Open Access
Add to BibSonomyAbstract
We experimentally investigate 28-GBd (84-Gb/s) and 37.3-GBd (112-Gb/s) polarization-switched quadrature phase-shift keying (PS-QPSK) signals. In single-channel transmission experiments over up to 12500 km ultra large effective area fiber, we compare their performance to that of polarization-division multiplexing quadrature phase-shift keying (PDM-QPSK) signals at the same bit rates. The experimental results show that PS-QPSK not only benefits from its better sensitivity but also offers an increased tolerance to intra-channel nonlinearities.
©2011 Optical Society of America
1. Introduction
Recently, Agrell and Karlsson [1,2] as well as Bülow [3] explored the optimal placement of signal constellation points in the four-dimensional (4-D) space spanned by the electric field in optical single-mode fibers. The four dimensions used for modulation are the inphase (I) and quadrature (Q) components in two orthogonal polarizations. By numerically optimizing the packing of spheres in 4-D space, the most power efficient 4-D modulation format was identified in [1]. It turned out to be a combination of 2-ary polarization-shift keying and quadrature phase-shift keying (QPSK) in each polarization and was named polarization-switched (PS) QPSK. Theoretically, it offers approximately a 1-dB improvement in sensitivity over polarization-division multiplexing (PDM) QPSK compared at a bit-error ratio (BER) of 10−3 for the same bit rate [2].
Recent numerical simulations indicate that PS-QPSK not only shows an improved sensitivity but also a better resilience against nonlinear impairments in wavelength-division multiplexing transmission systems. This was investigated by numerical simulations for symbol rates of 28 GBd and 37 GBd over 1800-km standard single-mode fiber (SSMF) as well as non-zero dispersion-shifted fiber links in [4] and for a symbol rate of 37 GBd over a 2000-km SSMF link in [5].
The back-to-back sensitivity improvement was recently demonstrated experimentally for 10-GBd (30-Gb/s) [6] as well as for 14.3-GBd (42.9-Gb/s) [7,8] and 13.5-Gbd (40.5-Gb/s) [9]. In all of these works, the employed PS-QPSK transmitter consists of two serial stages. First, a QPSK signal is generated by using an I/Q-modulator. A second stage consisting of a 3-dB coupler, two synchronized Mach-Zehnder modulators and a polarization-beam combiner switches the polarization state. The reported results of transmission experiments in [6]- [9] show either a similar or even improved nonlinear tolerance of PS-QPSK compared to PDM-QPSK, depending on symbol rate and link configuration.
In this contribution, we report on the generation of 28-GBd (84-Gb/s) and 37.3-GBd (112-Gb/s) PS-QPSK signals by using a single integrated dual-polarization (DP) I/Q-modulator. The performance of the generated signals is evaluated in back-to-back and single-channel transmission experiments over up to 12500 km ultra large effective area fiber (ULAF) and compared to that of PDM-QPSK signals at the same bit rates. This is an extension of our previously published work in [10].
2. Experimental setup
The experimental setup is shown in Fig. 1(a) . In the transmitter, an external cavity laser (ECL) with a linewidth of ~100 kHz is used as a light source. The transmitted wavelength is 1550.116 nm for 84-Gb/s and 1550.918 nm for 112-Gb/s signals. The ECL is followed by an integrated DP I/Q-modulator (Fujitsu FTM7977HQ) consisting of two I/Q-modulators, one for each polarization. The four driving signals are generated by a four-channel bit-pattern generator (BPG). For 84-Gb/s PS-QPSK the three bit patterns IX, QX and IY are de Bruijn binary sequences of length 215. The bit pattern QY for the quadrature component of the y-polarization is programmed to be an XOR-combination of the other three bit patterns as described in [1]. However, at the symbol rate of 37.3 GBd required for a bit rate of 112-Gb/s, our BPG (SHF12103A) does not allow for user-defined patterns but can output four independent pseudo-random binary sequences (PRBS) at a bit rate of 37.3 Gb/s (after custom calibration by the manufacturer). We therefore utilized a special property of PRBSs toemulate the XOR operation required for the generation of PS-QPSK signals. As pointed out in [11], d1⊕d2 = d3 when d1, d2 and d3 are circularly shifted PRBSs with an appropriate cyclic shift. This allowed us to use four appropriately delayed PRBSs of length 215-1 for driving the DP I/Q-modulator. The constellation diagrams of the generated 84-Gb/s PS-QPSK signals are shown in Fig. 1(c) as well as the eight resulting signal states per transmitted symbol as a plot of the φx / φy phase plane in Fig. 1(e). For comparison, Fig. 1(b) and (d) show constellation diagrams and phase plane with sixteen symbol states for 84-Gb/s PDM-QPSK signals.
Fig. 1 (a) Experimental setup. The insets show optical eye diagrams of 112-Gb/s PS-QPSK signals before and after the 50-GHz interleaving filter. The acronyms stand for ECL: external cavity laser, BPG: bit-pattern generator, EDFA: erbium-doped fiber amplifier, ILV: interleaver, AO: acousto-optic, EQ: equalizer, VOA: variable optical attenuator, LO: local oscillator, BD: balanced detector. (b), (c), (d), (e) Constellation diagrams as well as plots of the φx / φy phase plane at maximum OSNR for back-to-back 84-Gb/s PDM-QPSK and PS-QPSK signals, respectively. (f) Back-to-back constellation diagram after the 2 × 2 MIMO equalizer adapted with a modified CMA and a decision-directed least mean square algorithm for 112-Gb/s PS-QPSK at maximum OSNR.
After amplification by an erbium-doped fiber amplifier (EDFA), the signal is passed through a 50-GHz interleaving filter (ILV). The insets of Fig. 1(a) show the impact of this narrow-band filtering on the optical eye diagram of a 112-Gb/s PS-QPSK signal. After the ILV some inter-symbol interference is observed at the symbol center. The signal is then launched into a recirculating fiber loop that incorporates three spans of ULAF (kindlyprovided by OFS). The fibers have an effective area of 126 µm2, a dispersion parameter of 20 ps/(nm⋅km) and a fiber loss of 0.184 dB/km at 1550 nm. The span and loop losses are compensated by EDFAs only. The EDFAs are followed by 2-nm band-pass filters to avoid out-of-band noise accumulation. For the 112-Gb/s experiments we added a loop-synchronous polarization scrambler to avoid any potential build-up of polarization dependent impairments. Furthermore, a programmable gain-equalizing filter was used to compensate for the gain tilt of the inline EDFAs. These two components were not available for the 84-Gb/s experiments.
At the receiver, a noise loading stage consisting of a variable optical attenuator (VOA) and an EDFA allows for artificial degradation of the optical signal-to-noise ratio (OSNR). The signal is filtered by a 0.45-nm optical band-pass filter before being fed to a polarization-diversity optical 90°-hybrid. The local oscillator (LO) laser is an ECL with ~100 kHz linewidth and is tuned to match the signal laser wavelength. After detection by four balanced photo-detectors (BD) the signals are digitized by a real-time oscilloscope with a sampling rate of 50 GSa/s and a bandwidth of 20 GHz. The digital signal processing is performed offline.
The signal processing first compensates any imbalance of the optical frontend by using a Gram-Schmidt orthogonalization procedure [12]. After chromatic dispersion (CD) compensation, the residual frequency offset between signal and LO laser is estimated and removed [13]. An adaptive 2 × 2 MIMO equalizer separates the signal polarizations and compensates for residual CD, polarization-mode dispersion and LO phase noise. The filter taps are initialized by using a modified constant modulus algorithm (CMA) described in [14]. After preconvergence is obtained, the equalizer is switched to decision-directed mode. The feedback criterion is based on the minimum Euclidean distance in the 4-D signal space. A constellation diagram after the equalizer is shown for 112-Gb/s PS-QPSK in Fig. 1(f). Finally, the bits are decided based on minimum Euclidean distance and errors are counted.
3. Results and discussion
The measured BER in a back-to-back configuration is shown in Fig. 2 . The solid lines without symbols denote the theoretical limit of the BER performance for an additive white Gaussian noise channel according to [2]. The symbols are measured results. For 84-Gb/s PS-QPSK signals [blue circles in Fig. 2(a)] we measured a required OSNR of ~11.6 dB at a BER of 10−3 which is less than 0.6 dB away from the theoretical limit. The measured required OSNR for 84-Gb/s PDM-QPSK signals [red squares in Fig. 2(a)] at the same BER is ~12.6 dB, which means about 0.6 dB implementation penalty compared to the theoretical limit and a ~1-dB penalty compared to PS-QPSK as predicted theoretically (0.97 dB) in [2].
Fig. 2 Back-to-back BER for a bit rate of (a) 84 Gb/s and (b) 112 Gb/s. Solid lines correspond to the theoretical noise-limited BER and symbols denote measured BER values.
For 112-Gb/s PS-QPSK signals [circles in Fig. 2(b)] we measured a required OSNR of ~13.3 dB at a BER of 10−3 which is approximately 1 dB away from the theoretical limit. We attribute this higher implementation penalty to bandwidth constraints of our available hardware (e.g. the 20-GHz analog-to-digital converters). Finally, we also measured the BER for 112-Gb/s PDM-QPSK signals, resulting in a required OSNR of 13.9 dB. Therefore, at 112 Gb/s we measured a sensitivity improvement of only 0.6 dB. However, the advantage of PS-QPSK in required OSNR at a symbol rate of 28-GBd could make it an attractive solution for impairment-aware adaptation of the bit rate per wavelength channel since switching between the two formats at constant symbol rate just requires a modified bit-to-symbol pre- and decoding and minor adaptation of the signal processing as proposed in [4,15].
In the transmission experiments we first optimized the launch power into the fiber link. The launch power was varied between −3 dBm and + 3 dBm in steps of 2 dB. It was found that the optimum launch power resulting in maximum reach was −1 dBm for PDM-QPSK and + 1 dBm for PS-QPSK. Figure 3(a) shows the BER after transmission over the ULAF link with optimum launch power for 84-Gb/s PS-QPSK and PDM-QPSK, respectively. Results for other launch powers are reported in [10]. At + 1 dBm launch power, the PS-QPSK signal achieves a BER below 10−3 for a link length of 12500 km. The maximum reach measured for the PDM-QPSK signal is 9500 km. This corresponds to about 31% increase in achievable reach, indicating that PS-QPSK signals not only provide a better sensitivity but also a better tolerance against intra-channel nonlinear effects compared to PDM-QPSK signals at 84 Gb/s.
Fig. 3 (a) BER as a function of transmitted distance for a bit rate of 84 Gb/s at a launch power of −1 dBm for PDM-QPSK and + 1 dBm for PS-QPSK. (b) BER as a function of launch power for a bit rate of 112 Gb/s at different transmitted distances.
Figure 3(b) shows the results for transmission of the 112-Gb/s signals. For each transmission distance (6000 km, 9000 km and 12000 km) the launch power is varied between −4 dBm and + 4 dBm. For launch power ≤ −2 dBm the BER is limited by the delivered OSNR at the receiver. In this linear transmission regime, PS-QPSK benefits from its enhanced sensitivity only for low BER (i.e. at short distance). On the contrary, there is no difference to PDM-QPSK for high BER (i.e. for longer transmitted distance). This corresponds well to the back-to-back characterization reported in Fig. 2(b). However, in the nonlinear transmission regime (i.e. for launch power > −1 dBm) PS-QPSK is able to achieve a lower BER due to its better resilience against intra-channel nonlinear effects. For optimum launch power and at a BER of 10−3 we measure a maximum reach of ~9000 km for 112-Gb/s PDM-QPSK (at 0 dBm) and of 11000 km for 112-Gb/s PS-QPSK. This corresponds to approximately 22% increased maximum reach. Since we observe a higher implementation penalty for 112-Gb/s PS-QPSK compared to 112-Gb/s PDM-QPSK the increase is not as pronounced as at a bit rate of 84 Gb/s.
4. Conclusion
We investigated the BER performance of 84-Gb/s and 112-Gb/s PS-QPSK signals and compared it to the performance of PDM-QPSK signals at the same bit rates. This comparison is made in a back-to-back configuration as well as in single-channel ultra long-haul transmission experiments over up to 12500 km ULAF. For a bit rate of 84 Gb/s we confirmed the theoretically predicted sensitivity advantage of approximately 1 dB at a BER of 10−3. We showed that an implementation of 112-Gb/s PS-QPSK is more challenging due to the required symbol rate of 37.3 GBd, resulting in only 0.6 dB measured sensitivity improvement. After single-channel transmission over the fiber link, it was observed that PS-QPSK not only benefits from its better sensitivity but also offers a better tolerance against intra-channel nonlinear effects resulting in an increased transmission reach compared to PDM-QPSK.
Acknowledgments
The authors would like to thank OFS for kindly providing the ultra large effective area fibers and SHF for configuring a customized version of our bit pattern generator.
References and links
1. M. Karlsson and E. Agrell, “Which is the most power-efficient modulation format in optical links?” Opt. Express 17(13), 10814–10819 (2009). [CrossRef] [PubMed]
2. E. Agrell and M. Karlsson, “Power-efficient modulation formats in coherent transmission systems,” J. Lightwave Technol. 27(22), 5115–5126 (2009). [CrossRef]
3. H. Bülow, “Polarization QAM modulation (POL-QAM) for coherent detection schemes,” in Proc. Opt. Fiber Commun. Conf. (2009), paper OWG2.
4. P. Poggiolini, G. Bosco, A. Carena, V. Curri, and F. Forghieri, “Performance evaluation of coherent WDM PS-QPSK (HEXA) accounting for non-linear fiber propagation effects,” Opt. Express 18(11), 11360–11371 (2010). [CrossRef] [PubMed]
5. P. Serena, A. Vannucci, and A. Bononi, “The performance of polarization switched-QPSK (PS-QPSK) in dispersion managed WDM transmission,” in Proc. 36th Eur. Conf. Opt. Commun. (2010), paper Th.10.E.2.
6. M. Sjödin, P. Johannisson, H. Wymeersch, P. A. Andrekson, and M. Karlsson, “Comparison of polarization-switched QPSK and polarization-multiplexed QPSK at 30 Gbit/s,” Opt. Express 19(8), 7839–7846 (2011). [CrossRef] [PubMed]
7. D. S. Millar, D. Lavery, S. Makovejs, C. Behrens, B. C. Thomsen, P. Bayvel, and S. J. Savory, “Generation and long-haul transmission of polarization-switched QPSK at 42.9 Gb/s,” Opt. Express 19(10), 9296–9302 (2011). [CrossRef] [PubMed]
8. C. Behrens, D. Lavery, D. Millar, S. Makovejs, B. C. Thompson, R. Killey, S. Savory, and P. Bayvel, “Ultra-long-haul transmission of 7×42.9Gbit/s PS-QPSK and PM-BPSK,” in Proc. 37th Eur. Conf. Opt. Commun. (2011), paper Mo.2.B.2.
9. L. E. Nelson, X. Zhou, N. Mac Suibhne, A. D. Ellis, and P. Magill, “Experimental comparison of coherent polarization-switched QPSK to polarization-multiplexed QPSK for 10 × 100 km WDM transmission,” Opt. Express 19(11), 10849–10856 (2011). [CrossRef] [PubMed]
10. J. K. Fischer, L. Molle, M. Nölle, D.-D. Groß, and C. Schubert, “Experimental investigation of 28-GBd polarization-switched quadrature phase-shift keying signals,” in Proc. 37th Eur. Conf. Opt. Commun. (2011), paper Mo.2.B.1.
11. M. Wrigth, “Comments on ‘Aspects of MLS measuring systems’,” J. Audio Eng. Soc. 43, 48–49 (1995).
12. I. Fatadin, S. J. Savory, and D. Ives, “Compensation of quadrature imbalance in an optical QPSK coherent receiver,” IEEE Photon. Technol. Lett. 20(20), 1733–1735 (2008). [CrossRef]
13. M. Selmi, Y. Jaouën, and P. Ciblat, “Accurate digital frequency offset estimator for coherent PolMux QAM transmission systems,” in Proc. 35th Eur. Conf. Opt. Commun. (2009), paper P3.08.
14. D. S. Millar and S. J. Savory, “Blind adaptive equalization of polarization-switched QPSK modulation,” Opt. Express 19(9), 8533–8538 (2011). [CrossRef] [PubMed]
15. J. Renaudier, O. Bertran-Pardo, H. Mardoyan, M. Salsi, P. Tran, E. Dutisseuil, G. Charlet, and S. Bigo, “Experimental comparison of 28Gbaud polarization switched- and polarization division multiplexed- QPSK in WDM long-haul transmission system,” in Proc. 37th Eur. Conf. Opt. Commun. (2011), paper Mo.2.B.3.
