Open Access
Add to BibSonomyAbstract
We demonstrate 20 Gsymbol/s, 256 QAM polarization multiplexed (pol-mux) 320 Gbit/s coherent transmission. By employing an LD-based injection locking circuit, we achieved low noise optical carrier-phase locking between the LO and the data signal. Furthermore, frequency domain equalization and digital back-propagation enabled us to realize precise compensation for transmitted waveform distortions. As a result, a 320 Gbit/s data was successfully transmitted over 160 km with a potential spectral efficiency of 10.9 bit/s/Hz. This is the highest symbol rate yet achieved in a pol-mux 256 QAM coherent transmission. In addition, we also describe a pol-mux 256 QAM transmission at a symbol rate of 10 Gsymbol/s.
© 2016 Optical Society of America
1. Introduction
Spectrally efficient digital coherent transmission realized by employing a multi-level modulation such as quadrature amplitude modulation (QAM) has been intensively studied with the aim of increasing information capacity. One attractive feature of this transmission system is carrier phase recovery with digital signal processing (DSP) that does not require an analog optical carrier phase synchronization circuit [1,2]. 64~1024 QAM coherent transmissions have already been demonstrated with a digital carrier phase estimation method [3–5]. With pol-mux 256 QAM, 8 Gbaud [6] and 16 Gbaud [7] coherent transmission experiments have been demonstrated. However, as the modulation multiplicity increases, the computational complexity becomes too high, thus degrading the precision of the phase estimation. Therefore, the demodulation performance is degraded in a higher order QAM transmission.
By contrast, analog optical carrier synchronization including an optical phase-locked loop (OPLL) and injection locking enables us to realize precise carrier-phase locking independently of modulation multiplicity. By using an OPLL, we have transmitted a 3 Gsymbol/s, 2048 QAM signal over 150 km with a potential spectral efficiency (SE) of 15.3 bit/s/Hz [8]. With laser diode (LD)-based injection locking, we have successfully demonstrated an 80 Gbit/s, 5 Gsymbol/s 256 QAM-150 km transmission [9]. This scheme enabled precise optical carrier phase locking with a very simple receiver configuration.
In the work described in this paper, by applying digital frequency domain equalization (FDE) [10] and a digital back-propagation (DBP) method [11] to our LD-based injection locking transmission system, we increased the symbol rate of polarization multiplexed (pol-mux) 256 QAM transmission to 20 Gsymbol/s. 320 Gbit/s data were transmitted over 160 km within an optical bandwidth of 24.5 GHz, resulting in a potential SE of 10.9 bit/s/Hz. This is the highest symbol rate yet achieved in a 256 QAM transmission. Furthermore, we also present a 160 Gbit/s, 10 Gsymbol/s 256 QAM transmission, and compare the two sets of transmission characteristics.
2. Experimental setup for 320 Gbit/s, 256 QAM injection locking coherent optical transmission
Figure 1 shows our experimental setup for a 320 Gbit/s, 256 QAM pol-mux coherent transmission with an injection-locking homodyne receiver. The coherent CW light source for the transmitter was a 4 kHz linewidth, InP-based external cavity LD (ECLD) emitting at 1538.8 nm with an external Bragg grating on a silica planar lightwave circuit [12]. The output of the ECLD was coupled to a LiNbO3 IQ modulator with an insertion loss of 27 dB including a loss due to low modulation depth for linear modulation, and modulated with a 20 Gsymbol/s, and modulated with a 20 Gsymbol/s, 256 QAM signal and a pilot tone signal from an arbitrary waveform generator (AWG). The AWG was driven at 60 Gsample/s with 8-bit resolution (effective number of bits (ENOB): 5.6 bits at 12 GHz) and a directly generated 1 Vpp 256 QAM signal. The peak-to-average power ratio (PAPR) of the data signal was 9.3 dB. We adopted a root raised-cosine Nyquist filter with a roll-off factor of 0.2 at the AWG that enabled us to reduce the bandwidth of the QAM signal to 24 GHz. The frequency of the pilot tone signal was shifted by 12.5 GHz against the carrier frequency. At the AWG, the waveform distortions caused by individual components such as the IQ modulator and AWG were pre-equalized by using a 99-tap finite impulse response (FIR) digital filter. We obtained the optimized FIR tap coefficient at the receiver DSP by using a training sequence before the transmission experiment. The QAM data were polarization-multiplexed with a polarization division multiplexing emulator. The pilot-embedded 256 QAM signal was transmitted over two 80 km spans of ultra large area (ULA) fiber with an effective area of 153 μm2, and a dispersion of 21 ps/(nm･km). The launch power was optimized at 1 dBm. The transmission losses of each span were 15.7 and 17.1 dB. The fiber loss was compensated for by using EDFAs and Raman amplifiers. The noise figure of the EDFA was 5 dB. The Raman amplifiers provided a 9.5 dB gain with a pump power of 350 mW in each span. Figure 2 shows the optical spectra of the QAM data and pilot tone signals measured with a 0.01 nm resolution bandwidth. The transmission bandwidth including the data and pilot tone is 24.5 GHz.
Fig. 2 Optical spectra of 20 Gsymbol/s 256 QAM data and pilot tone signals (0.01 nm resolution bandwidth).
At the receiver, the transmitted signal was split into two paths after an EDFA. In one path, the pilot tone signal was extracted by an etalon filter with a 50 MHz bandwidth, and then injected into a local oscillator (LO) with an injection power of 0 dBm, where the locking range was 1 GHz [13]. We used a frequency-tunable ECLD with a linewidth of 4 kHz as an LO whose configuration was the same as that of the transmitter except that an isolator was removed. The output power of the LO was 11 dBm. The output signal of the ECLD (fLO) was frequency downshifted by 12.5 GHz with an optical frequency shifter (OFS), which was used as an LO signal for homodyne detection. The lengths of these two optical paths were adjusted within an accuracy of 1 m. The QAM and LO were homodyne-detected by using a polarization-diverse 90-degree optical hybrid and four balanced photo-detectors (B-PDs). Finally, the detected data signals were A/D-converted using a digital oscilloscope (40 Gsample/s, 16 GHz bandwidth) with an ENOB of 5.5 bits at 16 GHz and demodulated with a DSP in an offline condition. The ENOB of the A/D is defined as the following equation [14],
where SNR is the signal-to-noise ratio of the digitized data signal. An SNR of 31 dB is needed to realize a BER below 1 x 10−5 for 256 QAM, which indicates that the minimum ENOB should exceed 4.9 bits. The ENOB of our digital oscilloscope is sufficient for this experiment. In the DSP, a 12 GHz low-pass digital filter with a rectangular shape was used to eliminate the pilot tone signal. Then, the pol-mux QAM data were polarization demultiplexed with an algorithm based on the Stokes vectors [15]. We used DBP to compensate for waveform distortions induced by nonlinear effects in the optical fiber such as self-phase modulation (SPM) and cross phase modulation (XPM) between the two polarizations. In addition, chromatic dispersion was compensated for simultaneously. We employed a split-step Fourier analysis of the Manakov equation [11], where such nonlinearities and dispersion were numerically compensated for by inverse fiber propagation in the DSP. Optimum parameters were chosen for the DBP, namely the transmission loss, dispersion and nonlinear coefficient of the transmission fiber, to minimize the BER of the compensated 256 QAM signal. The step size was also optimized at 10 km. After that, we adopted FDE to compensate for residual distortions caused by hardware imperfections in the receiver. In the FDE process, we used a 256 QAM signal with a fixed pattern as a training signal, and transmitted it only once. During transmission the training signal became distorted due to the non-ideal frequency response of individual components such as the IQ modulator and B-PDs. The transfer function of the transmitted training signal Htrans(ω) was then compared with the non-distorted training signal Hideal(ω) in the frequency domain. In this process, we obtained two distortion functions Fdist(ω) = Htrans(ω)/Hideal(ω) independently for the data for the two polarizations. Thus, we were able to compensate the data signal for waveform distortion of data for both polarizations by dividing the spectrum by the distortion function. Here, the frequency resolution of FDE was 4.88 MHz with an FFT size of 8192 (4096 symbols x 2 oversampling). We also used an adaptive FIR filter to compensate for the time-varying waveform distortions of the data for each polarization by minimizing the error vector magnitude (EVM) with a decision directed least mean square (DD-LMS) algorithm. Here, the tap number was 99, which was the maximum value in our DSP. Its frequency resolution was 200 MHz. Finally, we calculated the BER from 164 kbit demodulated signals.3. Experimental results
First, we show the injection locking characteristics. Figure 3(a) shows the beat spectrum between the injection locked LO and the pilot tone as an intermediate frequency (IF) signal with a 2 MHz span. The linewidth of the IF signal was less than 10 Hz. Figure 3(b) shows the single-side band (SSB) phase noise spectrum. The phase noise variance (RMS) of the IF signal, estimated by integrating the SSB noise power spectrum from 10 Hz to 1 MHz, was only 0.2 degrees.
Fig. 3 IF spectrum of beat between pilot and injection-locked LO in (a) 2 MHz span and, (b) its SSB phase noise spectrum (10 Hz~1 MHz).
Here, we show the improved waveform distortion compensation performance that we realized by using FDE. Figures 4(a) and 4(b) show the constellation maps of 20 Gsymbol/s 256 QAM data signal under a back-to-back condition at an OSNR of 43 dB with FIR filter and with FDE, respectively. Here, the constellation map consists of continuous data with a symbol length of 20480. By employing FDE, the EVM decreased from 2.52% to 1.87%. This is due to the ability of FDE to compensate for waveform distortions with higher frequency resolution than an FIR filter.
Fig. 4 Constellations for a 20 Gsymbol/s, 256 QAM signal under a back-to-back condition when equalized with (a) an FIR filter and (b) FDE.
Figure 5 shows the optical spectra of the 20 Gsymbol/s 256 QAM data signal before and after a 160 km transmission measured with a 0.1 nm resolution. The optical signal-to-noise ratio (OSNR) of the data signal was degraded from 43 to 35 dB during the 160 km transmission.
Fig. 5 Optical spectra of 20 Gsymbol/s 256 QAM data signal before and after 160 km transmission (0.1 nm resolution bandwidth).
Figure 6(a) shows the BER characteristics of a 20 Gsymbol/s, 256 QAM transmission. The blue line shows the results under a back-to-back condition. The red line shows the transmission results. An error floor is observed under a back-to-back condition even with a high OSNR. After the transmission, a BER below the 20% forward error correction (FEC) threshold (2x10−2) was realized. Here, 320 Gbit/s data were transmitted within an optical bandwidth of 24.5 GHz, which corresponds to a potential spectral efficiency (SE) of 10.9 bit/s/Hz taking account of the 20% FEC overhead. Figure 6(b) shows the constellation for a 20 Gsymbol/s, 256 QAM signal after a 160 km transmission at an OSNR of 35 dB. The EVM was 2.74%. By applying DBP, the EVM of the transmitted 20 Gbaud 256 QAM data was improved from 2.79% to 2.74%. Its small difference may be attributed to the use of ULA fiber with a low nonlinearity.
Fig. 6 (a) BER characteristics for pol-mux, 20 Gsymbol/s, 256 QAM (320 Gbit/s)-160 km transmission, (b) constellations for 160 km transmission.
Figure 7 shows the RF spectra of demodulated 20 Gsymbol/s 256 QAM data under a back-to-back condition at the maximum OSNR. The SNR of the RF spectrum is about 28 dB. This is insufficient to achieve a BER below 1x10−4. The error floor observed in Fig. 6(a) was attributed to the insufficient SNR of the demodulated 20 Gsymbol/s 256 QAM data, which is mainly due to the insufficient SNR of the data signal generated by the AWG.
We also demonstrated a pol-mux 256 QAM transmission at 10 Gsymbol/s with the same experimental setup. Here, the data signal bandwidth was 12.6 GHz including a pilot tone. Figure 8 shows the optical spectra of a pol-mux 10 Gsymbol/s 256 QAM signal before and after transmission. The OSNR of the data was degraded from 46 to 38 dB during the 160 km transmission, where the launch power was 1 dBm.
Fig. 8 Optical spectra of 10 Gsymbol/s 256 QAM data signal before and after a 160 km transmission (0.1 nm resolution bandwidth).
Figure 9 shows the BER characteristics of a 10 Gsymbol/s 256 QAM-160 km transmission. Under a back-to-back condition, error-free demodulation was realized when the OSNR was about 38 dB. After transmission, the OSNR penalty was 4 dB at a BER of 2x10−3 (7% FEC threshold). The BERs for both sets of polarization data were below the FEC threshold. Here, 160 Gbit/s data were transmitted within an optical bandwidth of 12.6 GHz, which corresponds to a potential SE of 11.9 bit/s/Hz taking account of the 7% FEC overhead.
Figures 10(a) and 10(b) show the constellations obtained for a 10 Gsymbol/s, 256 QAM signal after back-to-back and 160 km transmissions, respectively. The EVMs are 1.35% and 2.1%. Figure 11 shows the RF spectrum of the homodyne-detected data under a back-to-back condition at the maximum OSNR. The SNR is approximately 31 dB, which is 3 dB higher than of the 20 Gsymbol/s data. This is sufficient for realizing a BER below 1x10−5.
Fig. 10 Constellations for a 10 Gsymbol/s, 256 QAM signal for (a) back-to-back, and (b) 160 km transmissions.
4. Conclusion
We successfully transmitted a 20 Gsymbol/s, 256 QAM signal over 160 km with an LD-based injection locking homodyne receiver. By using FDE and digital back-propagation, precise equalization can be achieved for waveform distortions. Thus, a 320 Gbit/s 256 QAM signal was transmitted with a potential SE of 10.9 bit/s/Hz. This is the highest symbol rate yet achieved in a pol-mux 256 QAM coherent transmission. Furthermore, a 10 Gsymbol/s, 160 Gbit/s data was successfully transmitted over 160 km with a potential SE of 11.9 bit/s/Hz.
Acknowledgments
This work is supported by “Research and Development Project toward 5G Mobile communication Systems” of the Ministry of Internal Affairs and Communications, Japan.
References and links
1. K. Kikuchi, “Coherent detection of phase-shift keying signals using digital carrier-phase estimation,” in Proceedings of the Optical Fiber Communication Conference (OFC, 2006), paper OTuI4. [CrossRef]
2. R. Noe, “PLL-free synchronous QPSK polarization multiplex/diversity receiver concept with digital I&Q baseband processing,” IEEE Photonics Technol. Lett. 17(4), 887–889 (2005). [CrossRef]
3. D. Qian, M. Huang, E. Ip, Y. Huang, Y. Shao, J. Hu, and T. Wang, “101.7-Tb/s (370×294-Gb/s) PDM-128QAM-OFDM transmission over 3×55-km SSMF using pilot-based phase noise mitigation,” in Proceedings of the Optical Fiber Communication Conference (OFC, 2011), paper PDPB5.
4. R. Schmogrow, D. Hillerkuss, S. Wolf, B. Bäuerle, M. Winter, P. Kleinow, B. Nebendahl, T. Dippon, P. C. Schindler, C. Koos, W. Freude, and J. Leuthold, “512QAM Nyquist sinc-pulse transmission at 54 Gbit/s in an optical bandwidth of 3 GHz,” Opt. Express 20(6), 6439–6447 (2012). [CrossRef] [PubMed]
5. M. Huang, D. Qian, and E. Ip, “50.53-Gb/s PDM-1024QAM-OFDM transmission using pilot-based phase noise mitigation,” in Proceedings of Optoelectronics and Communications Conference (OECC, 2011), paper PDP1.
6. R. Maher, L. Galdino, D. J. Elson, D. Lavery, A. Alvarado, and P. Bayvel, “Algorithms and reach enhancement for ultra high bandwidth transceivers,” in Proceedings of the Optical Fiber Communication Conference (OFC, 2016), paper Th3A.1. [CrossRef]
7. H. Chien, Z. Jia, and J. Yu, “256-Gb/s single-carrier PM-256QAM implementation using coordinated DD-LMS and CMA equalization,” in Proceedings of the European Conference on Optical Communications (ECOC, 2015), paper Mo.3.3.2. [CrossRef]
8. S. Beppu, K. Kasai, M. Yoshida, and M. Nakazawa, “2048 QAM (66 Gbit/s) single-carrier coherent optical transmission over 150 km with a potential SE of 15.3 bit/s/Hz,” Opt. Express 23(4), 4960–4969 (2015). [CrossRef] [PubMed]
9. K. Kasai, Y. Wang, S. Beppu, M. Yoshida, and M. Nakazawa, “80 Gbit/s, 256 QAM coherent transmission over 150 km with an injection-locked homodyne receiver,” Opt. Express 23(22), 29174–29183 (2015). [CrossRef] [PubMed]
10. K. Ishihara, T. Kobayashi, R. Kudo, Y. Takatori, A. Sano, E. Yamada, H. Masuda, and Y. Miyamoto, “Coherent optical transmission with frequency-domain equalization,” in Proceedings of the European Conference on Optical Communications (ECOC, 2008), paper We2E3. [CrossRef]
11. C. Paré, A. Villeneuve, P.-A. Bélanger, and N. J. Doran, “Compensating for dispersion and the nonlinear Kerr effect without phase conjugation,” Opt. Lett. 21(7), 459–461 (1996). [CrossRef] [PubMed]
12. L. Stolpner, S. Lee, S. Li, A. Mehnert, P. Mols, S. Siala, and J. Bush, “Low noise planar external cavity laser for interferometric fiber optic sensors,” Proc. SPIE 7004, 700457 (2008). [CrossRef]
13. Y. Wang, K. Kasai, M. Yoshida, and M. Nakazawa, “120 Gbit/s injection-locked homodyne coherent transmission of polarization-multiplexed 64 QAM signals over 150 km,” Opt. Express 22(25), 31310–31316 (2014). [CrossRef] [PubMed]
14. R. H. Walden, “Analog–to-digital converter survey and analysis,” IEEE J. Sel. Areas Comm. 17(4), 539–550 (1999). [CrossRef]
15. B. Szafraniec, B. Nebendahl, and T. Marshall, “Polarization demultiplexing in Stokes space,” Opt. Express 18(17), 17928–17939 (2010). [CrossRef] [PubMed]
