Abstract
We transmit a mix of 260-Gb/s polarization-division-multiplexed 16-ary quadrature-amplitude modulation (PDM-16QAM) and 130-Gb/s polarization-division-multiplexed quadrature-phase-shift-keying (PDM-QPSK) channels at a 50-GHz channel spacing in a dispersion-managed (DM) system with standard single-mode-fiber (SSMF) spans. We study the impact of pulse shaping, time interleaving of polarizations and maximum likelihood (ML) detection techniques on the performance of the system. We show that the pulse shaping and ML detection can increase the transmission distances of the PDM-16QAM channels and PDM-QPSK channels by 50% and 10%, respectively. With 20% overhead hard-decision forward-error-correction (FEC) coding, we successfully transmit the 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK channels over 960-km and 4,160-km, respectively, in the DM system.
©2012 Optical Society of America
1. Introduction
With the recent commercialization of 100-Gb/s technologies using polarization-division-multiplexed quadrature-phase-shift-keying (PDM-QPSK) and digital coherent detection, the focus of the optical communication industry is moving beyond 100-Gb/s. 16-ary quadrature-amplitude modulation (16QAM) with digital coherent detection is considered a promising candidate for such systems [1–4]. Compared with QPSK, 16QAM requires higher optical signal-to-noise ratios (OSNRs) and is more susceptible to hardware imperfections and has a higher implementation penalty than QPSK. In addition, 16QAM is more sensitive to fiber nonlinearities, which becomes more severe in dispersion-managed (DM) links over legacy fiber infrastructure [5]. Most of the 16QAM system experiments demonstrated so far used dispersion-compensation-fiber (DCF) free transmission links and/or new low loss and low nonlinearity fibers [1–4], which requires green field deployments. However, many existing optical networks have DCF inline, and in addition, when upgrading these networks to bit rates beyond 100-Gb/s, channels of different modulation formats and bit rates may co-exist. Therefore it is important to investigate the transmission performance of these systems where 16QAM co-propagating with other modulation formats.
In this paper, we investigate the transmission performance of mixed 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK channels at a 50-GHz channel spacing in legacy DM standard single-mode-fiber (SSMF) spans using all-Raman amplification [6]. We investigate the impact of pulse shaping, time interleaving of polarizations and maximum likelihood (ML) detection techniques on the performance of the system. We show that pulse shaping and maximum likelihood detection can reduce nonlinear impairments from both fiber nonlinearities and transmitter imperfections, and time interleaving polarizations by half symbol period can increase the nonlinear transmission performance. By using all these techniques, we can increase the transmission distances by 50% and 10% for the 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK channels, respectively, in the modulation format mixed DM system.
2. Pulse shaping and maximum likelihood detection
Pulse shaping has been used more and more in optical communication systems. It can be used either to increase the signal tolerance to fiber nonlinearities [4,7], or to increase the spectral efficiency (SE) of optical communication systems [5,7–9]. Using the Nyquist pulse shaping, which shapes the spectrum of a signal according to the Nyquist criterion, close to symbol rate channel spacing can be achieved [8,9]. Pulse shaping can be performed either with optical approaches or using digital signal processing (DSP) based electrical approaches. We find that the Nyquist pulse shaping can also increase the signal tolerance to fiber nonlinearities. Figure 1 shows the simulated spectra of a 32.5-Gbaud QPSK signal with and without pulse shaping. The spectra after the pulse shaping has a raised-cosine (RC) shape with a roll-off factor of 0.05. The eye-diagram after the pulse shaping looks like a return-to-zero (RZ) eye-diagram and signal amplitudes between the centers of symbols have a large variation. We will show later that those features will help to improve nonlinear transmission performance of a signal.
A detector makes decision on the transmitted signal based on observation of the received signal. One optimum decision method is called the maximum a posteriori probability (MAP) method, which maximums the probability of a correct decision [10]
where si and ri are the transmitted and received signals, is the decided signal, and P(si|ri) is the posterior probability, which is the probability that signal si is transmitted if signal ri is received. It was found that MAP detection can also be used to mitigate fiber nonlinearity induced impairments [11]. If all the transmitted symbols P(si) are equiprobable, according to Baye’s rule, the MAP detection becomes the same as the maximum likelihood (ML) detection, which is defined asP(si|ri) is the likelihood function, which is the probability that signal ri is received if signal si is transmitted. The ML detection is easier to implement than the MAP detection.In an additive white Gaussian noise (AWGN) channel, the ML detection is equivalent to the minimum distance detection, which minimizes the Euclidian distance
The minimum distance detection can be further simplified to the detection using fixed threshold boundaries, which is widely used in current optical communication systems. For example, square detection boundaries are used in most QPSK and square 16QAM systems.
For a signal with nonlinear distortions from transmitters or fiber nonlinearities, it is hard to find the optimum threshold boundaries. In this case, using the ML detection can obtain much better performance than the conventional detection method using the fixed detection boundaries under the AWGN channel assumption.
3. Experimental setup
The experiment was done in a L-band transmission system [12], and the experimental setup is shown in Fig. 2 . Sixteen L-band channels at a 50-GHz channel spacing ranging from 1577.44 nm to 1583.69 nm were separated into two groups, one group was used for PDM-16QAM channels and the other was for PDM-QPSK channels. The channel under test was from a tunable external cavity laser (ECL) of ~100-kHz linewidth and all the other channels were from distributed feedback (DFB) lasers. Figures 2(b) and 2(c) depict the setups for the 16QAM and QPSK transmitters. Two 4x1 mutiplexers multiplexed four delay-decorrelated copies of a 8.125-Gb/s 215-1 pseudo-random bit-sequence (PRBS) to generate four 32.5-Gb/s bit sequences, and then we combined outputs from and of the two multiplexers to form two 32.5-Gaud 4-level electrical signals, which were decorrelated with a delay line and used to drive the inphase and quadrature branches of a nested Mach-Zehnder modulator (MZM) to generate a 130-Gb/s optical 16QAM signal. For the QPSK transmitter, we used another 4x1 multiplexer to obtain two 32.5-Gb/s bit sequences from four delay-decorrelated copies of a 8.125-Gb/s 215-1 PRBS, and we delay-decorrelated the two bit sequences of the multiplexer, and , to drive the inphase and quadrature branches of another nested MZM to generate a 65-Gb/s optical QPSK signal. The 4-level electrical inphase eye-diagram and optical eye-diagrams of the 16QAM and QPSK signals are given in the insets of Fig. 2. Finisar waveshapers were used for pulse shaping, which were programmed to flatten the signal spectra to improve system performance [4,6,8]. Note that pulse shaping can also be done with DSP. After the waveshapers, the 16QAM and QPSK signals were combined by a 50/100 GHz interleaver. One polarization multiplexer (PolMux) with a 336-symbol delay between the two polarization tributaries was used to generate 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK signals. We tuned the delay line between the two polarization tributaries to time align or interleave the signals in the two polarizations by half symbol to investigate the effect of polarization interleaving on the transmission performance of the system [13].
Transmission experiments were performed in a 4x80-km DM all-Raman amplified SSMF recirculating loop. The dispersion of each span was compensated by a DCF with a residual dispersion per span near 30 ps/nm at 1580 nm, and a DCF with −300 ps/nm was used for dispersion pre-compensation. Both SSMF and DCF were backward Raman pumped, with the launch power to the DCF about 2-dB lower than that to the spans. The signal spectrum was flattened by a dynamic gain equalizing filter (DGEF) after each loop and the losses of switches and DGEF were compensated by an erbium-doped-fiber amplifier (EDFA).
In the receiver, the signal was first filtered by an amplified spontaneous emission (ASE) noise filter and amplified by an EDFA, then mixed with a free-running ECL local oscillator (LO) in a polarization diversity 90° hybrid, followed by four balanced detectors with bandwidths of 40 GHz. The four signal components were captured by two 2-channel 80-GSamples/s real-time oscilloscopes with 30-GHz bandwidths. The captured signal was digitally processed offline. The sampling skew was first corrected and the signal was synchronously re-sampled to 2 samples per symbol. After dispersion compensation, a butterfly equalizer with 19 taps was used for polarization demultiplexing and inter-symbol interference compensation. For PDM-QPSK, the equalizer was adapted via the constant modulus algorithm (CMA), and for PDM-16QAM, it was first adjusted via the CMA for pre-convergence and then finely tuned with the multi-modulus algorithm (MMA). A phase increment estimation algorithm and the Viterbi & Viterbi algorithm were used for carrier frequency estimation and phase recovery for QPSK, respectively. For 16QAM, carrier frequency and phase estimation was performed with a 2nd-order phase-locked loop (PLL) [14,15]. We performed symbol detection with both the conventional detection method, which uses square boundaries to identify symbols, and the ML detection. The ML detection was performed with a lookup table, which was set up with training symbols. Bit-error ratios (BERs) were calculated using direct error-counting.
4. Experimental results
Figure 3 depicts the spectra of 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK with and without the pulse shaping using the waveshaper. We programed the waveshaper with the measured signal spectra to generate square shaped spectra with the bandwidth of the symbol rate. Figure 3 shows that signal spectra after the waveshaper become flatter (The waveshapers were programmed to have a “V” shape transfer function), but due to the resolution of the optical spectrum analyzer and the waveshaper, the spectra are not sharp in the edges. The optical eye-diagrams of the 16QAM and QPSK signals after the waveshaper are given in Fig. 4 . Compared with the eye-diagrams in Fig. 2, the eye-diagrams of the signals after the waveshaper look like those of RZ signals, which agrees with simulation results in Fig. 1.
Back-to-back BER performance of the 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK is plotted in Fig. 5 . For 16QAM, the pulse shaping improves the back-to-back performance by about 1.0 dB and the ML detection can reduce the required OSNR by additional 0.5 dB. However, both pulse shaping and the ML detection have little impact on the performance of QPSK. This is because there are much larger distortions in the 16QAM than in the QPSK (cf. constellation of the 16QAM signal in inset of Fig. 2). At a BER of 10−3, compared with the theoretical result, the 16QAM signal has about 5.0-dB implementation penalty (without the waveshaper and ML detection), whereas the QPSK signal only has 1.5-dB penalty. More benefits can be obtained from these techniques if there are larger distortions.
Q2-factors (calculated from BERs) of a central 260-Gb/s PDM-QPSK channel without and with the waveshaper versus launch power after 960-km transmission are given in Figs. 6(a) and 6(b), respectively. The results of both the polarization aligned and polarization interleaved (two polarizations are time interleaved by half symbol period) cases with and without the ML detection are plotted in the figures. It shows that time interleaving the symbols in two polarizations improves the nonlinear transmission performance for both the signals with and without the pulse shaping, and the ML provides additional 0.5 dB improvement and the improvement tends to be larger at higher powers, indicating that there are nonlinearity induced signal distortions. The pulse shaping not only increases the achievable Q2-factor but increase the tolerable launch power as well. Using the pulse shaping and polarization interleaving, the optimum launch power is increased by about 1.5 dB, to about −7 dBm per channel. The optimum Q2-factor is improved by 2.0 dB, from about 5.0 dB to 7.0 dB when all these techniques are applied.
We then used −7 dBm launch power per channel and measured the performance of a central PDM-16QAM and PDM-QPSK channel versus transmission distances and the results are shown in Fig. 7 . For the case without the waveshaper, the symbols in the two polarizations were aligned, and the symbols were interleaved when the waveshaper was used. The ML detection was used for all the cases. It shows that above a 20% overhead hard-decision FEC threshold (Q2-factor of 6.73 dB) [16], the pulse shaping and the polarization interleaving techniques can increase the transmission distance by 50% for 260-Gb/s PDM-16QAM, from 640 km to 960 km, and by about 10% for 130-Gb/s PDM-QPSK, from 3840 km to 4160 km. We also plot the result with all the channels carrying PDM-QPSK in Fig. 7(b), and it shows that PDM-16QAM has a similar impact on the neighboring PDM-QPSK channels as PDM-QPSK.
The Q2-factors of all eight 260-Gb/s PDM-16QAM channels after 960-km transmission and eight 130-Gb/s PDM-QPSK channels after 4,160-km transmission are illustrated in Fig. 8 , where both the pulse shaping and ML detection are used. All the channels achieve performance above the 20% overhead hard-decision FEC threshold.
5. Conclusion
We have studied the transmission of mixed 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK channels at a 50-GHz channel spacing in DM SSMF spans. We showed that the pulse shaping and polarization interleaving can increase the nonlinear transmission performance of PDM-16QAM and PDM-QPSK in the DM system and the ML detection provides additional benefit for 16QAM. We also found that PDM-16QAM has a similar impact on the neighboring PDM-QPSK channels as PDM-QPSK in such system. By using these techniques, we have successfully transmitted 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK channels over 960 km and 4,160 km in the DM system, respectively.
References and links
1. A. H. Gnauck, P. J. Winzer, S. Chandrasekhar, X. Liu, and D. W. Peckham, “10 x 224-Gb/s WDM transmission of 28-Gbaud PDM 16-QAM on a 50-GHz grid over 1,200 km of fiber,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2010), paper PDPB8.
2. K. Schuh, F. Buchali, D. Roesener, E. Lach, O. Bertran-Pardo, J. Renaudier, G. Charlet, H. Mardoyan, and P. Tran, “15.4 Tb/s transmission over 2400 km using polarization multiplexed 32-Gbaud 16-QAM modulation and coherent detection comprising digital signal processing,” in Proc. European Conference on Optical Communication, (Geneva, Switzerland, 2011), paper We.8.B.4.
3. S. Makovejs, E. Torrengo, D. S. Millar, R. I. Killey, S. J. Savory, and P. Bayvel, “Comparison of pulse shapes in a 224Gbit/s (28Gbaud) PDM-QAM16 long-haul transmission experiment,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2011), paper OMR5.
4. B. Châtelain, C. Laperle, K. Roberts, X. Xu, M. Chagnon, A. Borowiec, F. Gagnon, J. C. Cartledge, and D. V. Plant, “Optimized pulse shaping for intra-channel nonlinearities mitigation in a 10 Gbaud dual-polarization 16-QAM system,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2011), paper OWO5.
5. C. Xie, “Fiber nonlinearities in 16QAM transmission systems,” in Proc. European Conference on Optical Communication, (Geneva, Switzerland, 2011), paper We.7.B.6.
6. C. Xie and G. Raybon, “Transmission of mixed 260-Gb/s PDM-16QAM and 130-Gb/s PDM-QPSK over 960-km and 4160-km dispersion-managed SSMF spans,” in Proc. European Conference on Optical Communication, (Amsterdam, the Netherlands, 2012), paper Mo.2.C.4.
7. C. Xie, G. Raybon, and P. J. Winzer, “Hybrid 224-Gb/s and 112-Gb/s PDM-QPSK transmission at 50-GHz channel spacing over 1200-km dispersion-managed LEAF® spans and 3 ROADMs,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2011), paper PDP D2.
8. G. Bosco, A. Carena, V. Curri, P. Poggiolini, and F. Forghieri, “Performance limits of Nyquist-WDM and CO-OFDM in high-speed PM-QPSK systems,” IEEE Photon. Technol. Lett. 22(15), 1129–1131 (2010). [CrossRef]
9. J. Wang and C. Xie, “Generation of spectrally efficient Nyquist WDM QPSK signals using DSP techniques at transmitter,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2012), paper OM3H.5.
10. J. G. Proakis, Digital Communications (McGraw-Hill, 2001), Chap. 5.
11. Y. Cai, D. G. Foursa, C. R. Davidson, J.-X. Cai, O. Sinkin, M. Nissov, and A. Pilipetskii, “Experimental demonstration of coherent MAP detection for nonlinearity mitigation in long-haul transmissions,” in Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest Series (CD) (Optical Society of America, 2010), paper OTuE1.
12. D. A. Fishman, W. A. Thompson, and L. Vallone, “LambdaXtreme® transport system: R&D of a high capacity system for low cost, ultra long haul DWDM transport,” Bell Labs Tech. J. 11(2), 27–53 (2006). [CrossRef]
13. C. Xie, “WDM coherent PDM-QPSK systems with and without inline optical dispersion compensation,” Opt. Express 17(6), 4815–4823 (2009). [CrossRef] [PubMed]
14. N. K. Jablon, “Joint blind equalization, carrier recovery, and timing recovery for high-order QAM signal constellations,” IEEE Trans. Signal Process. 40(6), 1383–1398 (1992). [CrossRef]
15. C. Xie and G. Raybon, “Digital PLL based frequency offset compensation and carrier phase estimation for 16-QAM coherent optical communication systems,” in Proc. European Conference on Optical Communication, (Amsterdam, the Netherlands, 2012), paper Mo.1.A.2.
16. F. Yu, C. Xie, and L. Zeng, “Application aspects of enhanced FEC for 40/100G systems,” in European Conference on Optical Communication, (Turino, Italy, 2010), workshop 11, 2011.