Open Access
Add to BibSonomyAbstract
We propose an adaptive detection technique in super-Nyquist wavelength-division-multiplexed (WDM) polarization-division-multiplexed quadrature-phase-shift-keying (PDM-QPSK) systems, where a QPSK signal is digitally converted to a quadrature n-level polybinary signal followed by a MLSE detector at the receiver, and study the performance of quadrature-duobinary and quadrature four-level polybinary signals using this detection technique. We change the level of the quadrature-polybinary modulation at the coherent receiver according to the channel spacing of a super-Nyquist system. Numerical studies show that the best performance can be achieved by choosing different modulation levels at the receiver in adaption to the channel spacing. In the experiment, we demonstrate the transmission of 3-channel 112-Gbit/s PDM-QPSK signals at a 20-GHz channel spacing, which is detected as a quadrature four-level polybinary signal, with performance comparable to PDM 16-ary quadrature-amplitude modulation (16QAM) at the same bit rate.
© 2015 Optical Society of America
1. Introduction
High spectral-efficiency (SE) optical communication networks have been an important technology to meet the demand for high capacity communication services [1]. There are two ways to increase SE. One is to use high order modulation formats and the other is to reduce channel spacing. In general, the increase of modulation levels quickly increases the required optical signal-to-noise ratios (OSNRs), transmitter complexities and implementation penalties. Meanwhile, super-Nyquist wavelength-division-multiplexing (WDM), which operates with a narrower than symbol-rate channel spacing, has received lots of attention recently, and typically sophisticated detection techniques such as maximum-likelihood-sequence estimation (MLSE) [2] are used to mitigate the inter-symbol interference (ISI). The mostly explored modulation format for super-Nyquist WDM is quadrature-phase-shift-keying (QPSK), which does not required a digital-to-analog converter (DAC) or a linear electrical driver at the transmitter. The transmitter is simpler than 8-ary quadrature-amplitude modulation (8QAM) and 16QAM. Polarization-division-multiplexed (PDM) QPSK transmission with 4-bit/s/Hz SE has been achieved by various detection techniques [3–9].
In this paper, we show that the SE of a super-Nyquist WDM QPSK system can be further increased by converting the QPSK signal to a quadrature n-level polybinary signal followed by an MLSE detector at the receiver. Without changing the transmitter, we adapt the detection scheme according to the channel spacing. Extensive simulations of a 3-channel 112-Gbit/s PDM-QPSK WDM system have been performed with the channel spacing ranging from 28-GHz to 15-GHz. We show that, to achieve the best performance, the levels of polybinary detection need to be changed when we vary the channel spacing of the super-Nyquist WDM QPSK system. Previous optical demonstrations of polybinary modulation with higher than 3 levels showed limited performances [10,11]. In one experiment, we demonstrate the detection of three 112-Gbit/s PDM-QPSK WDM channels multiplexed at a 20-GHz channel spacing, achieving a 5-bit/s/Hz SE [12]. No electrical pulse shaping is employed at the transmitter. The received QPSK signal is digitally converted to a quadrature four-level polybinary (Q-4PB) signal followed by an MLSE detector. The measured back-to-back (BTB) OSNR at bit-error ratio (BER) of 10−3 is 19.6 dB. We successfully transmit the signal over 960-km standard-single-mode-fiber (SSMF) link with erbium-doped-fiber amplifiers (EDFAs) at 7% hard-decision forward-error-correction (HD-FEC) threshold (3.8x10−3), which has similar performance with the 16QAM system.
2. Adaptive polybinary detection
Known as a class of partial response signaling, polybinary modulation formats with n levels are usually generated by a series of delay-&-add operations, as shown in Fig. 1(a). For instance, the conversion from a QPSK constellation to a Q-4PB constellation is shown in Fig. 1(b). Regardless of the number of levels in a polybinary signal, each polybinary symbol only carries one bit information. However, the spectral width of a polybinary symbol decreases with the increase of levels n, which can be described by the power-spectral density of a polybinary signal [10,13]
where and are the amplitude and symbol period of input non-return-to-zero (NRZ) pulses. In a super-Nyquist WDM system, we can adjust and design the number of levels of a polybinary symbol with respect to the channel spacing to obtain the best performance. Moreover, the implementation of quadrature-polybinary modulation formats can be performed with digital signal processing (DSP) at the coherent receiver, instead of the transmitter.
Fig. 1 (a) Generation of poly-binary signals with delay-&-add operations. (b) Conversion from a QPSK constellation to a Q-4PB constellation using a z-transfer function.
The proposed super-Nyquist WDM system with quadrature-polybinary detection is shown in Figs. 2(a)-2(b). At the transmitter, QPSK channels are combined at a narrower than symbol-rate channel spacing, where the QPSK signals are shaped by a WDM multiplexer. No electrical pulse shaping is required, as shown in Fig. 2(a). At the receiver, the QPSK signals are detected as quadrature-polybinary signals, as shown in Fig. 2(b).
Fig. 2 (a) Transmitter side of the super-Nyquist PDM-QPSK scheme. (b) Receiver side of the super-Nyquist PDM-QPSK scheme with quadrature-polybinary detection. (c) Constellations before and after carrier recovery for QPSK, QDB and Q-4PB detection of a QPSK signal at the receiver.
In the receiver side DSP, a narrow-bandwidth low-pass filter is used to reduce inter-channel crosstalk from neighboring channels. A finite impulse response (FIR) filter compensates chromatic dispersion (CD) in the transmission link. Then the quadrature n-level polybinary conversion is realized by an n-symbol delay-&-add filter together with a butterfly equalizer employing the multi-modulus algorithm (MMA). Note that, constant modulus algorithm (CMA) cannot work well for a quadrature-polybinary signal anymore, which is due to the multi-level feature as well as the statistic-unevenly distributed constellation (i.e. denser in the middle region). Afterwards, carrier frequency and phase recovery is performed. Finally, a MLSE detector is used for error detection. The tap number of MLSE is equal or higher than n-1.
Figure 2(c) visualizes the detection of a QPSK signal as different levels of quadrature-polybinary signals. For QPSK detection, the level n equals to 2. For QDB detection, the level n equals to 3. For Q-4PB detection, the level n equals to 4. The constellations at phase 1 and phase 2 in the DSP procedure is shown accordingly, which proves the feasibility of the adaptive quadrature-polybinary detection scheme.
3. Simulations at different channel spacings
In order to evaluate the performance of the quadrature-polybinary detection scheme in super-Nyquist WDM systems, simulations are conducted on a BTB 3-channel 112-Gb/s PDM-QPSK system with different channel spacings. In comparison, the performance of 3-channel PDM-8QAM and PDM-16QAM systems with the same bit rate is studied as well.
At the transmitter, each channel is modulated with a de-correlated 215-1 pseudo-random bit sequence (PRBS) to generate 28-Gbaud QPSK, 18.656-Gbaud 8QAM or 14-Gbaud 16QAM signals. No digital pulse shaping is employed here. The lasers have a 100-kHz linewidth. The 3 channels are combined together with a WDM multiplexer, which shapes the spectrum of signals. The channel spacing ranges from 15 GHz to 28 GHz. At the receiver, the local oscillator (LO) laser has a 100-kHz linewidth and the frequency offset from the transmitter laser is 200 MHz in each channel. A low-pass filter is used to reduce inter-channel crosstalk. Both the WDM multiplexer and the low-pass filter are modeled as a 5th-order super-Gaussian filter with the optimum bandwidth, which is found by sweeping the 3-dB bandwidth values.
In the receiver side DSP, the PDM-QPSK signals are detected as QPSK signals, QDB signals or Q-4PB signals, followed by a 7-tap MLSE detector implemented on the in-phase and quadrature paths. Note that in principle a 3-tap MLSE detector is sufficient to detect QPSK signals, QDB signals and Q-4PB signals. However, there is still some additional ISI left due to the deviation of the achieved spectral pattern from the target pattern. We choose a 7-tap MLSE detector for the tradeoff between complexity and performance. The PDM-8QAM or PDM-16QAM signals employing the same MMA are detected with symbol-by-symbol (SbS) detection, as shown in Fig. 3(a). Carrier recovery in both schemes is realized by a second-order phase-locked loop (PLL) [14].
Fig. 3 (a) Receiver side DSP procedure for QPSK, 8QAM and 16QAM systems. (b) The required OSNR at BER of 10−3 v.s. channel spacing in 3-channel 112-Gbit/s PDM-QPSK (detected as QPSK, QDB and Q-4PB signals), PDM-8QAM and PDM-16QAM systems.
The required OSNRs at BER of 10−3 as a function of channel spacings are shown in Fig. 3(b). When the channel spacing is less than 25-GHz, QDB starts to have better performance than QPSK. Further reducing the channel spacing to less than 20-GHz, Q-4PB starts to perform better than QDB. To achieve the best performance, different channel spacings need quadrature-polybinary detection with different levels. Therefore, by applying the adaptive quadrature-polybinary detection scheme on super-Nyquist WDM QPSK systems, we can achieve a much narrower than symbol-rate channel spacing, with performance comparable to 8QAM or 16QAM systems at the same bit rate.
4. Experiments and discussions
4.1 Experiment setup
We conduct a 3-channel 112-Gbit/s PDM-QPSK experiment at a 20-GHz channel spacing, which is compared with a PDM-16QAM WDM system at the same bit rate and spectral efficiency. The BTB experimental configuration is shown in Fig. 4(a). At the transmitter, three external cavity lasers (ECLs) with ~100-kHz linewidth spaced at 20 GHz are used. The outputs of the ECLs for channels 1 and 3 are combined together with a 3-dB polarization maintaining coupler and then sent to one nested Mach-Zehnder (MZ) modulator and the output of the ECL for channel 2 is sent to another nested MZ modulator. The in-phase and quadrature arms of both modulators are driven by electrical signals directly from a DAC with a peak-to-peak voltage Vpp = 600 mV. The QPSK system is modulated by 28-Gbit/s binary electrical signals for in-phase and quadrature with de-correlated 215-1 PRBS, and the 16QAM system is modulated by 14-Gbit/s 4-level pulse-amplitude modulation (4PAM) electrical signals for in-phase and quadrature with de-correlated 217-1 PRBS. The eye diagrams of the electrical driving signals are shown as the insets. No electrical pulse shaping is used for both systems. The outputs of the two modulators are multiplexed together with a wavelength selective switch (WSS) at a 20-GHz channel spacing. The 3-dB bandwidth of the WSS is 14 GHz. The spectra of QPSK and 16QAM WDM signals, which are measured at 0.01 nm resolution, are shown in Fig. 4(b). The signals are then polarization multiplexed with a polarization multiplexer (PolMux) with ~66.39-ns time delay between the two polarization tributaries.
Fig. 4 (a) Experiment setup in BTB configuration. (b) The spectra of QPSK and 16QAM WDM signals. (c) The transmission loop setup.
The transmission is performed in a 4x80-km SSMF recirculating loop with EDFAs, as shown in Fig. 4(c). A dynamic gain equalizing filter (DGEF) is used after each loop to block amplified spontaneous emission (ASE) noises. At the receiver, a tunable optical attenuator and two EDFAs with an optical filter in between are used to adjust the signal OSNR in BTB configuration. The received signal is mixed with a free-running ECL LO in a polarization-and-phase diversity coherent receiver, with its four outputs captured by an 80-GSa/s real-time oscilloscope with a 30-GHz bandwidth, and 2 million points are captured for offline processing.
The offline processing schemes for the Q-4PB detection and 16QAM detection are similar with simulations. The sampling skew is first corrected and the signals are resampled to 2 samples per symbol. A 5th-order super-Gaussian filter is used to filter out inter-channel crosstalk from neighboring channels. For QPSK systems, after the 3-symbol delay-&-add operation, a 19-tap MMA equalizer is used for Q-4PB shaping and polarization demultiplexing. For 16QAM systems, a 19-tap MMA equalizer is used for ISI mitigation and polarization demultiplexing. The laser frequency offset and phase noises are estimated by the second-order PLL. Finally, a 7-tap MLSE detector or a SbS detector is employed to calculate BERs.
4.2 Results and discussions
The BTB and transmission experimental results are discussed. In Fig. 5, we compare the BTB performance of the 3-channel 112-Gbit/s PDM-QPSK system at a 20-GHz channel spacing using the QPSK, QDB and Q-4PB detection with that of the PDM-16QAM system. The optimized 3-dB bandwidth of the low-pass filter is 15.4 GHz for QPSK detection, 21 GHz for QDB detection and Q-4PB detection, and 16.8 GHz for 16QAM detection. It is observed that, with the same 7-tap MLSE detector, Q-4PB detection has significant performance improvement than QDB and QPSK detection in the experiment. Meanwhile, the measured OSNR at BER of 10−3 is 19.6 dB for Q-4PB detection and 20.6 dB for 16QAM detection. About 1-dB OSNR improvement is observed with Q-4PB detection than 16QAM at the same bit rate, which may be due to the modulator nonlinearity and implementation penalties for a 16QAM system.
Fig. 5 BTB results of the 112-Gbit/s super-Nyquist PDM-QPSK system (detected as QPSK, QDB, and Q-4PB signals) and 112-Gbit/s PDM-16QAM system at 20-GHz channel spacing.
The transmission results comparing 112-Gbit/s super-Nyquist PDM-QPSK system (detected as Q-4PB signals) with PDM-16QAM system at 20-GHz channel spacing are shown in Fig. 6. The dependence of the BER performance on signal launch power after 960-km transmission is shown in Fig. 6(a). In the low OSNR region, Q-4PB detection tends to have slightly worse performance than 16QAM. In the nonlinear region, Q-4PB detection has similar performance with 16QAM, since MLSE is implemented separately on the in-phase and quadrature paths. The optimal total launch power for both schemes is around 2 dBm, and we set it as the launch power for measuring BER at different transmission distances, as shown in Fig. 6(b). It is observed that the super-Nyquist PDM-QPSK system using the Q-4PB detection scheme has comparable performance with the PDM-16QAM system.
Fig. 6 (a) BER vs. total launch power after 960-km transmission. (b) BER vs. distance at 2-dBm total launch power.
5. Conclusion
We propose an adaptive quadrature-polybinary detection technique combined with an MLSE detector, where the channel spacing of a super-Nyquist WDM QPSK system can be much smaller than the symbol rate, and use this technique to study the performance of quadrature-duobinary and quadrature four-level polybinary signals. The transmitter has a simple structure, which does not require a DAC or a linear electrical driver. Numerical results show that the best performance can be achieved by adapting the levels of the quadrature-polybinary detection at the receiver to the channel spacing. In the experiment, we demonstrate a 112-Gb/s super-Nyquist WDM PDM-QPSK system at 20-GHz channel spacing, achieving a 5-bit/s/Hz SE. The required OSNR at BER of 10−3 for the Q-4PB scheme is 19.6 dB. Over 960-km transmission is achieved for the super-Nyquist PDM-QPSK system at a 5-b/s/Hz SE with Q-4PB detection.
As no DAC and linear amplifiers are required at transmitters, the adaptive detection technique for super-Nyquist WDM QPSK systems could be used as an alternative to higher-level modulation in some application scenarios, such as the upstream path in asymmetric passive optical networks. Considering that MLSE equalizers for 10-Gb/s on-off-keying systems were commercialized more than 10 years ago [15], there should be no technical difficulties to implement the MLSE technique in coherent receivers today.
Acknowledgments
This work was supported in part by 863 program (2012AA011301), NSFC project (61271189, 61201154, 61302085), Ministry of Education-China Mobile Research Foundation (MCM20130132), RFDP Project (20120005120019), and Fund of State Key Laboratory of Information Photonics and Optical Communications (BUPT).
References and links
1. X. Liu, S. Chandrasekhar, and P. J. Winzer, “Digital signal processing techniques enabling multi-Tb/s superchannel transmission: an overview of recent advances in DSP-enabled superchannels,” IEEE Signal Process. Mag. 31(2), 16–24 (2014). [CrossRef]
2. G. D. Forney Jr., “Maximum likelihood sequence estimation of digital sequences in the presence of intersymbol interference,” IEEE Trans. Inf. Theory 18(3), 363–378 (1972). [CrossRef]
3. J.-X. Cai, Y. Cai, C. R. Davidson, D. G. Foursa, A. Lucero, O. Sinkin, W. Patterson, A. Pilipetskii, G. Mohs, and N. Bergano, “Transmission of 96 100-Gb/s bandwidth-constrained PDM-RZ-QPSK channels with 300% spectral efficiency over 10610 km and 400% spectral efficiency over 4370 km,” J. Lightwave Technol. 29(4), 491–498 (2011). [CrossRef]
4. J. Li, Z. Tao, H. Zhang, W. Yan, T. Hoshida, and J. C. Rasmussen, “Spectrally efficient quadrature duobinary coherent systems with symbol-rate digital signal processing,” J. Lightwave Technol. 29(8), 1098–1104 (2011). [CrossRef]
5. J. Zhang, J. Yu, N. Chi, Z. Dong, J. Yu, X. Li, L. Tao, and Y. Shao, “Multi-modulus blind equalizations for coherent quadrature duobinary spectrum shaped PM-QPSK digital signal processing,” J. Lightwave Technol. 31(7), 1073–1078 (2013). [CrossRef]
6. J. Yu, J. Zhang, Z. Dong, Z. Jia, H.-C. Chien, Y. Cai, X. Xiao, and X. Li, “Transmission of 8 × 480-Gb/s super-Nyquist-filtering 9-QAM-like signal at 100 GHz-grid over 5000-km SMF-28 and twenty-five 100 GHz-grid ROADMs,” Opt. Express 21(13), 15686–15691 (2013). [CrossRef] [PubMed]
7. J. Zhang, J. Yu, Z. Jia, and H. C. Chien, “400 G transmission of super-Nyquist-filtered signal based on single-carrier 110-GBaud PDM QPSK with 100-GHz grid,” J. Lightwave Technol. 32(19), 3239–3246 (2014). [CrossRef]
8. S. Chen, C. Xie, and J. Zhang, “Comparison of advanced detection techniques for QPSK signals in super-Nyquist WDM systems,” IEEE Photon. Technol. Lett. 27(1), 105–108 (2015). [CrossRef]
9. C. Xie and S. Chen, “Quadrature Duobinary Modulation and Detection,” in Proc. OFC, paper W4K.6 (2015).
10. S. Walklin and J. Conradi, “Multilevel signaling for increasing the reach of 10 Gb/s lightwave systems,” J. Lightwave Technol. 17(11), 2235–2248 (1999). [CrossRef]
11. J. J. V. Olmos, L. F. Suhr, B. Li, and I. T. Monroy, “Five-level polybinary signaling for 10 Gbps data transmission systems,” Opt. Express 21(17), 20417–20422 (2013). [CrossRef] [PubMed]
12. S. Chen, C. Xie, and J. Zhang, “Advanced detection of super-Nyquist WDM QPSK signals with 5-bit/s/Hz spectral efficiency,” in Proc. OFC, paper Th2A.11 (2015).
13. A. Lender, “Correlative digital communication techniques,” IEEE Trans. Commun. Technol, COM 12(4), 128–135 (1964). [CrossRef]
14. C. Xie and G. Raybon, “Digital PLL based frequency offset compensation and carrier phase estimation for 16-QAM coherent optical communication systems,” in Proc. ECOC, paper Mo.1.A.2 (2012). [CrossRef]
15. A. Farbert, S. Langenbach, N. Stojanovic, C. Dorschky, T. Kupfer, C. Schulien, J.-P. Elbers, H. Wernz, H. Griesser, and C. Glingener, “Performance of a 10.7 Gb/s receiver with digital equaliser using maximum likelihood sequence estimation,” in Proceedings of ECOC, paper Th4.1.5 (2004).
