Open Access
Add to BibSonomyAbstract
We propose and demonstrate a novel first-order polarization mode dispersion (PMD) compensator separating principal-states-of-polarization (PSP) control from differential-group-delay (DGD) control by using two independent feedback monitoring signals. To verify the proposed operating principle, we fabricated an automatically adaptive module-type 40-Gb/s PMD compensator on printed circuit boards. The results show that unlike previous typical compensation method of controlling alternately PSP and DGD, the proposed scheme provides independent continuous PSP tracking regardless of DGD control by delay line under rapidly varying PSP condition, thus resulting in stable compensated signal with reduction in compensation time.
©2012 Optical Society of America
1. Introduction
With growing demand for larger capacity in backbone fiber-optic networks, bit rate per a wavelength channel has recently been evolving from 10 Gb/s to 40 Gb/s or higher [1,2]. In such high-speed optical transmission system, polarization mode dispersion (PMD) is becoming a main factor limiting its performance [3–12]. PMD means pulse distortion due to differential group delay (DGD) between two mutually orthogonal principal states of polarization (PSP) occurring in a birefringent fiber-optic link, which is induced mainly by environment change, birefringent optical components, and asymmetric fiber core. Under a same DGD, higher bit-rate system suffers more impairment due to smaller signal pulsewidth. Therefore, PMD compensation is essential for transmission systems at 40 Gb/s or beyond.
Various methods for PMD compensation have been extensively studied and developed. As shown in Fig. 1(a) , the existing conventional methods use a single feedback signal such as degree of polarization (DOP) or half-clock RF signal for monitoring degree of signal distortion, which is extracted at the output of differential delay line (DDL) that provides inverse birefringence to cancel out the link DGD [13–18]. This single feedback signal compels the PMD compensator to alternately control a polarization controller (PC) and a delay line. In other words, a PC for aligning PSP to PBS axes and a delay line for removing DGD are operated by turns. The problem is that because of this alternate control the compensator cannot adjust the PC while controlling the relatively slow delay line. The PSP may rapidly change during the DGD control and thus the compensator may miss vestige of the PSP, causing misalignment of PSP and PBS axes. Accordingly, the birefringence of the compensator adds undesirably to the link DGD, resulting in the performance worsening than that without the compensator. The measured maximum speed of PSP in installed field fiber has been reported to be on the order of 10 ms [19–21].
Fig. 1 Schematic diagrams of (a) previous typical and (b) proposed first-order PMD compensators. PC: Polarization controller, PBS: Polarization beam splitter, RFD: RF power detector. DDL: Differential delay line.
To solve this problem, we propose a novel compensator using two independent feedback signals. One is used only for controlling a PC to align PSP to PBS, and the other for adjusting a delay line to offset DGD. Then, the two PSP and DGD controls can be separated and operated independently of each other due to use of the two separate feedback signals, thus preventing PSP tracking loss regardless of a slow delay line even though PSP changes rapidly and considerably during the delay line motion. To experimentally demonstrate the proposed scheme, we fabricated a first-order 40-Gb/s PMD compensator in module type on printed circuit boards. The compensator showed a very fast response time of less than 2 μs for PSP tracking and, regardless of relatively slow delay line control, maintained stable compensated output signal under randomly fast varying PSP condition.
2. Operating principle
Figure 1(b) shows a simplified structure of the proposed first-order PMD compensator. An optical signal transmitted through a PMD-impaired fiber link becomes distorted and has DGD between its two mutually orthogonal fast and slow PSPs. To compensate for PMD means to align the PSPs to PBS axes with a PC and remove the DGD with a delay line. For this purpose and to solve the problem of the previous method described above, the suggested scheme uses two feedback signals. One is RF feedback signal bandpass-filtered at clock frequency from optical-to-electrical converted signal via photodetector (PD) that detects light signal tapped at one path inside DDL to control only a PC for alignment of PSP to PBS. The other is DOP or half-clock RF signal extracted at output of DDL to control only a delay line for offsetting DGD. Therefore, the control and alignment of PSP to PBS can be operated independently of the control and elimination of DGD. The PSP and PBS alignment, implying separation of fast and slow PSPs via PBS, is accomplished by minimizing the RF feedback signal, reported in our earlier work [22], whereas the DGD elimination is done by maximizing the DOP. The two PSP and DGD tracking processes use conventional dithering algorithm of comparing current and previous detected values. The RF feedback signal for PSP separation does not suffer effect of DDL birefringence since it is extracted at one path inside DDL, whereas DOP extracted behind DDL may be affected by birefringence of DDL when PSP and PBS are misaligned by PSP tracking loss. A quarter-wave plate (QWP) and a half-wave plate (HWP) composing the PC are used for aligning PSP to PBS. The QWP transforms arbitrary elliptical two PSPs on Poincare sphere into linear PSPs located on the equator of the sphere, which are then rotated 180-degree by the HWP to be aligned to the eigen states of DDL. The proposed compensator first starts scan process. The PC is scanned to find an angle state of two wave plates of minimum RF feedback power, to which the PC state is moved and real-time PSP tracking begins. Then, the delay line for DGD tracking is operated in a way to maximize DOP while maintaining alignment of PSP and PBS via the real-time PSP tracking. The advantage of the proposed scheme is that the compensator can control the PC simultaneously while moving the delay line by separating the two controls by using the two independent feedback signals. Therefore, alignment of PSP to PBS can be maintained constantly during and regardless of delay line control, thus preventing PSP tracking loss, removing injurious effect of DDL birefringence, and allowing stable continuous compensation. In addition, compensation time is considerably reduced. When equal step size for control is given, scanning time of the previous scheme is decided by number of step of PC control multiplied by that of delay line control, whereas scan time of the proposed approach is determined by merely sum of step numbers of the two controls.
Figure 2(a) shows simulated RF power from one PBS axis, bandpass-filtered at various frequencies, as a function of angle θ between output PSP and PBS when given 2-GHz BPF bandwidth, 20-ps DGD and 30-degree launch angle to input PSP for 40-Gb/s non-return-to-zero (NRZ) modulation format. The simulation was carried out using the model described in [22]. At θ=nπ/2 (n: integer), two orthogonal PSPs are completely aligned to PBS axes and separated into two fast and slow PSPs via PBS. As shown in Fig. 2(a), only the RF signal filtered at 40 GHz shows convergence to minimum at θ=nπ/2, which can be reached by adjusting the PC in a way to minimize the RF feedback signal by dithering process. At other frequencies, no minimum-converging points can be found at θ=nπ/2. Therefore, RF signal filtered at clock frequency can be used as a feedback signal for separation of two fast and slow PSPs via PBS. Figure 2(b) shows RF power filtered at 40 GHz with 2-GHz bandwidth as a function of the angle θ for various DGDs. Regardless of DGD value, the RF power filtered at clock frequency shows minimum when PSP are aligned with PBS. The largest contrast or dynamic range in RF feedback power is obtained at a 12.5-ps DGD from which almost symmetric contrast with DGD is observed as shown in Fig. 2(b). Figure 2(c) shows simulated DOP at the DDL output in regard to DGD for various launch angles to input PSP after completion of PSP and PBS alignment. This DGD means residual DGD during compensation, which is sum of fiber DGD and differential delay time, and can be reduced to zero by adjusting DDL in a direction against birefringence of fiber link. The graphs show that DOP decreases with increasing DGD and the steepest inclination is obtained for a 45-degree launch angle. Convergence to DOP maximum means complete DGD canceling, which can be attained by adjusting the delay line in a way to maximize the DOP by dithering algorithm. If a return-to-zero (RZ) signal is used, the center of BPF for feedback signal should be twice the clock frequency due to a half times smaller pulsewidth than NRZ signal but the control algorithm is the same as the case of NRZ signal. We assume fully compensated chromatic dispersion (CD). If residual CD is large, the RF feedback signal may be affected by RF components induced by CD.
Fig. 2 Simulated results. (a) Filtered RF power spectra versus angle between PSP and PBS axes for given 2-GHz filter bandwidth, 20-ps DGD, and 30-degree launch angle for 40-Gb/s NRZ signal. (b) RF power filtered at 40 GHz versus angle between PSP and PBS axes for various DGD values. (c) DOP of DDL output signal versus DGD for various launch angles while keeping PSP parallel to PBS axes.
3. Experiment and results
Figure 3 shows configuration of the proposed PMD compensator fabricated in module type on printed circuit boards and an experimental setup to evaluate its performance. The compensator consists of LiNbO3 (LN) PC, DDL, and two separated PSP and DGD control parts. The LN PC of less than 100-ns response time is composed of a quarter-wave plate (QWP) and a half-wave plate(HWP) that are rotated by a driver with four linear amplifiers of over 1-MHz bandwidth. The driver is controlled by a PSP controller in a way to minimize the RF feedback signal filtered at clock frequency through iterative comparison of current and previous detected values for PSP and PBS alignment. The RF power detector (RFD) used converts RF power into DC voltage with a sensitivity of 1 mV/μW and a detection frequency range of 33 to 50 GHz. The electrical bandpass filter (EBPF) of waveguide type has a bandwidth of 2 GHz and a center frequency of 40 GHz. The gain of the RF amplifier behind 40-Gb/s photodetector is 30 dB over a 37 to 45 GHz range. The tapping rate from one arm of DDL for PSP control was set at 20% and the losses of two arms of DDL were balanced by a variable optical attenuator (VOA). The motorized delay line provides differential delay time between two eigen states of DDL from −45 ps to 45 ps with insertion loss variation of less than 0.35 dB over entire travel range. The PBS used has an extinction ratio of over 30 dB. The optical bandpass filter (OBPF) based on fiber Bragg grating (FBG) with a 0.25-nm bandwidth provides larger DOP dynamic range and higher resolution to tracking DGD. The DOP meter measures Stokes parameters to determine DOP with a 700-kHz bandwidth. The PSP and DGD controllers comprise a digital signal processor (DSP) of 160-MHz clock and analog-to-digital (A/D) and digital-to-analog (D/A) converters of 50-MHz sampling rate and 12-bit resolution. The compensator begins first the scanning of angles of two wave plates to find and move to a point of minimum RF feedback power where PSP and PBS are aligned and real-time PSP tracking starts. Then, DGD tracking starts to find maximum DOP by adjusting the delay line while keeping PSP parallel to PBS axes by the PSP tracking. The step angle of the two wave plates was tunable and set at 1 degree for fast PSP scan and at 0.5 degree for precise PSP tracking. At each step angle, the microprocessor reads the feedback power 32 times and takes an average of them for high accuracy. The step delay of the delay line can also be varied and was set to be 0.3 ps for DGD tracking.
Fig. 3 Schematics of (a) fabricated PMD compensator and experimental setup for evaluation and (b) BER measurement. LD: Laser diode, IM: Intensity modulator, EMUX: Electrical multiplexing, PPG: Pulse pattern generator, LN PC: LiNbO3 polarization controller, EBPF: Electrical bandpass filter, OBPF: Optical bandpass filter, BA: Booster amplifier, PA: Preamplifier, CDR: Clock and data recovery, DEMUX: Demultiplexer.
To evaluate the performance of the manufactured compensator module, we constructed an experimental setup as shown in Fig. 3. The four channel 10-Gb/s signals from pulse pattern generator (PPG) were electrically multiplexed to produce a 40-Gb/s NRZ signal with pseudo-random bit sequence (PRBS) of 231-1 pattern length before driving a Mach-Zehnder intensity modulator (IM) to modulate a light at 1552.5 nm from DFB LD to generate a 40-Gb/s optical NRZ signal. The polarization controller PC1 was adjusted to set polarization of the modulator output at 45 degrees to eigen states of a first-order PMD emulator to give equal optical power to the two eigen states that act as two PSPs of fiber link with PMD. The emulator (JDSU PE4) produces DGD between its two PSPs. The polarization controller PC2 was used to change the PSP of the output signal of the emulator. Figure 3(b) shows an experimental setup to measure bit error rate (BER) before and after PMD compensation. To obtain BER, a distorted optical signal by DGD of the emulator is varied in power by an attenuator before passing through a two-stage preamplifier (PA) and an optical filter of 0.5-nm bandwidth and the PMD compensator. The optical-to-electrical converted signal via 40-Gb/s photodetector in Fig. 3(b) was sent to a clock and data recovery (CDR) circuit to yield 40-Gb/s clock and data which entered a 1:4 electrical demultiplexer (DEMUX) to produce 10-Gb/s clock and data that were inputted to an error detector for BER measurement. For various DGD values, the BER was measured with variation of the input power to the PA before and after PMD compensation.
Figure 4(a) shows oscilloscope view of temporal variation of RF feedback signal measured by RF detector during scan and real-time tracking for a 25-ps DGD. The scan time was measured to be 50 ms, which can be varied and become larger with smaller step angle and larger average number. Following scan, the real-time tracking process was shown to begin where minimum feedback RF power was obtained. Figure 4(b) shows a measured response time of PSP control which is defined to be time taken from point of applying control voltage to LN PC until when RF detector detects the filtered feedback signal. The result showed a very fast response time of less than 2 μs, which is expected to allow tracking of rapid PSP variation up to submillisecond in fiber link.
Fig. 4 (a) RF feedback signal measured timewise by digital oscilloscope during scan and tracking processes when a filter of 2-GHz bandwidth and 40-GHz center frequency is used. (b) Time variation of RF feedback signal when applying a control voltage to LN PC to find a response time of PSP control.
Figures 5(a) and 5(b) show RF feedback signal detected while scanning two wave plates up to 180 degrees for 10-ps and 20-ps DGDs, respectively. The bottom in the graphs shows the contour of the RF power. The results show that regardless of DGD values, the convergent minimum point occurs two times and four times during 180-degree rotation of QWP and HWP, respectively. The two PSPs were confirmed to be aligned to PBS at those minimums and real-time tracking starts at one of them.
Fig. 5 3-dim and 2-dim view of measured RF feedback signal as a function of angles of half-wave and quarter-wave plates for given (a) 10-ps and (b) 20-ps DGDs, respectively.
Figure 6 shows direct real-time PSP tracking without scan process and convergence to minimum of RF feedback signal when given a worst case of 45-degree launch angle to input PSP (γ=0.5) and 25-ps DGD. Finding and moving to a minimum was accomplished by dithering process to minimize the feedback signal via comparison of current and previous detected values. The time taken from the worst distortion to PSP separation was measured to be approximately 1 ms, which implies that the time taken from any distorted state to PSP separation can be kept below 1 ms. This adaptation time to minimum depends on size of step angle of the waveplate and number of feedback signal data read by DSP for average. The bottom left and right insets show eye diagrams of the distorted signal by 25-ps DGD and the compensated signal after PSP separation via PBS, respectively. The EBPF of 2-GHz bandwidth was found to show best performance compared to that of other 1-GHz and 4-GHz bandwidth because the larger bandwidth showed larger feedback dynamic range but worse alignment accuracy of PSP and PBS.
Fig. 6 Oscilloscope view of RF feedback signal showing convergence to minimum by real-time PSP tracking started from when given a worst distortion of 45-degree launch angle and 25-ps DGD. The insets show eye diagrams of PMD-impaired signal and compensated signal.
Figure 7 shows DOP measured before the compensator with DGD varied up to 25 ps by using a polarization analyzer with or without a FBG optical bandpass filter (OBPF) before the compensator. The reason why, to obtain Fig. 7, the OBPF in Fig. 3(a) was placed before the compensator and then DOP was measured before the compensator is that in configuration as Fig. 3(a), the total DGD after the compensator is sum of DGD by the emulator and differential delay time by the delay line of the compensator and thus each DOP after the compensator for each fixed total DGD value cannot be obtained since the delay line moves automatically and adaptively. The measurement was done to find an optimum filter (OBPF) condition to enhance DOP dynamic range. The three DOP curves showed decrease with DGD increase regardless of the filter use. This is due to the fact that different wavelength within an optical signal band produces, via interaction with DGD, different polarization state at output of birefringent fiber link, represented as polarization diffusion on Poincare sphere. The extent of the polarization diffusion intensifies as DGD increases, thus worsening DOP. In the case of using the sideband filter whose center is 0.2 nm off that of the optical signal, the DOP showed larger dynamic range than that without filter because the center components in the signal band become weakened and comparable to its side components through the filtering due to the filter shape similar to Gaussian type. The larger DOP dynamic range provides higher resolution and accuracy in one step move in DGD tracking. On the other hand, the filter whose center wavelength is identical to that of the optical signal showed rather smaller DOP dynamic range than the filter-less case because via the filtering the side components in the signal band become more weakened than the center components due to the filter shape so that extent of depolarization is mitigated and hence DOP is increased more than the case without filter under the same DGD.
Fig. 7 Measured DOP versus DGD with or without optical filter. The sideband filtering provides the largest dynamic range and the highest resolution to DGD tracking.
Figure 8(a) shows time variation of PSP (s1, normalized Stokes parameter) and DOP before compensation and DOP after compensation under 25-ps DGD condition. To simulate fast PSP change in fiber link, the motorized PC2 (HP, scan rate 8) behind the PMD emulator was used to provide its fastest speed of random polarization fluctuation. We used the polarization analyzer (HP) to measure states of polarization (SOP) and DOP with its highest 2-ms sampling rate. The normalized Stokes parameter s1 in Fig. 8 shows how fast PSP changes and its maximum speed was measured to be about 0.1/ms. The DOP before compensation was measured to be an average of 0.66 for 25-ps DGD with slight fluctuation. Regardless of the fast PSP variation, the measured DOP after compensation in Fig. 8(a) was maintained within a range of 0.980 to 0.997 with compensated open-eye signal, showing that the speed of the compensator with less than 2 μs response time in PSP control is fast enough to track and keep the PSP fluctuating rapidly at 0.1/ms aligned constantly to PBS axes in real time. Figure 8(b) shows measured bit error rate (BER) before and after PMD compensation with the input power to the preamplifier (PA) varied by an attenuator under various DGD values. The results show that for 12.5-ps DGD, the PMD compensation yielded a 7.2 dB power gain at a 10−11 BER and a 3.4 dB Q-value gain at an input power of −24.5 dBm where optical signal-to-noise ratio (OSNR) after the PA was measured to be 22.6 dB in 0.2-nm resolution.
Fig. 8 (a) Temporal variation of PSP and DOP measured before and after PMD compensator. (b) Measured BER before and after PMD compensation.
4. Conclusion
In summary, we proposed a novel scheme for first-order PMD compensation that allows separation of PSP control from DGD control by using two independent feedback monitoring signals. We also experimentally demonstrated its operating principle by using an automatically adaptive 40-Gb/s first-order PMD compensator fabricated in module type on printed circuit boards. The results showed that the independent PC control for adjusting PSP from delay line control for tuning DGD was achieved by using both RF power bandpass-filtered at clock frequency and DOP as two separate feedback signals, thereby allowing independent continuous PSP tracking irrespective of DGD tracking and leading to stable compensated output signal with compensation time reduction. The fabricated compensator was found to have a response time of below 2 μs in PSP control and to successfully track PSP fluctuating rapidly in random way at a 0.1/ms speed in normalized Stokes parameter. The previous conventional compensation method shows limitation in tracking fast-changing PSP in fiber link due to the alternate control of PSP and DGD forced by a single feedback signal and thus its performance becomes worse as speed of delay line is slower. The proposed scheme, however, is expected to be able to track PSP varying rapidly on the order of microseconds regardless of the delay line speed due to separation of the two PSP and DGD controls enabled by the two independent feedback monitoring signals. To enhance speed of delay line, another type of delay line such as multi birefringent elements with polarization controllers inbetween can be used at one path of DDL if its differential delay time is guaranteed to vary continuously with monotone increasing or decreasing between adjacent steps of the delay line. The proposed scheme has disadvantage in terms of complexity and cost. If large higher-order PMD exists in fiber link, it may affect the tracking performance of this compensator. To mitigate this effect, another a few birefringent elements such as PMF and polarization controllers can be used in front of the proposed first-order compensator.
References and links
1. M. Chacinski, U. Westergren, L. Thylen, B. Stoltz, J. Rosenzweig, R. Driad, R. E. Makon, J. Li, and A. Steffan, “ETDM transmitter module for 100-Gb/s ethernet,” IEEE Photon. Technol. Lett. 22(2), 70–72 (2010). [CrossRef]
2. G. Raybon, P. J. Winzer, A. A. Adamiecki, A. H. Gnauck, A. Konczykowska, F. Jorge, J. Dupuy, L. L. Buhl, C. R. Doerr, R. Delbue, and P. J. Pupalaikis, “All-ETDM 80-Gbaud (160-Gb/s) QPSK generation and coherent detection,” IEEE Photon. Technol. Lett. 23(22), 1667–1669 (2011). [CrossRef]
3. C. D. Poole and R. E. Wagner, “Phenomenological approach to polarisation dispersion in long single-mode fibres,” Electron. Lett. 22(19), 1029–1030 (1986). [CrossRef]
4. C. D. Poole, R. W. Tkach, A. R. Chraplyvy, and D. A. Fishman, “Fading in lightwave systems due to polarization-mode dispersion,” IEEE Photon. Technol. Lett. 3(1), 68–70 (1991). [CrossRef]
5. H. Bülow, “System outage probability due to first- and second-order PMD,” IEEE Photon. Technol. Lett. 10(5), 696–698 (1998). [CrossRef]
6. H. Kogelnik, P. J. Winzer, L. E. Nelson, R. M. Jopson, M. Boroditsky, and M. Brodsky, “First-order PMD outage for the hinge model,” IEEE Photon. Technol. Lett. 17(6), 1208–1210 (2005). [CrossRef]
7. 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]
8. N. Gisin, R. Passy, J. C. Bishoff, and B. Perny, “Experimental investigations of the statistical properties of polarization mode dispersion in single mode fibers,” IEEE Photon. Technol. Lett. 5(7), 819–821 (1993). [CrossRef]
9. L. E. Nelson, R. M. Jopson, H. Kogelnik, and G. J. Foschini, “Measurement of depolarization and scaling associated with second-order polarization mode dispersion in optical fibers,” IEEE Photon. Technol. Lett. 11(12), 1614–1616 (1999). [CrossRef]
10. J. M. Gené and P. J. Winzer, “First-order PMD outage prediction based on outage maps,” J. Lightwave Technol. 28(13), 1873–1881 (2010). [CrossRef]
11. M. Karlsson, “Polarization mode dispersion induced pulse broadening in optical fibers,” Opt. Lett. 23(9), 688–690 (1998). [CrossRef] [PubMed]
12. B. W. Hakki, “Polarization mode dispersion compensation by phase diversity detection,” IEEE Photon. Technol. Lett. 9(1), 121–123 (1997). [CrossRef]
13. F. Heismann, D. A. Fishman, and D. L. Wilson, “Automatic compensation of first order polarization mode dispersion,” in Proc. European Conference on Optical Communication, ECOC98, 529–530 (1998).
14. R. Noe, D. Sandel, M. Yoshida-Dierolf, S. Hinz, V. Mirvoda, A. Schopflin, C. Gungener, E. Gottwald, C. Scheerer, G. Fischer, T. Weyrauch, and W. Haase, “Polarization mode dispersion compensation at 10, 20, and 40 Gb/s with various optical equalizers,” J. Lightwave Technol. 17(9), 1602–1616 (1999). [CrossRef]
15. H. Sunnerud, M. Chongjin Xie, M. Karlsson, R. Samuelsson, and P. A. Andrekson, “A comparison between different PMD compensation techniques,” J. Lightwave Technol. 20(3), 368–378 (2002). [CrossRef]
16. N. Y. Kim, D. Lee, J. Park, and N. Park, “Comparisons on PMD-compensation feedback methods for bandwidth-rich transmission formats,” IEEE Photon. Technol. Lett. 16(6), 1597–1599 (2004). [CrossRef]
17. L. Xu, H. Miao, and A. M. Weiner, “All-order polarization-mode-dispersion (PMD) compensation at 40 Gb/s via hyperfine resolution optical pulse shaping,” IEEE Photon. Technol. Lett. 22(15), 1078–1080 (2010). [CrossRef]
18. X. S. Yao, X. Chen, T. J. Xia, G. Wellbrock, D. Chen, D. Peterson, P. Zhang, A. Belisle, L. Dong, and T. Yu, “In-service light path PMD (polarization mode dispersion) monitoring by PMD compensation,” Opt. Express 18(26), 27306–27318 (2010). [CrossRef] [PubMed]
19. H. Bülow, W. Baumert, H. Schmuck, F. Mohr, T. Schulz, F. Kuppers, and W. Weiershausen, “Measurement of the maximum speed of PMD fluctuation in installed field fiber,” in Proc. Optical Fiber Communication Conference, OFC 99, 83–85 (1999).
20. D. S. Waddy, L. Chen, and X. Bao, “Polarization effects in aerial fibers,” Opt. Fiber Technol. 11(1), 1–19 (2005). [CrossRef]
21. K. Ogaki, M. Nakada, Y. Nagao, and K. Nishijima, “Fluctuation differences in the principal states of polarization in aerial and buried cables,” in Proc. Optical Fiber Communications Conference, OFC 2003, 14–15 (2003).
22. K. H. Han and W. J. Lee, “Tracking and separation of time-varying principal states of polarization in optical fiber link for first-order PMD compensation and its filter-dependent performance,” Opt. Fiber Technol. 14(4), 268–274 (2008). [CrossRef]
