High-rate long-haul fiber communications systems will require high-speed efficient signal regeneration. By introducing a simple semiconductor-optical-amplifier wavelength converter into the folded ultrafast nonlinear interferometer, we demonstrate a polarization-insensitive wavelength-maintaining 3R all-optical regenerator. The extinction ratio between the on and off states is 21 dB. Bit-error rate data show no error floor and <0.3-dB power penalty when compared with a baseline measurement. Moreover, changes in input polarization cause no change in performance. The regenerator maintained a constant polarization and constant output power, independent of the input polarization.
© 2006 Optical Society of America
As channel line rates in optical networks become faster, penalties from dispersion and nonlinearities in optical fiber become more severe. Long-haul propagation in systems operating at high channel-rates will require ultrafast regeneration that re-amplifies, reshapes, and retimes the optical signal (3R regeneration). In today’s wavelength-division multiplexed fiber networks, full 3R data regeneration requires the costly and inefficient demultiplexing of the wavelength channels down to electronic rates and requires an optical-to-electronic-to-optical conversion of each wavelength channel. Researchers have shown all-optical switching, on the other hand, to be capable of high-rate regeneration >84 Gbit/s [1, 2]. All-optical switches, when applied to 3R regeneration, use the network data pulses as control pulses to transfer the data pattern onto locally generated pulses that are transform limited, have low jitter and, a high signal-to-noise ratio (SNR) . Desired properties of all-optical switches used in 3R regeneration include data-pattern independence, ultrafast operation, insensitivity to network data pulse polarization, stability against vibrations and other environmental changes, wavelength-maintaining operation, low-power requirements, and constant output polarization.
Researchers have adapted a number of all-optical switches to 3R regeneration, but many do not satisfy all the properties listed above. The nonlinear optical loop mirror (NOLM) has been demonstrated as a regenerator , but is sensitive to acoustic vibration. Moreover, at a fixed pulse width, the contrast between switched and unswitched pulses degrades with higher data rates due to the effects of the control pulses on the counterpropagating signal pulses. A semiconductor-optical-amplifier- (SOA-) based ultrafast nonlinear interferometer (UNI)  has been used to demonstrate 3R regeneration at 80 Gbit/s , but long-lived carrier density recovery times in the SOA (~100’s ps) result in pattern-dependent amplitude patterning of the switch output when using intensity-modulated data formats . The SOA-based interferometer in Ref.  requires low input pulse energies, but is bit inverting and does not maintain the input wavelength. It has been demonstrated successfully in all-optical regeneration .
In Ref. , we presented the folded ultrafast nonlinear interferometer (FUNI) all-optical switch for 3R regeneration. The geometry of the FUNI’s design makes it insensitive to the environmental changes and acoustic vibrations that alter polarizations within the switch itself. Moreover, the nearly instantaneous response of the FUNI’s nonlinear medium, dispersion-shifted optical fiber (DSF), gives the FUNI its ultrafast operation and eliminates pattern-dependent amplitude patterning at the output. The FUNI also maintains a constant output polarization. The FUNI has been demonstrated to operate at 40 Gbit/s with ~2 ps pulses . When operated at 10 Gbit/s, the short pulse widths in the FUNI suggest other regeneration applications. One application area is in access points of time-division multiplexed networks. Another is in free-space communications where short broad-spectrum pulses are required for high-sensitivity optical receivers. There are two possible disadvantages in the FUNI’s operation: its output wavelength is different from that of the incoming data, and its output power is sensitive to changes in the incoming data polarization.
In this letter, we demonstrate a wavelength-maintaining polarization-insensitive FUNI 3R regenerator by incorporating an SOA wavelength converter into the switch design. The SOA wavelength converter shifts the incoming data wavelength to an intermediate wavelength, which the FUNI then regenerates back onto the original data wavelength. Like the FUNI, this SOA wavelength converter  has been demonstrated at rates up to 40 Gbit/s . By itself, the SOA wavelength converter cannot serve as a 3R regenerator. It does not, for example, retime the incoming network data. Moreover, in Ref.  there is a ~2-dB penalty in the bit-error rate (BER) due to patterning caused by the SOA. Also, we have observed that the output power of the SOA wavelength converter can drop as much as 0.6 dB with a change in the polarization of the incoming data. It does, however, have the advantage of simple and stable operation.
Incorporating the SOA wavelength converter into the FUNI eliminates some of the problems that each device suffers from separately, thus enabling stable wavelength-maintaining all-optical 3R regeneration. The nonlinear power response of the FUNI reduces amplitude patterning on incoming data pulses introduced by the SOA wavelength converter. Also, because the SOA wavelength converter outputs a fixed polarization, the sensitivity of the FUNI to the input signal polarization is reduced, providing a polarization-insensitive regeneration operation. In this way, the wavelength-maintaining FUNI, in addition to re-amplifying, retiming, and reshaping, also repolarizes the data pulses, producing an output of a single polarization and amplitude independent of the input polarization. These properties lead to stable and simple operation, requiring no active monitoring of the input polarization or of the polarizations within the switch itself.
2. Device description
Figure 1 shows a block diagram of the wavelength-maintaining polarization-insensitive FUNI (WM-FUNI). In order to regenerate the network data onto the network data wavelength, we incorporated a wavelength converter (shown in the dashed box) that shifted the network data to an intermediate wavelength, set by the CW source. In the wavelength converter, incoming data pulses enter an SOA along with the CW beam. The SOA is an Alcatel A1901SOA biased with a 200 mA current. In the SOA, the network data pulses broaden the spectrum of the CW beam through cross-phase modulation (XPM) and cross-gain modulation (XGM). A filter then notches out the CW and band-passes one side of the expanded CW spectrum. This wavelength converter is discussed at greater length in Ref.  and Ref. .
The wavelength converted data pulses enter the FUNI. The local clock source is a pulse train synchronized with the data pulses and at the original network data wavelength. First, consider the FUNI’s operation in the absence of data pulses. The clock pulse source produces a polarization that transmits completely through the polarizer. We align the birefringent fiber (BRF) at 45° to the polarizer’s transmission axis, thus splitting the clock pulse into two temporally-separated orthogonal polarizations. The separation time between the polarizations is 10 ps. Both components propagate through 1 km of DSF, with a dispersion zero at 1551.4 nm, after which a Faraday mirror reflects the pulses and rotates them by 90°. This 90° polarization rotation compensates, during the return pass through the DSF and BRF, all birefringence accumulated in the forward pass. The polarizer then blocks the temporally-recombined clock pulses because they are now rotated by 90° from their original polarization. The FUNI turns on when a data pulse enters the FUNI with the clock pulse. Each data pulse copropagates with one of the two temporally-separated orthogonal clock pulse polarizations. During propagation through the DSF, the nonlinear interaction between data and clock pulses induces a 180° relative phase shift between the orthogonal clock components. This phase shift rotates the clock pulse polarization so that, upon arrival at the polarizer, the clock pulse is aligned to the polarizer transmission axis and exits out port 3 of the circulator. A bandpass filter (BPF2) passes the clock pulse and blocks the data pulse. Thus, the on-off keyed data pattern is regenerated onto the clock pulse wavelength. Counterpropgating data pulses interact with both orthogonal clock polarizations, so the contrast problem with the NOLM at higher data rates, mentioned in the introduction, does not appear in the FUNI. There may, however, be unwanted nonlinear polarization rotation effects at higher data rates.
3. Experiment and results
A 231-1 pseudorandom bit pattern was modulated onto a 10-GHz pulse train centered at 1547.2-nm with a pulse intensity full-width at half maximum (IFWHM) of 2-ps. A 1557.1-nm CW source provided the intermediate wavelength for the output data pulse sequence of the SOA wavelength converter. The pulse energy of the incoming data pulses was 93 fJ/pulse and the CW power was 3.2 mW (measured at the SOA input). The output pulses of the SOA wavelength converter had an IFWHM of 4.7 ps and center wavelength of 1557.8 nm. These pulses entered the FUNI data pulse port. The clock pulses had pulse widths of 2 ps, a repetition rate of 10 GHz, and a center wavelength of 1547.2 nm, and entered the FUNI’s clock source port. Figure 2 shows autocorrelations of the output of the wavelength converter and of the WM-FUNI regenerator. The IFWHM of the WM-FUNI’s output was 3.2 ps. The time-bandwidth product (IFWHM) of the wavelength-converter output was 0.56 and of the WM-FUNI was 0.53, compared with a time-bandwidth product of 0.44 for unchirped Gaussian pulses. To ensure these regenerated pulses propagate we plan future transmission tests in a recirculating loop.
In 3R regeneration, the switch must restore the pulse quality. To verify the regenerative qualities of the WM-FUNI, we tested the output pulse quality after the all-optical switching operation with a bit-error rate test using an optically pre-amplified direct-detection receiver. The BER curves plot the bit-error rate measured by the receiver versus the optical power at the receiver input. Figure 3 shows BER curves for the output of the wavelength converter (taken at A in Fig. 1), for the output of the WM-FUNI (taken at B in Fig. 1), and for the transmitter output going directly into the receiver (“Back-to-back”). The data quality is maintained and that the output pulses remain well-matched to the receiver. There is a 0.8-dB penalty for the wavelength converter compared to the back-to-back at BER=10-9. The WM-FUNI restores the BER to <0.3-dB penalty compared to back-to-back performance.
A 3R regenerator should also be insensitive to the polarization of the incoming data pulses. In the SOA wavelength converter, a change in data polarization has been experimentally measured to cause a 0.6-dB drop in output power and a 0.5-dB error penalty as measured at point A in Fig. 1. The FUNI, without the SOA wavelength converter, is also sensitive to the data polarization. Polarization fluctuations at point A in Fig. 1 can cause up to a 2.1-dB drop in output power and a 0.6-dB penalty in the BER as measured at point B. The cascade of the wavelength converter and the FUNI, however, is polarization independent: both the measured power drop and measured error penalty are 0 dB. The FUNI’s nonlinear response curve, shown in Fig. 4, dampens amplitude patterning in the wavelength converter’s output that was caused by data-polarization changes and gain saturation in the SOA. By interpolating a raised cosine curve to the data in Fig. 4, we estimate that a ±0.5 dB variation in the output power at the curve’s peak leads to only a ±0.06 dB variation in the output power.
We demonstrate the WM-FUNI’s polarization insensitivity with the input polarization varying over all polarizations. We took a BER curve of the WM-FUNI with the polarizations manually adjusted to the optimal bit-error rate and compared it to a BER curve of the WM-FUNI with a polarization scrambler at the data input. The BER curves in Fig. 5 show <0.1 dB penalty between the two curves. The wavelength converter provides the FUNI with a constant polarization at point A, giving the combined switch this polarization insensitivity. To our knowledge this regenerator is the first demonstrated with both constant output polarization and <0.1 dB sensitivity to input polarization.
In conclusion, we have demonstrated a wavelength-maintaining polarization-insensitive FUNI 3R regenerator. In addition to re-amplifying, retiming, and reshaping the incoming optical data, as is typically demonstrated in 3R regenerator designs, the wavelength-maintaining FUNI also repolarizes the data by giving a constant output polarization and amplitude independent of the input polarization. This design was tested at 10 Gbit/s with pulse widths of ~3 ps. The wavelength-maintaining FUNI is well-suited for all-optical 3R regeneration of high-rate short-pulse data streams. It shows <0.3-dB power penalty when compared to the baseline BER at 10-9. The wavelength-maintaining FUNI, unlike its two constituent switches, is insensitive to input polarization, shows no drop in power with changes in incoming data polarization, and incurs a 0-dB performance penalty. Both the wavelength converter and the FUNI are stable with respect to acoustic vibrations and other environmental effects, and the wavelength-maintaining FUNI requires no active monitoring of the input polarization or of polarizations within the switch itself. In all, the wavelength-maintaining FUNI 3R regenerator simultaneously is data-pattern independent, has ultrafast operation, is stable against vibrations and other environmental changes, is wavelength-maintaining, is insensitive to network data pulse polarization, and gives constant output polarization.
This work was supported by the Air Force Research Laboratory (AFRL), under Air Force contract #FA8721-05-C-0002. Opinions, interpretations, recommendations and conclusions are those of the authors and are not necessarily endorsed by the United States Government.
References and links
1. 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 Photonics Technol. Lett. 13, 469–471 (2001). [CrossRef]
2. J. Leuthold, G. Raybon, Y. Su, R. Essiambre, S. Cabot, J. Jaques, and M. Kauer, “40 Gbit/s transmission and cascaded all-optical wavelength converter over 1 000 000 km,” Electron. Lett. 38, 890–892 (2002). [CrossRef]
3. O. Leclerc, B. Lavigne, C. Duchet, C. Janz, and E. Desurvire, “All-optical 3R regeneration,” in Technical Digest of the Conference on Lasers and Electro-Optics, (Optical Society of America, Washington DC, 2000), pp. 407–408.
4. M. Jinno, “All optical signal regularizing/regeneration using a nonlinear fiber Sagnac interferometer switch with signal-clock walk-off,” J. Lightwave Technol. 12, 1648–1659 (1994). [CrossRef]
5. N. S. Patel, K. L. Hall, and K. A. Rauschenbach, “Interferometric all-optical switches for ultrafast signal pocessing,” Appl. Opt. 37, 2831–2842 (1998). [CrossRef]
6. 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 (1999). [CrossRef]
7. B. S. Robinson, J. D. Moores, and D. T. Moriarty, “Pattern independent semiconductor-based interferometric all-optical switching using pulse-position modulation,” in Technical Digest of the Conference on Lasers and Electro-Optics, (Optical Society of America, Washington DC, 2000), pp. 695–696.
8. Y. Ueno, S. Nakamura, K. Tajima, and S. Kitamura, “3.8-THz wavelength conversion of picosecond pulses
using a semiconductor delayed-interference signal-wavelength converter (DISC),” IEEE Photonics Technol. Lett. 10, 346–348 (1998). [CrossRef]
9. S. J. Savage, B. S. Robinson, S. A. Hamilton, and E. P. Ippen, “All-optical pulse regeneration in an ultrafast nonlinear interferometer with Faraday mirror polarization stabilization,” Opt. Lett. 28, 13–15, (2003). [CrossRef]
10. B. S. Robinson, S. A. Hamilton, S. J. Savage, and E. P. Ippen, “40 Gbit/s all-optical XOR using a fiberbased folded ultrafast nonlinear interferometer,” in Technical Digest of the Optical Fiber Communications Conference (Optical Society of America, Washington DC, 2002), pp. 561–563.
11. J. Leuthold, D. M. Marom, S. Cabot, J. J. Jaques, R. Ryf, and C. R. Giles, “All-optical wavelength conversion using a pulse reformatting optical filter,” J. Lightwave Technol. 22, 186–192 (2004). [CrossRef]
12. M. L. Nielsen, B. Lavigne, and B. Dagens, “Polarity-preserving SOA-based wavelength conversion at 40 Gbit/s using bandpass filtering,” Electron. Lett. 39, 1334–1335 (2003). [CrossRef]