulsewidth tunability are discussed together in this subsection. Four-wave mixing spectra at the output of the AND gate is shown in Fig. 4. The input fourWDM channels are converted to wavelengths from 1550.1 nm (ch1),…, 1547.7 nm (ch4). Almost the same conversion efficiency is obtained for all channels. A larger number of channels with wider wavelength conversion range could be also possible provided a HNLF with higher nonlinear coefficient or polarization maintaining is used as demonstrated in Ref. . As seen in Fig. 4, channel order is inverted after the FWM. This is common feature in parametric process using a single pump as seen in Ref. [8–10]. To rearrange the channel order, an additional wavelength conversion stage could be used after the present system. This method is, dependently on the setting of the pump wavelength, able to bring the channels back to the original wavelengths, or to convert the channels to other wavelength regions for larger wavelength conversion range. Other method is using two-pump operating scheme as demonstrated in Ref. . By properly arranging the pump wavelengths, it is possible for multiple-channel wavelength conversion without the inversion of channel order.
Figure 5 shows eye diagrams of the converted signals at all four channels as the delay line is widely tuned. The NRZ signals are successfully converted to RZ modulation format, and the different pulsewidths corresponding to the delay settings Δt of 20 ps, 40 ps, 60 ps, and 80 ps are observed simultaneously at all channels. The clear eye openings without patterning effect indicate the good performance of the proposed conversion with wide tunability range of the pulsewidth. The data pattern independence makes the present system transparent to the input bit-rate which is potential for high speed applications. The rising and falling time of the pulsewidth-tuned RZ clock as well as the converted RZ signals is strongly dependent on the pulsewidth of the input control RZ clock. Thus, a wider tuning range is also achieved once a narrower-width input clock is used instead of the one of 20 ps-width in this experiment.
The conversion performance is further investigated by measuring the bit error rate (BER) of the output signals. The BER characteristics of the converted signal at various pulsewidths for channel 3 are shown in Fig. 6 (a), whereas, the receiver sensitivity at BER=10-9 for all channels with respect to the back-to-back NRZ inputs are shown in Fig. 6 (b). Compared to the back-to-back, the converted RZ signals at the pulsewidth of 20 ps, 40 ps, 60 ps and 80 ps have negative power penalties of -2.0 dB, -1.9 dB, -1.7 dB, and -1.4 dB, respectively. It is interesting to note that beside the BER improvement as compared to the NRZ, the converted RZ signal whose pulsewidth is narrower has higher receiver sensitivity. The similar results about the dependence of receiver sensitivity on signal pulsewidth have been found in numerical analysis  and experimental research [32, 33]. This comes from a reason that at the same average power the narrow pulse signal has higher extinction ratio than that of the wide one. Since the photocurrent at optical receiver is proportional to the incident optical power, the above extinction ratio is further improved as the signal is converted to electrical domain. Further reason is the bandwidth limitation of the electrical components in the receivers causing the rise in intersymbol interference (ISI) and the drop in receiver sensitivity when detecting the signal with wider pulsewidth. The narrower pulsewidth signal thus owns the better sensitivity. As seen in Fig. 6 (b), due to the difference in the OSNRs of input NRZ signals, the receiver sensitivities of the converted signals are slightly varied from channel to channel. However, the power penalties against the back-to-back remain nearly unchanged among converted channels.
Although the RZ signal with narrower pulsewidth have shown a higher sensitivity, it is not always true that the narrow pulsewidth signal is better for fiber-optics communication systems. Our previous investigation on transmission characteristic of RZ signal with widely-tuned pulsewidth [22, 23] has shown that for each transmission distance with particular cumulative dispersion there is existence of an optimal pulsewidth for the converted RZ signal to obtain highest receiver sensitivity, and the optimal pulsewidth is shifted to the wider pulses when the dispersion becomes larger. This has raised the importance of the pulsewidth tunability for performance optimization in RZ transmission systems.
4.2. Signal regeneration
This subsection describes two signal regenerative features of the proposed MCMF-OSP: 1) retiming for the NRZ inputs with large timing jitter, and 2) performance improvement for signals distorted by NAOF and CD. Experimental setups to create timing jitter by using WDM interleaver, and signal distortion induced by NAOF and CD for WDM-NRZ signals before the MCMF-OSP are shown in Fig. 7.
4.2.1. Tolerance to timing jitter
It is well-known that the NRZ modulation format is advantageous to transmission systems due to its resistance to timing jitter. The timing jitter is, however, still a critical problem in WDM-channel optical signal processing. Since the misalignment easily occurs among input channels due to the wavelength-dependent propagation delay, it is hardly to achieve a reference point for the control signal without resynchronization of all channels before performing the signal processing. In the best case, where there is no delay among channels, the timing jitter of one channel becomes double with respect to the other channel of the same jitter.
The investigation on the timing jitter tolerance is done by adjusting the time offset of local RZ clock relative to the center of incoming WDM-NRZ signals. Two cases of setting for time offset are considered: input NRZ signals are time aligned in case 1, and misaligned in case 2. The alignment or misalignment among input NRZs is made by using aWDM signal interleaver comprised of an AWG, optical delay lines (DLs), and a WDM coupler as shown in Fig. 7(a).
In case 1, the comparisons in terms of BER variations at channel 2 are shown in Fig. 8(a) for 20 ps, 40 ps converted signals, and in Fig. 8(b) for 60 ps and 80 ps converted signals as the time offset is changed in a wide range. Figure 9, as an example, shows the changes in eye patterns of 20 ps and 40 ps converted signals accordingly with various values of time offset. Since the signals at different pulsewidths have different receiver sensitivities as seen in Fig. 6, to compare the BER characteristics among such signals the received power in each case of pulsewidth is adjusted so that its BER reaches the same level of 10-10 at zero time offset.
As seen in Fig. 8, the large timing jitters of ±30 ps and ±20 ps (±30% and ±20% of 10 GHz bit period) of input NRZ signal cause the error rates increased by a factor of only 10 for pulsewidths of 20 ps, 40ps, and 60 ps, 80 ps, respectively. In wide ranges of time offset, which are around 50 ps in case of Fig. 8(a) and 30 ps in case of Fig. 8(b), the bit error rates remain constant at 10-10 level for different-pulsewidth signals. Similarly in term of eye patterns, as seen in Fig. 9, the converted signals with pulsewidth of 20 ps and 40 ps at time offset of ±20 ps are observed almost the same waveform as at zero time offset in Fig. 5. This is an attractive property for signal processing functions because it allows a penalty-free operation even with input signals of large timing jitters which has been hardly achieved in other schemes . This property is due to the AND logic function based on optical parametric processing, and the characteristic of NRZ modulation format. As long as the RZ clock pulse falls into the flat top region of the mark level of the NRZ signal, the conversion performance remains unchanged. That is also the reason why the narrower pulsewidth signal owns the penalty-free operation at 10-10 BER level in a wider range of time offset than that of wider pulsewidth one.
However, once the RZ clock steps out of the flat top region of the NRZ mark level, the conversion performance is degraded in a quicker manner with the narrower pulse clock when the time offset is set to larger amount as seen in Fig. 8. This is because at certain time offset there is always a larger portion of wider pulsewidth RZ clock remains in mark level region than that of narrower pulsewidth one. This can be seen in Fig. 9. The 40 ps converted signal has a clearer eye patterns than those of 20 ps one at ±30 ps time offset. This point again emphasizes the need of the pulsewidth management of RZ modulation format not only for the improvement of transmission performance [22, 23], but also for the tolerance to timing jitter of NRZ signals incoming to the processor.
In case 2 of the misalignment among input NRZs, eye patterns of input WDM-NRZ signals with various timing jitters respect to the time reference and resulting eye patterns of output signals converted to 20 ps RZs are shown in Fig. 10. The time delays within input channels are set by adjusting DLs in the WDM interleaver, and a time delay of around 50 ps is created between channel 1 and 4 as seen in the leftside eye patterns of Fig. 10. The RZ clock pump is adjusted to the overlapped point of all input channels, which is the time reference for the retiming. As shown in the resulting eye patterns at the right side of Fig. 10, the signals with various time jitters are shifted back to the time reference. Thus, a perfect synchronization has been done for all input data channels. Note here that the acceptable 50 ps delay between channel 1 (1554.1 nm) and channel 4 (1556.5 nm) allows a WDM-channel transmission over around 1.2 km standard single-mode fiber (SSMF), 17 ps/nm/km dispersion, prior to theMCMF processor without resynchronization. Although the bit alignment may cause nonlinear crosstalks in case the convertedWDM signals are propagated on transmission links right after theMCMF processor, the feature of time regeneration ofMCMF-OSP is particularly useful in applications where bit-level synchronization amongWDM channels is required, for example, in bit-parallelWDM data transmission systems or in WDM optical packet switching [24, 25]. It is also necessary to note that the bit alignment will be no longer a problem in transmissions with chromatic dispersion due to the wavelength-dependent propagation delay, especially with RZ format signals.
4.2.2. Tolerance to NAOF and CD
Narrow-band optical filtering (NAOF) and fiber chromatic dispersion (CD) are considered two of the most signal degradation factors in fiber-optics communications. The former factor is applied to WDM networks with add-drop or routing capabilities in which signal channels are passed through cascaded optical filters used for multiplexing and demultiplexing [34–36]. On the other hand, the latter factor is a critical issue in every fiber-optic transmission systems. In the experiment, two parallel 25 GHz channel-spacing AWGs (0.2 nm) are used as NAOF in Fig. 7(b). Various distances of SSMF, on the other hand, is considered CD medium as shown in Fig. 7(c). A WDM interleaver is needed after SSMF for the signal alignment at the input of the MCMF-OSP.
Experimental results regarding to tolerance to NAOF factor are shown in Fig. 11 when signals are converted to 20ps RZs. As seen in Fig. 11(a) for channel 2, the NAOF makes slow the leading and trailing edges of the input NRZ (the inset) and the receiver sensitivity inclined by 3 dB with a error floor at 10-9 BER level. However, after the signal processing, since the output converted signals follow the pulse shape of the RZ clock pump, the signal degraded by NAOF is recovered in terms of BER in Fig. 11(a), and the leading and trailing edges as seen in Fig. 11(b)–(e) for all four channels. For instance, at channel 2, the 20 ps converted RZ signal owns a better sensitivity by 3.9 dB at BER of 10-9 and a linear BER characteristic than those of the degraded and BER-floored NRZ signal. Note here that with a non-degraded NRZ input the sensitivity improvement is only 2 dB as it is converted to 20 ps RZ signal as shown in Fig. 6. These importantly lead to the signal regeneration for the proposed MCMF-OSP.
In the case of SSMF transmission before processing, due to the NRZ signals at different wavelengths to travel at different velocities over SSMF, it is necessary to align the signals prior to the optical processor by the WDM interleaver as shown in Fig. 7(c). In practical systems and networks, the time alignment for signals must be controlled by relying on synchronization information of the WDM channels. Thus, it is essential to collect the synchronization from all input channels. This is easy to achieve once the proposed MCMF-OSP is applied at the network nodes which are the central points of synchronization information due to the necessity for network control and management. It can also be realized by using multiwavelength clock recovery schemes as in Ref. [28, 29].
Similarly to the effect of NAOF, the CD accumulated along SSMF causes the slow leading and trailing edges of NRZ signals, and signal degradation. Figure 12 compares the receiver sensitivity at 10-9 BER of NRZ signal (channel 3) with its signals converted to different-pulsewidth RZs as the WDM-NRZ signal is transmitted up to 50 km SSMF corresponding to 850 ps/nm CD accumulation. There are significant improvements in the sensitivity of converted RZ signals respect to the input degraded NRZ. This sensitivity improvement keeps increasing in the cases of 20 ps, 40 ps and 60 ps converted signals as the amount of CD becomes larger. For example, when the CD is from zero to 850 ps/nm, by converting to 20 ps RZ the sensitivity improvement increases from 2.0 dB to 2.9 dB. The improvement can also be observed in Fig. 13 over the eye patterns of the converted signals (Fig.13(b)–(d)) in comparison with that of the NRZ after 50 km SSMF (Fig.13(a)). This feature indicates that it is possible to overcome and compensate the signal degradation induced by fiber accumulated CD by converting the signal to RZ modulation format with narrower pulsewidth. Due to the slow leading and trailing edges of the NRZ signal with large CD, the conversion performance declines in the case of 80 ps converted RZ as the CD is increased as seen in Fig. 12.
We have proposed and demonstrated a multiple-channel multiple-function optical signal processor (MCMF-OSP) which realizes at the same time wavelength-waveform conversions, pulsewidth tunability, and signal regeneration. The proposed system simply uses optical parametric processing to perform AND logic function between WDM-NRZ signals and a pulsewidth tunable RZ clock. The concept of the proposed MCMF-OSP can be realized dependently on how to implement the individual modules. The results of the proof-of-concept experiment show that beside the wavelength and waveform conversions, the pulsewidth tunability for the converted RZ signals is achieved in a wide range from 20% to 80% duty cycles by tuning the pulsewidth of the RZ clock signal. A large tolerance up to ± 30 ps timing jitter (± 30% of 10 GHz bit interval) is obtained as pulsewidth is tuned at 20 ps and 40 ps. Last but not least, the MCMF-OSP also offers a further generation for input signals distorted by narrow-band optical filtering or fiber accumulated chromatic dispersion.
References and links
1. R. J. Manning, A. D. Ellis, A. J. Poustie, and K. J. Blow, “Semiconductor laser amplifier for ultrafast all-optical signal processing,” J. Opt. Soc. Am. B 14, 3204–3216 (
2. S. Bigo, O. Leclerc, and E. Desurvire, “All-optical fiber signal processing and regeneration for soliton communications,” IEEE J. Sel. Top. Quantum Electron. 3, 1208–1223 (
3. A. E. Kelly, I. D. Phillips, R. J. Manning, A. D. Ellis, D. Nesset, D. G. Moodie, and R. Kashyap, “80 Gbit/s all-optical regenerative wavelength conversion using semiconductor optical amplifier based interferometer,” Electron. Lett. 35, 1477–1478 (
4. Y. Ueno, S. Nakamura, and K. Tajima, “Penalty-free error-free all-optical data pulse regeneration at 84 Gb/s by using a symmetric-Mach-Zehnder-type semiconductor regenerator,” IEEE Photon. Technol. Lett. 13, 469–471 (
5. Y. Liu, E. Tangdiongga, Z. Li, S. Zhang, H. de Waardt, G. D. Khoe, and H. J. S. Dorren, “Error-free all-optical wavelength conversion at 160 Gb/s using a semiconductor optical amplifier and an optical bandpass filter,” J. Lightwave Technol. 24, 230–236 (
6. N. Chi, L. Xu, K. S. Berg, T. Tokle, and P. Jeppesen, “All-optical wavelength conversion and multichannel 2R regeneration based on highly nonlinear dispersion-imbalanced loop mirror,” IEEE Photon. Technol. Lett. 14, 469–471 ( 2002).
7. Y. Huang, I. Glesk, R. Shankar, and P. R. Prucnal, “Simultaneous all-optical 3R regeneration scheme with improved scalability using TOAD,” Opt. Express 14, 10339–10344 (
2006). http://www.opticsinfobase. org/oe/abstract.cfm?URI=oe-14-22-10339. [CrossRef]
8. S. Watanabe, S. Takeda, and T. Chikama, “Interband wavelength conversion of 320 Gb/s (32×10 Gb/s) WDM signal using a polarization-insensitive fiber four-wave mixer,” in Proc. European Conference on Optical Communications (ECOC), ( 1998), 85–86.
9. P. Devgan, R. Tang, V. S. Grigoryan, and P. Kumar, “Highly efficient multichannel wavelength conversion of DPSK signals,” J. Lightwave Technol. 24, 3677–3682 (
10. J. Yamawaku, H. Takara, T. Ohara, K. Sato, A. Takada, T. Morioka, 0. Tadanaga, H. Miyazawa, and M. Asobe, “Inter-band wavelength conversion of 25 GHz-spaced 1.03 Tbit/s (103×10 Gb/s) DWDM signals with small guard band and low crosstalk in PPLN waveguide,” in Proc. Conference on Lasers and Electro-Optics (CLEO), ( 2003), CThPDB2.
11. E. Yamazaki, A. Takada, J. Yamawaku, and T. Morioka, “Simultaneous and arbitrary wavelength conversion of WDM signals using multiple wavelength quasi phase matched LiNbO3 waveguide,” in Proc. Optical Fiber Communication Conference (OFC), ( 2004), FL6.
12. L. Noel, X. Shan, and A. D. Ellis, “Four WDM channel NRZ to RZ format conversion using a single semiconductor laser amplifier,” Electron. Lett. 31, 277–278 (
13. H. S. Chung, R. Inohara, K. Nishimura, and M. Usami, “All-optical multi-wavelength conversion of 10 Gbit/s NRZ/RZ signals based on SOA-MZI for WDM multicasting,” Electron. Lett. 41, 230–232 (
14. C. H. Kwok and C. Lin, “Simultaneous 4×10 Gb/s NRZ-to-RZ modulation format conversion in nonlinear optical loop mirror with a photonic crystal fiber,” IEEE Photon. Technol. Lett. 19, 1825–1827 (
16. Ch. Kouloumentas, P. Vorreau, L. Provost, P. Petropoulos, W. Freude, J. Leuthold, and I. Tomkos, “All-fiberized dispersion-managed multichannel regeneration at 43 Gb/s,” IEEE Photon. Technol. Lett. 20, 1854–1856 (
17. P. V. Mamyshev, “All-optical data regeneration based on self-phase modulation effect,” in Proc. European Conference on Optical Communications (ECOC), ( 1998), 475–476.
18. Y. Yu, X. Zhang, J. B. Rosas-Fernandez, D. Huang, R. V. Pemty, and I. H. White, “Simultaneous multiple DWDM channel NRZ-to-RZ regenerative format conversion at 10 and 20 Gb/s,” Opt. Express 17, 3964–3969 (
2009). http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-17-5-3964. [CrossRef]
19. Y. Yu, X. Zhang, J. B. Rosas-Fernandez, D. Huang, R. V. Pemty, and I. H. White, “Single SOA based 16 DWDM channels all-optical NRZ-to-RZ format conversions with different duty cycles,” Opt. Express 16, 16166–16171 (
2008). http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-16-20-16166. [CrossRef]
20. A. Sano, Y. Miyamoto, T. Kataoka, and K. Hagimoto, “Long-span repeaterless transmission systems with optical amplifiers using pulse width management,” J. Lightwave Technol. 16, 977–985 (
21. L.-S, Yan, S. M. R. Motaghian Nezam, A. B. Sahin, J. E. McGeehan, T. Luo, Q. Yu, and Alan E. Willner, “Performance optimization of RZ data format in WDM systems using tunable pulse-width management at the transmitter,” J. Lightwave Technol. 23, 1063–1067 (
22. M. Matsuura, N. Kishi, and T. Miki, “Performances of a widely pulsewidth-tunable multiwavelength pulse generator by a single SOA-based delayed interferometric switch,” Opt. Express 13, 10010–10021 (
2005). http://www.opticsexpress.org/abstract.cfm?uri=oe-13-25-10010. [CrossRef]
23. H. Nguyen Tan, M. Matsuura, and N. Kishi, “Transmission performance of a wavelength and NRZ-to-RZ format conversion with pulsewidth tunability by combination of SOA- and fiber-based switches,” Opt. Express 16, 19063–19071 (
2008). http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-16-23-19063. [CrossRef]
24. D. Zhou, B. C. Wang, R. J. Runser, I. Glesk, and P. R. Prucnal, “Perfectly synchronized bit-parallel WDM data transmission over a single optical fiber,” IEEE Photon. Technol. Lett. 13, 382–384 (
25. H. J. S. Dorren, M. T. Hill, Y. Liu, N. Calabretta, A. Srivatsa, F. M. Huijskens, H. deWaardt, and G. D. Khoe, “Optical packet switching and buffering by using all-optical signal processing methods,” J. Lightwave Technol. 21, 2–12 (
26. K. I. Kang, T. G. Chang, I. Glesk, and P. R. Prucnal, “Comparison of Sagnac and Mach-Zehnder ultrafast all-optical interferometric switches based on a semiconductor resonant optical nonlinearity,” Appl. Opt. 35, 417–426 (
28. C. Johnson, K. Demarest, C. Allen, R. Hui, K. V. Peddanarappagari, and B. Zhu, “Multiwavelength all-optical clock recovery,” IEEE Photon. Technol. Lett. 11, 895–597 (
29. V. Mikhailov and P. Bayvel, “All-optical multiwavelength clock recovery using integrated semiconductor amplifier array module,” Electron. Lett. 37, 232–234 (
30. J. Lasri, P. Devgan, V. S. Grigoryan, and P. Kumar, “Multiwavelength NRZ-to-RZ conversion with significant timing-jitter suppression and SNR improvement,” Opt. Commun. 240, 293–298 (
31. I. Brener, M. Chou, E. Chaban, K. Parameswaran, M. Fejer, and S. Kosinski, “Polarization-insensitive parametric wavelength converter based on cascaded nonlinearities in LiNbO3 waveguides,” in Proc. Optical Fiber Communication Conference (OFC), ( 2000), TuF11.
32. G. W. Lu, L. K. Chen, and C. K. Chan, “Novel NRZ-to-RZ format conversion with tunable pulsewidth using phase modulator and interleaver,” in Proc. Optical Fiber Communication Conference (OFC), ( 2006), JThB32.
33. H. Nguyen Tan, M. Matsuura, and N. Kishi, “Pulsewidth tunable NRZ-to-RZ data format conversion by combination of SOA- and fiber-based switches,” in Proc. OptoElectronics and Communications Conference (OECC/ACOFT), ( 2008), TuF-5.
34. M. Vasilyev, I. Tomkos, M. Mehendale, J.-K. Rhee, A. Kobyakov, M. Ajgaonkar, S. Tsuda, and M. Sharma, “Transparent ultra-long-haul DWDM networks with broadcast-and-select OADM/OXC architecture,” J. Lightwave Technol. 21, 2661–2672 (
35. C. Caspar, H. M. Foisel, R. Freund, U. Kruger, and B. Strebel, “Cascadability of arrayed-waveguide grating (de)multiplexers in transparent optical networks,” in Proc. Optical Fiber Communication Conference (OFC), (
1997), TuE2. [CrossRef]
36. Y. Jiang, X. Tang, J. C. Cartledge, and K. Roberts, “Electronic pre-compensation of narrow optical filtering for OOK, DPSK and DQPSK modulation formats,” J. Lightwave Technol. 27, 3689–3698 (