Abstract
We investigate polarization mode dispersion (PMD) and polarization dependent loss (PDL) impairments in polarization division multiplexing (PDM) signals with optical polarization demultiplexing and direct detection. We find that the time alignment between the bits in the two polarizations has a significant impact on the PMD impairments, and PMD impairments also depend on the bandwidth of PDM signals, whereas PDL impairments have little dependence on the relative time alignment between the two polarizations and the signal bandwidth. We show that with a proper configuration of the polarization demultiplexing, the PDL-induced crosstalk between the two polarizations can be completely eliminated. The combined effects of PMD and PDL are also studied, and we find that, in the presence of concatenated PMD and PDL, the impairment from one effect does not enhance that from the other.
©2009 Optical Society of America
1. Introduction
Increasing the spectral efficiency of fiber-optic communication systems is an effective way to meet the ever growing demand for transmission capacity and fully exploit the potential of current deployed fiber infrastructure. Polarization division multiplexing (PDM), which transmits two channels with orthogonal states of polarization (SOPs) at an identical wavelength, can double the spectral efficiency and therefore attracts much attention [1-3]. PDM signals are more sensitive to polarization effects in fiber-optic communication systems than signals with a single polarization. Two important polarization effects in a fiber-optic communication system are polarization mode dispersion (PMD) and polarization dependent loss (PDL). PMD mainly arises from the random birefringence in fibers and optical components, in which signals with different SOPs travel at different speeds. PDL usually occurs in optical components, such as isolators and couplers, whose insertion loss varies with the SOPs of input signals. In addition to the penalties from PMD-induced pulse broadening [4] and PDL-induced variation of optical signal to noise ratio (OSNR) [5-8], the main impairments caused by PMD and PDL in PDM signals are the crosstalk between the two polarizations. At the receiver side of PDM systems, the two channels are separated with polarization demultiplexing either in the electronic domain with coherent detection or in the optical domain with direct detection. In the coherent detection scheme, the PMD-induced impairments can be well compensated with digital signal processing [9-11], thus PDL is the primary source of system degradation and has been carefully studied [11] [12]. However, in the direct detection scheme, both PMD and PDL effects put significant limitations on system performance. So far, only limited work has been done on separate PMD or PDL penalties for PDM signals [13-19], and there is no detailed assessment on the combined penalties in literature.
In this paper, we study the PMD and PDL impairments first separately and then jointly, in PDM signals with optical polarization demultiplexing and direct detection. We first analyze the crosstalk induced by PMD and PDL in PDM signals, and then experimental results on 10-Gb/s return-to-zero (RZ) and non-return-to-zero (NRZ) on-off-keying (OOK) signals are presented.
2. Crosstalk Mechanisms
2.1 PMD
PMD causes the output SOP of a fully polarized input signal to vary with frequency. This depolarization means that the two channels originally having orthogonal polarizations cannot be completely separated with a polarization beam splitter (PBS) in the presence of PMD as signals have a certain bandwidth, thus coherent crosstalk is induced in the PDM system. Assuming there is only PMD, after transmission, the signal at the output of a polarizer (one port of a PBS) can be expressed as
where is the input complex signal, and are the Jones matrices to describe PMD and the polarizer, respectively. We assume that a PDM signal with channels and at two orthogonal polarization states is launched into a first-order PMD emulator (PMDE). If a PBS aligned at the SOPs of the center frequency components of the signal is used to demultiplex the two channels, from Eq. (1) we can get the output signal at the port Awhere θ is the angle between the input SOP of and one principal state of polarization (PSP) of the PMDE, ω is the deviation of the angular frequency component of the signal from its center frequency, and are the input signals of and , respectively, and is the differential group delay (DGD) of the PMDE. Three conclusions can be obtained from Eq. (2): 1) When(PSP launching), there is no crosstalk and when , there is the largest frequency dependent crosstalk. The crosstalk increases monotonically when θ increases from to or decreases from to; 2) The crosstalk from to depends on the bandwidth of . with a larger bandwidth will be more depolarized and thus induces larger crosstalk to ; and 3) As most frequency components of a signal are contained in the transition period of pulses, the relative time delay between the two polarization channels has a big impact on the crosstalk.2.2 PDL
Except for the principal axes launching, after passing through a PDL emulator (PDLE), two originally orthogonally polarized signals become non-orthogonal. We assume that the PDLE is aligned with x and y axes (y axis is the least lossy axis), and a PDM signal is launched into the PDLE as shown in Fig. 1(a) . The normalized input signals are
where θ is the angle between the SOP of and y axis of the PDLE in the Jones space.After the PDLE, the normalized output signals are
where α is related to the PDL value Γ (dB) asThe angle between the output SOPs of channels and can be expressed as
which indicates that the output angle reaches its local minimum if the launched polarization is at , as shown in Fig. 1(b). The loss of orthogonality due to PDL will cause crosstalk between the two channels if only one PBS is used. However, by using two PBSs with each one set to block the signal from the other polarization, the crosstalk can be completely eliminated, at the cost of power reduction of the desired channel [1]. Note that PDL is generally considered as frequency-independent and it alone will not reshape a fully polarized signal; in the presence of modest PMD, the pulse distortion effects related to PDL is considerably small [20] [21]. Therefore, the PDL related intra-channel waveform degradation is not discussed in this paper, but it is included in the experimental results.2.3 PMD and PDL
A real system has both PMD and PDL, but the combined effects of PMD and PDL on a PDM signal have never been investigated before. Two simple cases are studied here. One is PDL after PMD and the other is PMD after PDL. Although PMD and PDL are distributed in a real system, these two simple cases can help us understand of the interaction between PMD and PDL impairments. The similar simplification of real systems has been implemented to study the combined effects of PMD and PDL on single polarization signals [22].
With a PDLE concatenated after a PMDE, from Eq. (6), when the PDLE input angle (relative to y axis of the PDLE) of a frequency component is θ, the output angle can be expressed as . This input-output transformation of the angles can be characterized as the solid curve in Fig. 2 , whose curvature increases as the PDL value increases. As the launch angle increases from to , the corresponding slope of the transformation curve increases from smaller than 1 to larger than 1. There is a sole critical angle whose corresponding slope is 1, i.e., where the output angle variation equals the input angle variation. The critical angle becomes larger as the PDL value increases, and it is always larger than as long as PDL is present.
For the PMD-degraded PDM signal launched into the PDLE, assuming that the launch angles of the center frequency component and the most deviated frequency component of the same channel are and , respectively, if both angles are smaller then the critical angle, the output angle difference will be smaller than the input angle difference, which means that a signal depolarized by PMD gets re-polarized by PDL [23].
If the launch angle of one channel is pushed beyond the critical angle and falls into the divergence zone in Fig. 2, the output SOPs of this channel are more divergent and larger intra-/inter-channel penalty is induced, however, as the launch angles of the two channels are complementary, the launch angle of the other channel will be pushed further into the convergence zone with more convergent SOPs and smaller penalty. Meanwhile, as indicated in Fig. 1(b), the output angle between the two channels increases towards as the launch angle of one channel is pushed further into the divergence zone, suggesting that the inter-channel crosstalk induced by the loss of SOP orthogonality is alleviated. Thus, to assess the largest PDL-induced penalty on a PMD-degraded PDM signal, the launch angle can be set as relative to the least lossy axis of the PDLE.
Figure 3 shows the relation of signal SOPs in the case of a PMDE concatenated after a PDLE. Based on Eq. (6), when the launch angle of is set to be relative to the PSPs of the PMDE in the Jones space, the launch angle of varies, depending on the relation between SOPs of and and the PSPs. The smallest angle between the SOP of and the PSPs of the PMDE is when the SOPs of and and the PSPs are in the same plane, and the largest angle is when the plane of SOPs of and is perpendicular to the PSPs in the Stokes space.
There is the largest PMD-induced penalty when the launch angles of the two channels are both , but it is hard to know in an experiment what the launch angle of is. Practically in most cases, as indicated in Fig. 3, the launch angle of is within [,], and the potential largest penalty difference between the worst case scenario and an angle-uncertain launch is the difference of PMD-induced intra-channel penalty (pulse broadening penalty) between and launching, thus we can evaluate the worst case while only one channel is assured to be launched at .
3. Experiment Results and Discussions
3.1 Experimental Setup
The experimental setup is shown in Fig. 4 . A distributed feedback (DFB) laser with a center wavelength of 1561.31nm is modulated with 231-1 pseudo-random binary sequence (PRBS) 10-Gb/s data to generate a 10-Gb/s NRZ-OOK signal. To generate a 50% duty cycle RZ signal, a pulse carver is inserted after the data modulator. After the modulators, the signal is split by a 50/50 coupler into two branches. A delay line (DL) is inserted in one branch to make the bit sequences of the two channels interleaved (the relative time delay is , where is the bit period and n is an integer) or synchronized (the relative time delay is ) in the time domain, and a variable optical attenuator (VOA) is applied in the other branch to equalize the power of the two channels. In each branch the signal is adjusted with a polarization controller (PC) to get the proper polarization, and then combined with the polarization beam combiner (PBC). The PDM signal, whose SOP is controlled with PC3 to demonstrate the worst-case scenarios, is launched to the PMDE or/and the PDLE and both the input and output of the emulators are monitored with the polarization analyzer. When both the PMDE and the PDLE are present, a PC is inserted in between. The output signal is combined with an amplified spontaneous emission (ASE) noise source to vary OSNR. A small portion of signal power is tapped out to an optical spectrum analyzer (OSA) for OSNR monitoring. The polarization demultiplexing is achieved with a PBS and a manually adjusted PC.
3.2 Measurement with Only PMD
The system PMD performance is measured in six cases: NRZ one-polarization/interleaved-PDM/synchronized-PDM and RZ one-polarization/interleaved-PDM/synchronized-PDM signals. Each channel is launched relative to the PSPs of the PMDE.
Figure 5 shows the measured OSNR penalties at bit-error-ratio (BER) of 10−3 versus DGD. The right insets of Fig. 5 show typical eye diagrams of the degraded signals. As expected, for the single polarization signal, RZ has better tolerance to PMD than NRZ [24]. However, for the PDM signals, the NRZ signal can tolerate more PMD than the RZ signal in the synchronized case, but has similar tolerance to PMD as the RZ signal in the interleaved case. This can be explained with the eye-diagrams, which show that the crosstalk from to mainly occurs in the transition period of the pulses of and it drops to minimum at the pulse center. When and are interleaved, the crosstalks for both RZ and NRZ signals are almost the same and located at the center of each bit slot, whereas in the synchronized case, the crosstalk peak is located closer to the bit center for the RZ signal than that for the NRZ signal. Figure 5 indicates that by properly choosing the modulation format and time delay, the PMD tolerance of a PDM signal can be improved.
3.3 Measurement with Only PDL
To assess the PDL impairment, each channel of the PDM signal is launched at relative to the least lossy axis of the PDLE, and three cases are measured. Case 1: the PC4 is optimized to get the maximum power from channel at one PBS port. Based on Eq. (6), there is a maximum crosstalk from channel and the amount is , which is defined as the ratio of signal power from the other channel to that of the desired channel. Case 2: the PC4 is optimized to block at one port, and there is only from this port. The crosstalk is eliminated, but the power of is reduced with a ratio of . Case 3: the PC4 is optimized to get equal crosstalk at both PBS ports, and the crosstalk is for each channel. Note that the demultiplexing setup in Case 2 will be used to analyze the joint effect of concatenated PDL and PMD in the next section.
Figure 6 shows the measured OSNR penalties of the PDM NRZ-OOK signal at BER of 10−3 as a function of PDL value. The result for the RZ signal is similar as the PDL-induced penalties are independent of bandwidth. This bandwidth-independence can be observed in the eye diagrams of Fig. 6, where the PDL value is 3dB. Figure 6 shows that there is a big difference in the PDL penalties among the three cases. With the crosstalk completely eliminated (Case 2), the PDM signal can tolerate a large PDL value. Note that the penalty for the time synchronized case is slightly larger than that for the time interleaved one. This is because the crosstalk in the interleaved case mainly occurs at the edge of the bit slot while in the synchronized case it occurs at the center; and in the latter case the crosstalk has a larger influence on BER.
3.4 Measurement with Concatenated PMD and PDL
The joint effects of PMD and PDL are measured in two regimes, i.e., the PMD-PDL regime and the PDL-PMD regime.
In the PMD-PDL regime, the PDM signal is at first launched into the PMDE then the PDLE, with the SOP of the center frequency component of each channel relative to the PSPs of the PMDE and the least lossy axis of the PDLE. At the receiver side, the polarization demultiplexing strategy is to maximally block the unwanted channel by setting the polarizer (one port of the PBS) orthogonal to the SOP of the center frequency component of that channel, as described in Case 2 of Section 3.3. The evolution of the SOPs of the two channels during propagation in the PMD-PDL regime is shown in Fig. 7 . Note that the frequency components deviated from the center one will suffer different loss inside PDLE, but they are symmetric (with the assumption of modest PMD and PDL) with the center one thus its influence on the overall power reduction is small. However, in our experiment there is not such a presumption that this effect is omitted.
As shown in Fig. 8 , although the PDM signal suffers the largest impairment in each emulator, the overall OSNR penalty is less than the sum of the two separate penalties, and as the PDL value is larger, the PMD-induced penalty is more suppressed. This phenomenon comes from the convergence of the SOPs of the different frequency components in the PMD-degraded signal.
Figure 9 illustrates the convergence in the form of the measured output degree of polarization (DOP) of one channel versus the PDL value, i.e., the output DOP increases with PDL value when the launch angle is (within the convergence zone); as comparisons, the results of launching for one channel and launching for the other channel are presented, also note that for PDM signals in the / launching case the output SOP between the center frequency components of both channels remains orthogonal. Therefore, in the worst joint penalty case (launching), the SOP variation induced by the previous PMDE can be partially suppressed by the PDLE, resulting in mitigated pulse broadening effect inside each channel and reduced inter-channel crosstalk as well.
In the PDL-PMD regime, the PDM signal is at first launched into the PDLE with the SOP of the each channel relative to the least lossy PDL axis, and then launched into the PMDE with the SOP of relative to the PSPs of the PMDE. As the launch angle of is within [,], and suffers less PMD-induced pulse broadening than , therefore the PMD-induced crosstalk from to is smaller than that from to . Since the inter-channel crosstalk has a larger impact on BER than intra-channel pulse broadening, the OSNR penalty of is smaller than that of , as shown in Fig. 10 .
With the given experimental setup, the exact launch angle of is unknown, however, the worst case scenario (when the PMDE launch angle of is also ) can still be evaluated with the measured results. Note that the angle between the center frequency components of the two channels is determined by the PDLE and irrelevant to the PMDE; meanwhile, the polarization demultiplexing strategy for is to set the polarizer at port B perpendicular to in the Jones space, thus the crosstalk from to is invariant with regard to how is affected by the PMDE. Therefore, the penalty difference for between launching and launching only comes from the different pulse broadening impairment of . Based on Eq. (6), for a PDM signal launched into a 3.2dB PDLE, the output inter-channel angle γ is , thus equals . Note that a 30.92ps PMDE can induce 1.2dB intra-channel pulse broadening penalty if the channel of the PDM signal is launched relative to the PSPs of the PMDE. We assume that happens to be launched at , the difference of pulse broadening penalty between launching and launching can be expressed as 1.2dB [25], i.e., 0.54dB. Taking this extra 0~0.54dB penalty into account, the overall OSNR penalty will still not be larger than the sum of the two separate penalties, which is similar to the PMD-PDL regime.
In a real system, PDM and PDL are distributed along the link, which is much complicated to analyze. The simple model of PMD and PDL interaction used in this paper shows that: the sum of the crosstalk-induced penalties from PMD and PDL alone sets the upper limit of the overall crosstalk-induced penalty observed at the end of the link. We believe the conclusion also applies to the distributed model.
4. Conclusion
We have studied the PMD and PDL impairments in PDM signals. We showed that the time synchronized PDM signals can tolerate more PMD than the time interleaved ones, and that signals with larger bandwidth are more sensitive to PMD. We found that PDL impairments have little dependence on the time delay between the two polarizations and signal bandwidth. We showed that PDL-induced crosstalk could cause severe penalties in PDM signals, but with the proper configuration of polarization demultiplexing, the PDL-induced crosstalk can be completely eliminated. The effects of concatenation of PMD and PDL on the PDM signal were studied, and we found that the impairment caused by the combination of PMD and PDL does not enhance the impairments caused by PMD and PDL alone.
Acknowledgment
This work was done at Crawford Hill Laboratory, Alcatel-Lucent, USA. The authors are grateful to acknowledge valuable discussions with R. W. Tkach and R. -J. Essiambre. Zinan Wang acknowledges the support from China Scholarship Council, National High-tech Research and Development Program of China (2007AA03Z447) and National Basic Research Program of China (2003CB314900).
References and links
1. A. R. Chraplyvy, A. H. Gnauck, R. W. Tkach, J. L. Zyskind, J. W. Sulhoff, A. J. Lucero, Y. Sun, R. M. Jopson, F. Forghieri, R. M. Derosier, C. Wolf, and A. R. McCormick, “1 Tb/s transmission experiment,” IEEE Photon. Technol. Lett. 8(9), 1264–1266 (1996). [CrossRef]
2. A. H. Gnauck, G. Charlet, P. Tran, P. J. Winzer, C. R. Doerr, J. C. Centanni, E. C. Burrows, T. Kawanishi, T. Sakamoto, and K. Higuma, “25.6-Tb/s WDM transmission of polarization-multiplexed RZ-DQPSK signals,” J. Lightwave Technol. 26(1), 79–84 (2008). [CrossRef]
3. G. Charlet, J. Renaudier, M. Salsi, H. Mardoyan, P. Tran, and S. Bigo, “Efficient mitigation of fiber impairments in an ultra-long haul transmission of 40Gbit/s polarization-multiplexed data, by digital processing in a coherent receiver,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2007), paper PDP17. http://www.opticsinfobase.org/abstract.cfm?URI=OFC-2007-PDP17
4. H. Sunnerud, M. Karlsson, C. Xie, and P. A. Andrekson, “Polarization mode dispersion in high-speed fiber-optic transmission systems,” J. Lightwave Technol. 20(12), 2204–2219 (2002). [CrossRef]
5. A. Mecozzi and M. Shtaif, “Signal-to-noise-ratio degradation caused by polarization-dependent loss and the effect of dynamic gain equalization,” J. Lightwave Technol. 22(8), 1856–1871 (2004). [CrossRef]
6. M. Shtaif and A. Mecozzi, “Polarization-dependent loss and its effect on the signal-to-noise ratio in fiber-optic systems,” IEEE Photon. Technol. Lett. 16(2), 671–673 (2004). [CrossRef]
7. C. Xie and L. F. Mollenauer, “Performance degradation induced by polarization dependent loss in optical fiber transmission systems with and without polarization mode dispersion,” J. Lightwave Technol. 21(9), 1953–1957 (2003). [CrossRef]
8. O. Vassilieva, T. Hoshida, X. Wang, J. Rasmussen, H. Miyata, and T. Naito, “Impact of polarization dependent loss and cross-phase modulation on polarization multiplexed DQPSK Signals,”in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest (CD) (Optical Society of America, 2008), paper OThU6. http://www.opticsinfobase.org/abstract.cfm?URI=OFC-2008-OThU6
9. J. Renaudier, G. Charlet, M. Salsi, O. B. Pardo, H. Mardoyan, P. Tran, and S. Bigo, “Linear fiber impairments mitigation of 40-Gbit/s polarization-multiplexed QPSK by digital processing in a coherent receiver,” J. Lightwave Technol. 26(1), 36–42 (2008). [CrossRef]
10. L. E. Nelson, S. L. Woodward, S. Foo, X. Zhou, M. D. Feuer, D. Hanson, D. McGhan, H. Sun, M. Moyer, M. O. Sullivan, and P. D. Magill, “Performance of a 46-Gbps dual-polarization QPSK transceiver with real-time coherent equalization over high PMD fiber,” J. Lightwave Technol. 27(3), 158–167 (2009). [CrossRef]
11. C. Laperle, B. Villeneuve, Z. Zhang, D. McGhan, H. Sun, and M. O’Sullivan, “WDM performance and PMD tolerance of a coherent 40-Gbit/s dual-polarization QPSK transceiver,” J. Lightwave Technol. 26(1), 168–175 (2008). [CrossRef]
12. M. Shtaif, “Performance degradation in coherent polarization multiplexed systems as a result of polarization dependent loss,” Opt. Express 16(18), 13918–13932 (2008), http://www.opticsinfobase.org/abstract.cfm?URI=oe-16-18-13918. [CrossRef] [PubMed]
13. L. E. Nelson, T. N. Nielsen, and H. Kogelnik, “Observation of PMD induced coherent crosstalk in polarization-multiplexed transmission,” IEEE Photon. Technol. Lett. 13(7), 738–740 (2001). [CrossRef]
14. L. Nelson and H. Kogelnik, “Coherent crosstalk impairments in polarization multiplexed transmission due to polarization mode dispersion,” Opt. Express 7, 350–361 (2000), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-7-10-350. [CrossRef] [PubMed]
15. D. van den Borne, N. E. Hecker-Denschlag, G. D. Khoe, and H. de Waardt, “PMD-induced transmission penalties in polarization-multiplexed transmission,” J. Lightwave Technol. 23(12), 4004–4015 (2005). [CrossRef]
16. Z. Wang, and C. Xie, “PMD and PDL tolerance of polarization division multiplexed signals with direct detection,” in Proc. ECOC 2008, Brussels, Belguim, 2008, paper We.3.E.2.
17. S. Hinz, D. Sandel, F. Wuest, and R. Noé, “PMD tolerance of polarization division multiplex transmission using return-to-zero coding,” Opt. Express 9, 136–140 (2001), http://www.opticsinfobase.org/abstract.cfm?URI=oe-9-3-136. [CrossRef] [PubMed]
18. D. van den Borne, S. L. Jansen, E. Gottwald, P. M. Krummrich, G. D. Khoe, and H. de Waardt, “1.6-b/s/Hz spectrally efficient transmission over 1700 km of SSMF using 40×85.6-Gb/s POLMUX-RZ-DQPSK,” J. Lightwave Technol. 25(1), 222–232 (2007). [CrossRef]
19. H. C. Ji, J. H. Lee, H. Kim, P. K. 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), http://www.opticsinfobase.org/abstract.cfm?URI=oe-17-3-1169. [CrossRef] [PubMed]
20. M. Shtaif and O. Rosenberg, “Polarization-dependent loss as a waveform-distorting mechanism and its effect on fiber-optic systems,” J. Lightwave Technol. 23(2), 923–930 (2005). [CrossRef]
21. C. Xie, L. F. Mollenauer, and L. Möller, “Pulse distortion induced by polarization-mode dispersion and polarization-dependent loss in lightwave transmission systems,” IEEE Photon. Technol. Lett. 15(8), 1073–1075 (2003). [CrossRef]
22. L. Chen, Z. Zhang, and X. Bao, “Combined PMD-PDL effects on BERs in simplified optical systems: an analytical approach,” Opt. Express 15(5), 2106–2119 (2007), http://www.opticsinfobase.org/abstract.cfm?URI=oe-15-5-2106. [CrossRef] [PubMed]
23. C. R. Menyuk, D. Wang, and A. N. Pilipetskii, “Repolarization of polarization-scrambled optical signals due to polarization dependent loss,” IEEE Photon. Technol. Lett. 9(9), 1247–1249 (1997). [CrossRef]
24. C. Xie, L. Möller, H. Haunstein, and S. Hunsche, “Comparison of system tolerance to polarization-mode dispersion between different modulation formats,” IEEE Photon. Technol. Lett. 15(8), 1168–1170 (2003). [CrossRef]
25. H. Kogelnik, R. M. Jopson, and L. E. Nelson, Optical Fiber Telecommunications IV-B (Academic, 2002), Chap.15.