Open Access
Add to BibSonomy
Abstract
In this study, we report, to the best of our knowledge, the first experimental demonstration of the real-time reception of a 106 Gbps PAM-4 transmission over an 80 km dispersion uncompensated standard single-mode fiber (SSMF) link. In the transmitter, a chromatic dispersion (CD) pre-compensation, aided by an optical IQ modulator, is implemented. The optimization guideline of driver signal amplitudes and bias voltages is proposed to enable optimal CD pre-compensation. In the receiver, a real-time PAM-4 module including simple feed-forward equalization (FFE) is used. After the transmission, the required optical signal-to-noise ratio (OSNR) at a bit error rate (BER) below 3.8 × 10−3 is determined to be 35 dB for 106 Gbps PAM-4 signals. The better BER of 6.9 × 10−4 is achieved successfully compared to the previously reported off-line reception. The demonstration confirms the feasibility of 80 km DCI based on CD pre-compensation under real-time reception.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
With increasing Internet traffic driven by cloud computing, virtual reality, and the Internet of Things, inter-data-center interconnections (DCI) of 100 Gbps or higher in a single lane are urgently required to support transmission distances of up to 80 km. Intensity modulation direct detection (IMDD) is preferred considering the requirements on low cost and power consumption. The simple implementation and low-complexity digital signal processing (DSP) of four-level pulse amplitude modulation (PAM-4) [1] makes it a more desirable solution than discrete multitone modulation and carrier-less amplitude and phase modulation. Meanwhile, a transmission distance of 80 km or higher prefers a more favorable C-band wavelength window owing to low loss and the mature wavelength multiplexing capability to upgrade to 400 Gbps or higher. However, fiber chromatic dispersion (CD) at the C-band is a significant impairment, and optical/electrical dispersion compensation methods must be employed to enable 100 Gbps PAM-4 transmission of up to 80 km.
Equalizations/compensations in the electrical domain are preferred over dispersion compensation fibers (DCFs) in DCIs because of their simple implementation. Several digital equalization schemes are applied in the receiver to mitigate CD and consequently enable high baud rate PAM-4, including maximum likelihood sequence estimation (MLSE) [2], a Volterra filter (VF) [3], and a combination of multiple equalizers and MLSE [4]. These highly complex digital signal processing (DSP) algorithms in the receiver result in power consumption concerns. Digital CD pre-compensation is a good alternative for mitigating CD [5], with the cost of the acceptable hardware and DSP complexity in the transmitter. In the previously reported CD pre-compensation experiments, off-line reception with complex DSP algorithms is commonly employed to achieve a bit error rate (BER) below hard-decision forward error correction (HD-FEC). The dual-drive Mach-Zehnder modulator (DDMZM) aided digital dispersion compensation is implemented to enable 128 Gbps PAM-4 transmission over DCF-free links, while MLSE is required in the receiver to equalize nonlinear distortion [6]. Recently, 112 Gbps PAM-4 transmission performance over 80 km SSMF is achieved thanks to digital CD management and an off-line VF algorithm [7]. Moreover, transmission over a 300 km single-mode fiber had been demonstrated, where a novel CD pre-compensation was aided by an IQ modulator with a phase-synchronized carrier [8]. These CD pre-compensation experiments require complex off-line DSP algorithms, including MLSE&VF, which leads to unacceptable complexity. Consequently, it is necessary to consider CD pre-compensation with real-time reception to evaluate the actual performance with acceptable complexity. Besides, the reported off-line experiments have a limited BER margin below the HD-FEC threshold. This indicates that it is challenging to achieve 80 km uncompensated SSMF transmission with the commercial PAM-4 module, where only simple feed-forward equalization (FFE) is available to compensate for impairments. Therefore, it is intriguing to investigate the real-time reception performance of PAM-4 transmission with CD pre-compensation and a commercial PAM-4 receiver with a simple FFE.
In this study, the real-time reception of a single-lane 106 Gbps PAM-4 transmission over an 80 km dispersion uncompensated standard single-mode fiber (SSMF) is demonstrated experimentally. In the transmitter, an optical IQ modulator is used to implement CD pre-compensation. The influence of the parameters, including the driver signal amplitude and the bias voltage of I/Q tributaries, was investigated by both simulation and experimentation. The proposed optimization guideline of the parameters is employed to obtain optimal CD pre-compensation. In the receiver, a commercial PAM-4 application-specific integrated circuit (ASIC) module, including a simple FFE, is used to compensate for possible impairments from the fiber link and transceiver. With the optimization of both the launched optical power and the CD pre-compensation parameters, 106 Gbps PAM-4 transmission over 80 km of SSMF is successfully achieved under real-time reception with the BER of 6.9 × 10−4.
2. Experimental setup
In the experimental setup, the PAM-4 signal transmission system enabled by CD pre-compensation was implemented at two specific baud rates of 26.5625 GBd and 53.125 GBd. A pseudo-random bit sequence of word length 215-1 was mapped to generate a PAM-4 signal. After up-sampling, root-raised cosine filters of roll-off 1 and 0.1 were used for pulse shaping under two specific baud rates. Considering that the receiver used square-law detection, the square root operation was applied to the transmitter. This nonlinear operation leads to the significant out-of-band components, as shown in the inset (c) of Fig. 1. The impact of the square root operation on the sampling rate and DAC bandwidth requirements is analyzed. The sixth-order Bessel filter is used for modeling the DAC bandwidth. The OSNR penalty as a function of DAC bandwidth is plotted in the inset (d) with 2 or 4 samples per symbol (sps). With the same bandwidth of 33 GHz in the experiments, the penalty of around 0.75 dB is introduced by comparing 2-sps to 4-sps. Meanwhile, around 2.7 dB penalty is induced by the DAC bandwidth limitation. The digital CD pre-compensation was implemented in the frequency domain, and the pre-compensated PAM-4 signal can be expressed as
Fig. 1. Experimental setup and DSP algorithms in the transceiver. The histograms of upre_I and upre_Q are shown in the insets (a) and (b), respectively. Inset (c): the spectrum before and after square root operation; Inset (d): osnr sensitivity vs. DAC bandwidth
The optically modulated signal was then amplified by an Erbium-doped fiber amplifier (EDFA) and then launched into an 80 km SSMF fiber link without DCF. In the receiver, another EDFA was utilized as a noise source to tune the optical signal-to-noise ratio (OSNR). A bandpass optical filter with a 3dB bandwidth of 100 GHz was used to remove out-of-band noise. The filtered signal was firstly sent to the detection module (MACOM ROSA, BPS56+MATA5819) for optical-to-electrical conversion and subsequently was equalized and decoded by a real-time receiver (MACOM, PRISM-50) which includes a timing-recovery module and a 15-taps T-spaced FFE. The receiver has a sampling rate of 53.125 GHz and a 3 dB bandwidth of around 20Ghz. Finally, the BER was obtained by error counting in real-time. Owing to the simple algorithms, the receiver module had low power consumption, which is crucial for DCI applications. In addition to real-time reception, the detected signal was also sampled by a 160 GSa/s oscilloscope (Keysight, DSAZ634A) for off-line processing, in which the same algorithms as in the real-time module were used.
3. Results and analysis
Two important operating parameters, namely the driver signal amplitude and bias voltage, need to be configured accurately to optimize CD pre-compensation. The histograms of the normalized I/Q components of the pre-compensated PAM-4 signal with an 80 km SSMF link were simulated, as shown in the insets of Fig. 1. For the Q tributary, the bias voltage should be set at the null point of the transmission curve of the modulator. For the I tributary, there was a prominent DC component due to the feature of PAM-4, which is dependent on the chromatic dispersion value of the fiber link. The AWG only outputs the AC component of I tributary for the following electrical amplifier. The bias voltage of I tributary should be tuned properly to compensate for the DC component of uQ. The rest bias voltage in the IQ modulator is configured to have pi/2 phase shift to combine I and Q tributaries to generate CD pre-compensated optical signal.
The influence of the bias voltages of the I/Q tributaries (represented by Bias I and Bias Q, respectively) on the OSNR penalty was investigated by numerical simulation. The 28 GBd PAM-4 system with the same DSP algorithms in the experimentation was also studied via numerical simulation. For the Q tributary, the optimal bias voltage is approximately ${V_\pi }$, as shown in Figs. 2(a) and 2(b). This result is consistent with the aforementioned analysis. For the I tributary, the optimal bias voltage is approximately 0.65${V_\pi }$ in the back-to-back (BTB) case. With the pre-compensation of the 80 km link, the optimal bias voltage is approximately 0.75${V_\pi }$. The optimal voltages between the BTB and transmission cases were different, which can be attributed to the difference in the probability distribution of the signal level. With these optimal bias voltages, the OSNR sensitivity, as a function of the driver signal amplitude, of the I/Q tributaries is presented, as shown in Fig. 2(c). The optimal amplitudes in the BTB and transmission cases were 0.72 ${V_\pi }$ and 0.82 ${V_\pi }$, respectively. The digital pre-distortion was less effective for the larger or smaller amplitudes, which subsequently degraded the CD pre-compensation performance. In addition, for a smaller amplitude, the unwanted carrier was increased in the PAM signal, which also diminished the receiver sensitivity. Figure 2(d) shows that the joint optimization of bias and I/Q amplitude is crucial to obtain the optimal transmission performance enabled by CD pre-compensation.
Fig. 2. The contour plots of OSNR penalty as a function of Bias I and Bias Q for BTB case (a) and 80 km transmission case (b). (c)The OSNR penalty as a function of I/Q amplitude. (d)The contour plots of OSNR penalty as a function of Bias I and I/Q amplitude
Table 1 compares the simulation and experimental results regarding the optimization of the amplitude and bias voltage. It is challenging to measure the absolute amplitude experimentally; therefore, the ratio of the optimal amplitudes between the BTB and transmission cases was used for comparison. The amplitude ratios obtained during the simulation and experimentation were 0.82/0.7, and 120/100, respectively. The difference between these two ratios was less than 3%. Considering the bias drift in the Mach-Zehnder modulator (MZM), it is preferable to evaluate the difference between the applied and null transmission voltages. In the simulation, the bias voltage had optimal values of 0.643${V_\pi }$ and 0.745${V_\pi }$ in the BTB and transmission cases (Fig. 2), respectively. Consequently, the differences between the applied and null transmission voltages were 0.357${V_\pi }$ and 0.255${V_\pi }$ in the BTB and transmission cases, respectively. In the experiment, the bias voltages at the BTB and transmission cases were optimized at 9 V and 10.2 V, respectively. The differences were calculated by subtracting the bias voltages from the null transmission voltage of 12.6 V and then normalized by ${V_\pi }\; $ of 8.5 V. The deviation of the optimal bias voltages between simulation and experimentation was less than 7% and 3% for the BTB and transmission cases, respectively. Therefore, the numerical simulation agreed with the experimentation very well. These optimization guidelines of amplitude and bias voltage are essential for effective and optimal CD pre-compensation and pre-distortion implementation.
Table 1. Comparison of optimal amplitude and bias voltage of simulation and experimental results.
The eye diagrams of the BTB and transmission cases at two baud rates are presented in Fig. 3. At a baud rate of 26.5625 GBd, the eye diagrams of the two cases are clearly open. At 53.125 GBd (Fig. 3(c) and 3(d)), the eye openings were severely degraded. Specifically, the BTB eye diagrams had noticeable distortion, which can be attributed to the possible bandwidth limitation of the devices or imperfect pre-emphasis. After 80 km of SSMF transmission, the eye diagrams were further degraded by possible residual dispersion and limited equalization capability of the receiver.
The BER as a function of the OSNR at two baud rates was experimentally studied with real-time reception and off-line processing. The BER at a baud rate of 26.5625 GBd is shown in Fig. 4(a). With real-time reception, the OSNR sensitivity of 25.9 dB and 28.5 dB at the HD-FEC threshold are obtained for BTB and 80 km transmission, respectively. The off-line processing (red lines) obtains better BER than real-time reception, which can be attributed to a higher bit resolution of the analog-to-digital converter and better computation accuracy in the DSP algorithms.
At a baud rate of 53.125 GBd, the OSNR sensitivities of 33.5 dB and 35 dB were obtained in the BTB and transmission cases, respectively. After PAM-4 transmission over an 80 km SMF enabled by CD pre-compensation, the lowest BER of 6.90 × 10−4 was obtained with real-time reception, which outperforms the previous off-line reception [7]. As shown in Fig. 4(b), after 80 km transmission, real-time and off-line reception has comparable performance. Under the low OSNR region, off-line reception has slightly worse BER. These results may be attributed to a long RF cable (about 30 cm) used to connect the optical receiver and oscilloscope in the off-line case. The cable introduces additional impairments at 53 GBd and consequently leads to BER degradation.
4. Conclusions
The real-time reception of a 106 Gbps PAM-4 signal over 80 km of SSMF transmission without DCF was experimentally demonstrated for the first time, to the best of our knowledge. The influence of the bias and amplitude of the driver signal in the transmitter on the CD pre-compensation was investigated in detail. The simulation results indicate that joint optimization of the amplitudes and bias voltages of the I/Q component of the pre-compensated signal is crucial for achieving optimal CD pre-compensation. Regarding the optimal amplitudes and bias voltages, the experimental results were consistent with the simulation prediction. With the commercial real-time DSP module including simple FFE equalization, 106 Gbps PAM-4 transmission over an 80 km SSMF link is demonstrated successfully with the better BER of 6.9 × 10−4 compared to the previous off-line experiment. This achievement can be mainly attributed to the optimal CD pre-compensation aided by joint optimization of bias and amplitude. If the automatic control of the amplitudes and bias voltages is utilized, the BER performance can be improved further. Finally, the feasibility of the CD pre-compensation route with real-time reception is confirmed by our demonstration successfully, which can encourage the industry to consider CD pre-compensation enabled 80 km DCI with the aid of the mature PAM-4 modules.
Funding
Shenzhen Municipal Science and Technology Innovation Council (JCYJ20190806142407195).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. K. Zhong, X. Zhou, T. Gui, L. Tao, Y. Gao, W. Chen, J. Man, L. Zeng, A. P. T. Lau, and C. Lu, “Experimental study of PAM-4, CAP-16, and DMT for 100 Gbps Short Reach Optical Transmission Systems,” Opt. Express 23(2), 1176–1189 (2015). [CrossRef]
2. H. Wang, J. Zhou, D. Guo, Y. Feng, W. Liu, C. Yu, and Z. Li, “Adaptive Channel-Matched Detection for C-Band 64-Gbit/s Optical OOK System over 100-km Dispersion-Uncompensated Link,” J. Lightwave Technol. 38(18), 5048–5055 (2020). [CrossRef]
3. H. Xin, K. Zhang, L. Li, H. He, and W. Hu, “50 Gbps PAM-4 Over Up to 80-km Transmission With C-Band DML Enabled by Post-Equalizer,” IEEE Photonics Technol. Lett. 32(11), 643–646 (2020). [CrossRef]
4. J. Zhou, H. Wang, L. Liu, C. Yu, Y. Feng, S. Gao, W. Liu, and Z. Li, “C-band 56 Gbit/s on/off keying system over a 100 km dispersion-uncompensated link using only receiver-side digital signal processing,” Opt. Lett. 45(3), 758–761 (2020). [CrossRef]
5. R. I. Killey, P. M. Watts, V. Mikhailov, M. Glick, and P. Bayvel, “Electronic dispersion compensation by signal predistortion using digital processing and a dual-drive Mach-Zehnder Modulator,” IEEE Photonics Technol. Lett. 17(3), 714–716 (2005). [CrossRef]
6. Q. Zhang, N. Stojanovic, C. Xie, C. Prodaniuc, and P. Laskowski, “Transmission of single lane 128 Gbit/s PAM-4 signals over an 80 km SSMF link, enabled by DDMZM aided dispersion pre-compensation,” Opt. Express 24(21), 24580–24591 (2016). [CrossRef]
7. S. Fu, C. Chen, F. Gao, X. Li, L. Deng, M. Tang, and D. Liu, “Digital chromatic dispersion pre-management enabled single-lane 112 Gbps PAM-4 signal transmission over 80 km SSMF,” Opt. Lett. 43(7), 1495–1498 (2018). [CrossRef]
8. Z. Liu, T. Xu, G. Saavedra, and P. Bayvel, “448-Gbps PAM4 Transmission over 300-km SMF-28 Without Dispersion Compensation Fiber,” in Optical Fiber Commununication Conference (Optical Society of America, San Diego, CA, 2018), p. W1J.6 [CrossRef] .
9. Q. Zhang, N. Stojanovic, J. Wei, and C. Xie, “Single-lane 180 Gbps DB-PAM-4-signal transmission over an 80 km DCF-free SSMF link,” Opt. Lett. 42(4), 883–886 (2017). [CrossRef]
