通信原理论文(中英文版)

Optimized Pulse Shaping for Intra-channel

Nonlinearities Mitigation in a 10 Gbaud Dual-Polarization 16-QAM System

Beno?t Chatelain1, Charles Laperle2, Kim Roberts2, Xian Xu1, Mathieu Chagnon1, Andrzej Borowiec2, Fran?ois Gagnon3, John C. Cartledge4, and David V. Plant1

1McGill University, Montreal, Quebec, Canada, H3A 2A7 2Ciena Corporation, Ottawa, Ontario, Canada, K2H 8E9

3école de technologie supérieure, Montreal, Quebec, Canada, H3C 1K3 4Queen’s University, Kingston, Ontario, Canada, K7L

3N6 Email: benoit.chatelain@mail.mcgill.ca

Abstract: An optimized pulse shape is shown to reduce intra-channel nonlinear effects in a 10 Gbaud dual-polarization 16-QAM EDFA-amplified system without optical dispersion compensation.

OCIS codes: (060.2330) Fiber optics communications; (060.4080) Modulation

1. Introduction

Intra-channel nonlinear distortion is an important source of signal degradation in optical

communication systems using advanced modulation formats such as quadrature amplitude modulation (QAM). Its main impact is to reduce the maximal power that can be launched into the fiber, thereby limiting optical signal-to-noise ratio (OSNR) levels at the receiver, reducing system margins and reducing the maximum propagation distance that can be achieved. In this paper, the nonlinear tolerance of an optimized pulse and a root-raised cosine (RRC) pulse is compared in terms of bit error rate (BER) performance, for varying propagation distances and launch powers. Experimental results show that the optimized pulse considerably reduces intra-channel nonlinear effects in a long haul, 10 Gbaud dual-polarization (DP) 16-QAM system relying on erbium-doped fiber amplifiers (EDFAs) and G.652 fiber (no optical dispersion compensation).

The improved nonlinear tolerance of the return-to-zero (RZ) pulse shape was recently highlighted in [1], for a 112 Gb/s DP-quadrature phase shift keying (QPSK) system without optical dispersion compensation. Comparing the RZ pulse to the non return-to-zero (NRZ) pulse, the authors reported an increase in single channel maximum propagation distance of 18%. However, the improved nonlinear tolerance achieved using the RZ pulse comes at the expense of increased spectral content. For instance, the bandwidth of a 50% RZ pulse is twice the bandwidth of the NRZ pulse or the RRC pulse with a roll-off factor (α) of 1. The use of a RZ pulse thus reduces spectral efficiency by a factor of two, and for a system using digital signal processing at the transmitter and receiver, at the Nyquist rate, it doubles the required bandwidth and sampling frequency of the digital-to-analog converters (DACs) and analog-to-digital converters (ADCs).

（a）

(b)

Figure 1. Impulse (a) and frequency (b) responses of the RRC and optimized pulses.

In comparison to the RRC pulse, the optimized pulse shape presented in [2] was shown to improve

the nonlinear performance of DP-QPSK systems without optical dispersion compensation and to increase the maximum transmission distance by as much as 22%. The specialized pulse was obtained by numerical optimization, with the primary objective formulated to reduce its width, but constrained to have a bandwidth equal to the bandwidth of a NRZ or RRC pulse (α = 1). Therefore, the optimized pulse can be used without penalizing spectral efficiency and without using higher speed DACs and ADCs. Fig. 1 exhibits the time and frequency characteristics of the optimized and RRC pulses. It can be seen that the optimized pulse is narrower than the RRC pulse with α = 1, and that its first-null bandwidth is equal to the baud rate. It is also apparent that the reduced width of the optimized pulse translates in higher energy content for frequencies between 6 GHz and 10 GHz. Through dispersion, the higher energy frequencycomponents produce a propagating pulse and a propagating waveform that is more broadened, with lower peak power excursions. Since intra-channel nonlinear distortion is proportional to peak power, the optimized pulse makes the system more tolerant to nonlinear effects. In this work and as in [2?5], the RRC pulse is used as the reference pulse for its ability to minimize pulse-induced intersymbol interference (ISI) and to minimize out-of-band power.

2. Experimental Setup

Fig. 2 illustrates the experimental setup. The symbol sources consist of random symbol sequences of length 214 and were followed by pulse shaping finite impulse response (FIR) filters. DACs with 6-bit resolution drive the in-phase-quadrature (I-Q) modulators associated with the two polarizations. The emitting wavelength was set to 1547.715 nm. A polarization scrambler (PS) was inserted after the polarization beam combiner (PBC) in order to randomly rotate the polarization state. A fiber link consisting of 15 spans of 80 km G.652 fiber was used, together with EDFAs at each span. At the receiver, a noise source was used to adjust the OSNR. The optical spectrum analyzer (OSA) served the purposes of measuring the OSNR. A coherent front-end integrates polarization beam splitters, optical hybrids, a local oscillator and photodetectors [6]. It provides four signals corresponding to the in-phase and quadrature components of the two polarizations. These baseband signals were sampled using a 50 GSa/s oscilloscope with 8-bit ADCs and post-processed in a personal computer. The linewidth of the transmitter and receiver lasers was 100 kHz, and their frequency offset was below 200 MHz. The same clock reference was used for the DACs and ADCs. The receiver signal processing functions include matched filtering and chromatic dispersion compensation (CD-1) by FIR filters, polarization recovery by a 13-tap fractionally spaced (T/2) butterfly equalizer (EQ), carrier offset removal and carrier phase recovery by a second order phase lock loop (PLL) and finally, detection. The constant modulus algorithm (CMA) is used for the equalizer coefficients pre-convergence. Once the signal is recovered, the system switches to a decision-directed modulus algorithm (DDMA) [7]. Two symbol pattern periods, totaling 215 symbols for each polarization, were used to produce BER statistics.

In what follows, the system performance is studied using either the RRC pulse or the optimized pulse presented in Fig. 1 as transmit FIR filters and receive matched filters.

Figure 2. Experimental setup.

3. Results and Discussion

Fig. 3 displays the electrical eye diagrams of the signal in-phase component taken at the transmitter, after the FIR pulse shaping filter. The inner eye openings of the optimized 16-QAM sequence eye diagram are reduced compare to those in the eye diagram obtained using the RRC pulse, indicating, as mentioned in [2], that the optimized pulse may be more sensitive to timing jitter. Fig. 3 (b) further shows that the eye diagram obtained with the optimized pulse is very similar to the eye diagram that would be obtained using an RZ pulse, and that the optimized pulse sequence returns close to a zero level between each symbol period. Since the RRC and the optimized pulses are designed to be used as matched filter, it is expected to visualize on the eye diagrams at the transmitter a certain amount of ISI. The RRC and optimized pulses produce a filtered sequence with zero pulse-induced ISI only at the receiver, after matched filtering.

Fig. 4 (a) reports the measured BER in a back-to-back configuration for a system using RRC filters and for a system using optimized filters. The linear performance of the RRC and optimized pulses is very similar and for the considered range of OSNR, the minimal and maximal penalty is 2.5 dB and 6.5 dB, respectively. Fig. 4 (b) and (c) show propagation results for 800 km and for 1200 km. At 800 km and considering a BER threshold of 8×10-3, the optimized pulse reduces the OSNR penalty by 0.6 dB and 2.8 dB, for launch powers of –2 dBm and 0 dBm, respectively. At 1200 km and for a launch power of –2 dBm an improvement of 1.2 dB is observed. At 1200 km and for a launch power of 0 dBm, the system with optimized pulse shaping almost reaches the BER threshold, while the system using RRC pulses is limited to a BER of 1.8×10-2. At this level, the optimized pulse outperforms the RRC pulse by 4.3 dB. It should be noted that the introduction of soft-decision forward error correction (FEC) [8] would be required to be able to operate at the considered BER threshold.

(a) (b)

Fig.3. Measured 16-QAM eye-diagrams at the transmitter when using: (a) the RRC filter, and (b) the optimized filter.

(a)

(b)

(c)

Figure 4. BER vs OSNR (noise bandwidth = 0.1 nm) for back-to-back (a), 800 km (b) and 1200 km (c).

For systems using multi-bit DACs operating at the Nyquist rate, the main advantage of the optimized pulse approach is that it is implemented at no extra hardware or computational cost, since the optimized pulse shaping function replaces the existing shaping filters. Yet another option would be to synthesise the optimized pulse transfer function by analog filters, thus enabling the use of simpler 2-bit DAC structures clocked at the baud rate.

4. Conclusion

The characteristics of an optimized pulse shape specifically designed to increase the nonlinear tolerance of long-haul optical transport systems were described. The specialized pulse was used in a 10 Gbaud DP-16-QAM transmission experiment and was demonstrated to significantly reduce the impact of intra-channel nonlinearities.

The authors would like to acknowledge the NSERC/Bell Canada Industrial Research Chair Program.

5. References

[1] E. Torrengo, S. Makovejs, D. S. Millar, I. Fatadin, R. I. Killey, S. J. Savory, P. Bayvel, “Influence of pulse shape in 112Gbit/s WDM PDM-QPSK transmission,” accepted to IEEE Photon. Technol. Lett., (2010).

[2] B. Chatelain, C. Laperle, D. Krause, K. Roberts, M. Chagnon, X. Xu, A. Borowiec, F. Gagnon, J. C. Cartledge, D. V. Plant, “SPM-tolerant pulse shaping for 40 Gb/s and 100 Gb/s dual-polarization QPSK systems,” accepted to IEEE Photon. Technol. Lett., (2010).

[3] M. Nakazawa, S. Okamoto, T. Omiya, K. Kasai, M. Yoshida, “256-QAM (64 Gb/s) coherent optical transmission over 160 km with an optical bandwidth of 5.4 GHz,” IEEE Photon. Technol. Lett., 22, 185–187 (2010).

[4] I. Fatadin, D. Ives, S. J. Savory, “Blind equalization and carrier phase recovery in a 16-QAM optical coherent system,” J. Lightw. Technol, 27, 3042–3049, (2009).

[5] H. Goto, K. Kasai, M. Yoshida, M. Nakazawa, “Polarization-multiplexed 1 Gsymbol/s, 128 QAM (14 Gbit/s) coherent optical transmission over 160 km using a 1.4 GHz Nyquist filter,” in Proc. OFC/NFOEC 2008. [6] K. Roberts, M. O’Sullivan, K.-T. Wu, H. Sun, A. Awadalla, D. J. Krause, C. Laperle, “Performance of dual-polarization QPSK for optical transport systems,” J. Lightw. Technol., 27, 3546–3559, (2009). [7] C. A. R. Fernandes, G. Favier, J. C. M. Mota, “Decision directed adaptive blind equalization based on the constant modulus algorithm,” Signal, Image and Video Processing, 1, 333–346, (2007).

[8] K. Onohara, Y. Miyata, T. Sugihara, K. Kubo, H. Yoshida, T. Mizuochi, “Soft decision FEC for 100G transport systems,” in Proc. OFC/NFOEC 2010.

本文来源：https://www.bwwdw.com/article/47x6.html