The impact of PDL-induced crosstalk on 100-Gb/s POLMUX RZ-DQPSK performance is investigated both experimentally and through numerical simulations in direct detection systems. We found that RZ time-interleaving, contrary to what happens with PMD, features a strong robustness to PDL. The detrimental effect of optical narrow filtering on PDL-induced penalty is also analyzed; RZ time-interleaving still proves the best solution to deal with the PDL issue.
© 2010 OSA
Spectrally-efficient multilevel formats, like differential quadrature phase-shift-keying (DQPSK), combined with polarization multiplexing (POLMUX) represent a proven solution to target the upgrade to next generation 100-Gb/s Ethernet transport . POLMUX implies a higher sensitivity to polarization effects, such as polarization mode dispersion (PMD) and polarization dependent loss (PDL). With respect to the single polarization case, penalties arise from PMD- and PDL-induced crosstalk between the demultiplexed channels and from OSNR degradation by PDL [2–4]. Moreover, the performance of the 25-Gbaud POLMUX DQPSK signal is affected by the narrow optical multiplexer and demultiplexer filtering of the 50-GHz grid in already deployed links and by further filtering of cascaded reconfigurable optical add drop multiplexers (ROADM’s).
Coherent detection schemes have widely proven to well compensate PMD and PDL [5,6], thanks to fast digital signal processing. On the other hand, less complex 100-Gb/s POLMUX systems based on direct detection can potentially enable a cost-effective terabit/s-capacity transport network [7,8]. Yet, in a POLMUX direct-detection scheme PMD- and PDL-induced crosstalk and narrow filtering severely affect system performance.
PMD- and PDL-induced penalties in direct-detection POLMUX schemes have been analyzed in literature so far in systems up to 20 Gb/s [4,9,10]. Only recently , the experimental assessment has been extended to 100-Gb/s POLMUX RZ-DQPSK: results confirm a decreased PMD tolerance (down to few ps) due to increased bit-rate. Moreover time interleaving of RZ pulses is demonstrated to be less robust than RZ time-overlapping towards PMD. Ref . takes into account only OSNR degradation due to power imbalance occurring when polarization components are aligned to PDL principal axes, whereas it disregards PDL-induced crosstalk, which causes a worse signal degradation than power imbalance, due to the orthogonality loss between the POLMUX components.
In this work we analyze the impact of PDL-induced crosstalk on 100-Gb/s POLMUX RZ-DQPSK direct-detection system performance. In particular, it is shown that, contrary to what happens with PMD, RZ interleaving proves a strong robustness to PDL, much higher than the RZ overlapping and NRZ formats. Through numerical simulations and experimental results we pursue a deep insight of this RZ-interleaving advantageous feature, not considered so far in POLMUX systems. Furthermore, when considering the impact of narrow filtering on PDL-induced penalty, we also find that the high tolerance of RZ interleaving to PDL is drastically reduced, though RZ interleaving still performs better than RZ overlapping and NRZ interleaved and overlapped formats.
2. Numerical simulations
The impact of PDL-induced crosstalk on 100-Gb/s POLMUX RZ-DQPSK has been first assessed by OSNR performance simulations; for comparison, also NRZ has been considered. In the simulations two independent 25-Gbaud DQPSK signals, obtained with nested DQPSK modulators, are polarization multiplexed. In the case of RZ-DQPSK, an RZ carver driven by a 12.5-Gb/s clock is also included. The cascade of a polarization controller (PC1), a lumped PDL element and a further polarization controller (PC2) is placed between the transmitter and the polarization demultiplexing stage. Through PC1 each channel of the POLMUX signal is launched at 45° with respect to the PDL-element axes, in order to consider the worst case for the PDL-induced crosstalk impairment. PC2 is then optimized to maximize the power of the demultiplexed channel, which is subsequently measured. This indeed is equivalent to evaluate the channel mostly affected by crosstalk, and hence mostly penalized, when a single-automatic polarization-controller standard strategy is adopted [12,13], i.e., when the crosstalk affecting the other channel is minimized. Instead, if a double polarization-controller demultiplexing is performed , the PDL-induced crosstalk can be almost eliminated on both channels, of course at the expense of the system complexity.
At the receiver side the ASE noise is added to the demultiplexed signal and then filtered. To investigate the effect of the optical filtering, simulations have been carried out setting two different bandwidths of the optical filter, that is, 1.4 nm (wide filter) and 0.25 nm (narrow filter). The demodulated signal is then received with a 23-GHz balanced differential detector.
In Fig. 1(a) wide-filter case OSNR penalty at 10−3 BER is shown. For NRZ formats the interleaving technique does not make much difference: the interleaved and overlapped POLMUX NRZ-DQPSK perform similarly, and present analogous penalty to the 10-Gb/s POLMUX NRZ-OOK reported in . It is thus confirmed that for NRZ the PDL penalty is quite independent of whether intensity or phase modulation is exploited, regardless of the bit rate. Furthermore, Fig. 1(a) shows that the application of RZ carving combined with time-interleaving makes the POLMUX RZ-DQPSK format much more robust to PDL than the NRZ format and the RZ overlapping. Indeed, with RZ interleaving only a 0.5-dB penalty is found with 3-dB PDL, whereas the NRZ interleaving shows 7.2-dB penalty and both NRZ and RZ overlapping floor.
The reason of such a better performance of RZ interleaving is the following: at the demultiplexing stage, due to PDL-induced deorthogonalization, in maximizing one channel the other channel is not completely blocked, leading to crosstalk. If the two POLMUX RZ-tributaries are overlapped this crosstalk is maximum, since occurs on the entire pulse, while for interleaved tributaries only a residual crosstalk occurs at the edges of the symbol, thus minimizing the impact on the detected channel. This behaviour was already noticed in  for the 10-Gb/s NRZ-OOK overlapped and interleaved cases. Yet, as also shown here in Fig. 1(a), the benefit of interleaving is almost negligible with NRZ format, as the pulse duration coincides with the entire symbol slot. The results here reported [Fig. 1(a)] instead clearly prove much more advantage in exploiting the interleaving technique combined with RZ format, thanks to the reduced RZ-pulse duration.
Figure 1(b) reports simulated curves obtained assuming a 0.25-nm narrow optical filter. Besides the 50-GHz grid of deployed systems, the optical bandwidth may in fact be reduced further by the presence of cascaded ROADM’s. By comparing Fig. 1(b) and Fig. 1(a), it is evident that the combined presence of PDL-induced crosstalk and tight optical filtering reduces the high tolerance of RZ interleaving towards PDL, e.g., 5.5-dB OSNR penalty now occurs with 3-dB PDL. On the contrary, the effect of narrow filtering only slightly increases OSNR penalty for the RZ-overlapped case and leaves almost unchanged NRZ penalty. Narrow filtering on the larger RZ-DQPSK spectrum causes in fact a temporal broadening of the interleaved pulses, which thus overlap significantly, giving a condition similar to NRZ. Nevertheless, results confirm that RZ-interleaved DQPSK is still to be preferred with respect to the three other formats as far as the PDL issue is concerned. For example, with 2-dB PDL a 2.4-dB OSNR penalty is foreseen for RZ- interleaved case, with respect to the 6.4-dB OSNR penalty of the RZ-overlapped case.
Actually the penalty curves reported in Fig. 1 refers to the channel mostly affected by PDL-induced crosstalk. On the other hand we verified by simulations that the orthogonal channel is not appreciably penalized, in agreement with what reported in  for NRZ case.
As already evidenced, Fig. 1-OSNR penalty are obtained when the SOPs of the POLMUX channels are oriented at 45° with respect to PDL axes, thus no PDL-induced power imbalance occurs, as both POLMUX channels are attenuated in the same manner. Yet, this PDL-induced attenuation on each channel must be taken into account in power budget planning of optical transmission systems. The additional power attenuation, with respect to the least-lossy PDL axis, is plotted in Fig. 2 as a function of PDL. For PDL values up to 3-4 dB, the PDL-induced attenuation is within 1.5 dB while, for very high PDL values the attenuation approaches 3-dB as each POLMUX channel looses one polarization component.
3. Experimental results
To experimentally evaluate the PDL-induced crosstalk impairment on 100-Gb/s POLMUX RZ-DQPSK and verify simulated results of Fig. 1, we exploited the set-up sketched in Fig. 3 . A DFB laser is RZ carved at 12.5 GHz and DQPSK modulated at 25 Gbaud. The modulated signal is split in two replica uncorrelated by a 1-km fiber and multiplexed through a polarization beam combiner (PBC).
The POLMUX signal at an overall bit-rate of 100-Gb/s is launched into a PDL emulator realized by means of a small ring (few centimeters long) of polarization maintaining (PM) fiber . By varying the PM-fiber ring radius different PDL values are obtained. Before the demultiplexing stage (PC2 + PBS) the signal is tapped and sent to a polarization analyzer to guarantee a 45° launch of the orthogonally polarized channels with respect to the PDL emulator axes. After demultiplexing, the signal is optically filtered either with a 1.4-nm or a 0.25-nm filter, in order to assess the impact of filtering on PDL-induced crosstalk penalty. The in-phase and quadrature components of the demultiplexed RZ-DQPSK are then demodulated with a Mach-Zehnder delay interferometer (MZDI) and received by a 23-GHz balanced differential detector. POLMUX RZ-DQPSK performance is evaluated in term of BER versus received OSNR, varied by means of an ASE adder at the receiver.
Figure 4 shows BER versus OSNR curves for PDL values from 0.6 dB to 2.7 dB, for the case of wide (a) and narrow (b) optical filtering. Corresponding PDL values in Fig. 4(a) and Fig. 4(b) are slightly different due to the precision in manually adjusting the PM-fiber ring radius. BER measurments refer to the channel whose power is maximized after demultiplexing and thus mostly affected by PDL-induced crosstalk. Measured BER curves clearly evidence the advantage of exploiting RZ interleaving (continuous lines) with respect to RZ overlapping (dashed lines), even in presence of narrow optical filtering. Figure 4(b) experimentally confirms that the narrow optical filtering combined with not negligible PDL significantly impairs the BER curves. Indeed, at 1.9-dB PDL, while RZ interleaving curve still presents limited penalty, the corresponding RZ overlapping one shows an error floor.
To better appreciate the beneficial impact of RZ interleaving on system performance in presence of PDL-induced crosstalk, in Fig. 5 we have plotted the measured OSNR required to achieve 10−3 BER versus PDL values, for the case of both wide (a) and narrow (b) optical filtering. Corresponding simulated curves are also reported (straight line). A good agreement between experimental and simulated results is found. Indeed the measured RZ-interleaved OSNR penalty in Fig. 5(a) proves lower compared to the corresponding one in Fig. 5(b) and almost negligible if compared to both the RZ-overlapped cases.
We have also measured the orthogonal channel performance (i.e. the channel with minimized crosstalk) and verified that no significant penalty occurs. Therefore the overall performance of the POLMUX system is limited by the measured PDL-induced penalties reported in Fig. 5.
In this work we have assessed both with numerical simulations and experiments the robustness of 100-Gb/s POLMUX RZ-DQPSK to PDL-induced crosstalk. As already shown in past works at lower bit rates, in direct detection schemes PDL is a significant source of impairment for system performance.
Yet, our investigation has evidenced that RZ interleaving demonstrates a potential solution to cope with the PDL issue, alternatively to PDL-compensation schemes [7,15], when demultiplexing is performed exploiting a single automatic polarization controller. Indeed, even at high PDL values (up to 5 dB) the OSNR penalty remains negligible if RZ interleaving is exploited. The results here reported thus clearly prove a more advantageous behaviour of the interleaving technique combined with RZ shaping compared to the NRZ case, thanks to the RZ reduced pulse duration.
Our analysis has also highlighted that narrow optical filtering, which occurs in deployed links including several ROADM’s, reduces the high robustness of RZ interleaving to PDL. Nevertheless, even with thigh filtering RZ interleaving still proves the best condition to deal with the PDL issue in POLMUX direct detection systems with single automatic polarization controller.
References and links
1. C. R. S. Fludger, T. Duthel, and C. Schulien, “Towards Robust 100G Ethernet Transmission,” Proceeding LEOS Summer Topical Meetings 2007 Digest of the IEEE, 224–225 (2007).
2. P. Boffi, M. Ferrario, L. Marazzi, P. Martelli, P. Parolari, A. Righetti, R. Siano, and M. Martinelli, “Measurement of PMD tolerance in 40-Gb/s polarization-multiplexed RZ-DQPSK,” Opt. Express 16(17), 13398–13404 (2008). [CrossRef] [PubMed]
3. T. Duthel, C. R. S. Fludger, J. Geyer, and C. Schulien, “Impact of polarization dependent loss on coherent POLMUX-NRZ-DQPSK,” Proceeding OFC 2008, paper OThU5, San Diego- CA, USA (2008).
6. C. R. S. Fludger, T. Duthel, D. van den Borne, C. Schulien, E.-D. Schmidt, T. Wuth, J. Geyer, E. De Man, G.-D. Khoe, and H. de Waardt, “Coherent equalization and POLMUX-RZ- DQPSK for robust 100-GE transmission,” J. Lightwave Technol. 26(1), 64–72 (2008). [CrossRef]
7. H. Wernz, S. Bayer, B. Olsson, M. Camera, H. Griesser, C. Fuerst, B. Koch, V. Mirvoda, A. Hidayat, and R. Noé, “112Gb/s PolMux RZ-DQPSK with fast polarization tracking based on interference control,” Proceeding OFC 2009, paper OtuN4, San Diego- CA, USA (2009).
8. S. Chandrasekhar, X. Liu, E. C. Burrows, and L. L. Buhl, “Hybrid 107- Gb/s polarization-multiplexed DQPSK and 42.7-Gb/s DQPSK transmission at 1.4-bits/s/Hz spectral efficiency over 1280 km of SSMF and 4 bandwidth-managed ROADMs,” Proceeding ECOC 2007, paper PD1.92, Berlin (2007).
9. H.-C. Ji, J. H. Lee, H. Kim, P. K. J. Park, and Y. C. Chung, “Effect of PDL-induced coherent crosstalk on polarization-division-multiplexed direct-detection systems,” Opt. Express 17(3), 1169–1177 (2009). [CrossRef] [PubMed]
11. S. Chandrasekhar, and X. Liu, “Experimental investigation of system impairments in polarization multiplexed 107-Gb/s RZ-DQPSK,” Proceeding OFC 2008, paper OThU7, San Diego- CA, USA (2008).
12. P. Martelli, P. Boffi, M. Ferrario, L. Marazzi, P. Parolari, S. Pietralunga, R. Siano, A. Righetti, and M. Martinelli, “Polarization stabilizer for polarization-division multiplexed optical systems,” Proceeding ECOC 2007, paper 6.6.5, Berlin (2007).
13. B. Milivojevic, A. F. Abas, A. Hidayat, S. Bhandare, D. Sandel, R. Noé, M. Guy, and M. Lapointe, “1.6-b/s/Hz 160-Gb/s 230-km RZ-DQPSK polarization multiplex transmission with tunable dispersion compensation,” IEEE Photon. Technol. Lett. 17(2), 495–497 (2005). [CrossRef]
14. A. B. dos Santos and J. P. von der Weid, “PDL effects in PMD emulators made out with HiBi fibers: building PMD/PDL emulators,” IEEE Photon. Technol. Lett. 16(2), 452–454 (2004). [CrossRef]
15. C.-S. Kim, B. Choi, J. S. Nelson, P. Z. Dashti, and H. P. Lee, “Novel PDL/PDG compensator for transmission optical devices using Sagnac interferometer” Proceeding OFC 2005, paper OFK7, Anaheim- CA, USA (2005).