Abstract
A novel automatic bias control (ABC) method for optical in-phase and quadrature (IQ) modulator is proposed and experimentally demonstrated. In the proposed method, two different low frequency sine wave dither signals are generated and added on to the I/Q bias signal respectively. Instead of power monitoring of the harmonics of the dither signal, dither-correlation detection is proposed and used to adjust the bias voltages of the optical IQ modulator. By this way, not only frequency spectral analysis isn’t required but also the directional bias adjustment could be realized, resulting in the decrease of algorithm complexity and the growth of convergence rate of ABC algorithm. The results show that the sensitivity of the proposed ABC method outperforms that of the traditional dither frequency monitoring method. Moreover, the proposed ABC method is proved to be modulation-format-free, and the transmission penalty caused by this method for both 10 Gb/s optical QPSK and 17.9 Gb/s optical 16QAM-OFDM signal transmission are negligible in our experiment.
© 2017 Optical Society of America
1. Introduction
With the development of optical fiber communication technology, coherent optical communication systems using advanced modulation format have attracted great attention due to the extremely large transmission capacity [1, 2]. In coherent optical communication systems, the optical in-phase and quadrature (IQ) modulators are often deployed as the electrical-to-optical up-convertors [3, 4]. A single-polarization optical IQ modulator needs at least three bias signals to control its two child Mach-Zehnder modulator (MZM) and one parent MZM [5, 6]. However, the stability of the optical IQ modulator is easily influenced by some environment factors, such as temperature and mechanical vibration even the bias voltages remain unchanged [7, 8]. The drifting bias conditions will significantly degrade the transmission performance. Therefore, the auto bias control (ABC) module is essential to achieve long-term stability. Several ABC approaches have been proposed recently, including optical power monitoring techniques [9, 10] and dither signal monitoring techniques [11–14]. For instance, a dither-free method by monitoring optical power and analyzing the variance of optical power has been reported in [9], and a dither-added way of monitoring the electrical power of RF signal and dither signal respectively has also been proposed in [10]. For dither signal monitoring method, the power spectrum of 1st and 2rd dither harmonic frequency signals are analyzed and calculated carefully by fast Fourier transform (FFT) operation [11–14]. However, to meet the requirement of higher speed, longer transmission distance and more complex modulation format in current coherent optical communication system, auto bias control method with modulation format free, higher sensitivity and lower complexity is highly desired.
In this paper, we experimentally demonstrate an intelligent and effective ABC approach for optical IQ modulator. In the proposed scheme, the optimal bias conditions of two child MZMs and one parent MZM are achieved by using two different low frequency sine wave dither signals and dither-correlation detection. The designed dither-correlation detection doesn’t need frequency spectral analysis which is essential in the conventional dither signal based ABC method, and the results show that directional bias adjustment could be easily realized, resulting in the lower algorithm complexity and faster convergence rate. Moreover, the proposed ABC scheme is proved to have higher sensitivity and be modulation-format-free. For demonstration and evaluation purpose, we implement the proposed ABC method in both 10 Gb/s optical single carrier QPSK and 17.9 Gb/s optical 16QAM-OFDM signal transmission systems, and the experimental results show that the transmission penalty caused by ABC method is negligible.
2. Principle of the proposed ABC method
Figure 1 shows the proposed auto bias control configuration based on dither-correlation technique. To monitor and control the biases, a low bandwidth photo-detector (PD) is used to detect a small proportion (about 10%) of the optical signal power. A low speed (25 kHz) analog-to-digital converter (ADC) is adopted to sample the output signal of the PD. Two generated sine waves with the frequency of 3.9 kHz and 4.9 kHz are applied to dither the bias signal of the two child MZMs (BiasI and BiasQ) respectively. Meanwhile, these two dither signals are sampled by other ADCs for signal processing. A microprocessor unit (MPU) is then used to monitor both the average optical signal power and the correlation integration of the transmitted dither signals and the detected dither signals for auto bias control. Moreover, due to the use of ultra-low bandwidth PD, the wide-band RF signals (Signal I and Signal Q) are considered as white Gaussian noise.
The output optical signal can be written as
where p represents the phase difference between the two branches of the parent MZM. is the half-wave voltage of the two-child MZMs at the bias port. and are the bias voltages when the bias of two-child MZMs are set to be at their peak transmission point, and and are the equivalent bias voltages for these two-child MZMs respectively. and are the dither signals which could be denoted aswhere the phase and amplitude differences between these two dither signals are ignored for simplicity, and and are the dither signals with the frequency of 3.9 kHz and 4.9 kHz respectively. Subsequently, the output optical power at the PD could be given byAround the optimum bias point (, and ), Eq. (3) can be rewritten as
where and are the errors of and respectively. It could be clearly observed in Eq. (4) that if is set to its optimum value which means , the electrical power of the dither fundamental frequency () in is minimum. Similarly, if is set to its optimum value, the electrical power of the dither fundamental frequency () in is minimum. Moreover, if is set to its optimum value which means , the electrical power of the dither inter-modulation frequency () are minimum. These feature are widely used in the conventional ABC methods based on frequency spectrum analysis [14]. In these methods, FFT based frequency spectrum analysis is adopted to calculate the power of the fundamental harmonic component and the inter-modulation frequency component. However, the bias adjustment around the optimum point are obviously non-directional and non-linear in these methods. Figures 2(a) and 2(b) show the electrical power of dither inter-modulation frequency () and low frequency RF signal curves in the conventional dither signal monitoring [14] and dither-free ABC scheme [15] respectively. In this simulation, the bias voltage of the parent MZM is scanned from to , where is the optimum bias value of the parent MZM, so the phase is varied from to , and the electrical power of dither inter-modulation frequency is normalized by the electrical power of when .To address these issues, an effective ABC approach based on dither-correlation detection is proposed in our scheme. The proposed correlation integration operations can be written as
where and are the initial phase of these two dither signals respectively. is the acquisition time, normally is set to be about 100 times of the dither signal’s period to reduce the effect of noise. , and are the correlation integral coefficients which are related to the bias states of two-child MZMs and one parent MZM respectively. According to the statistical theory, () can represent the correlation degree between original dither signals and the photo-detected dither signals. As shown in Eq. (4), when the bias of parent MZM is set to the optimum value, the dither inter-modulation frequency () are minimum, so and are uncorrelated and . Figure 2(c) shows the curve around the optimum point, and in this simulation the value is also normalized by the when . Obviously, this curve is linear and monotonous, and the zero point corresponds to the optimum bias point of the parent MZM. Therefore, directional bias adjustment could be easily realized in our scheme. Actually, in the proposed method, only two bias voltage values are required to achieve the optimum point, resulting in the faster convergence rate. However, many bias voltage values need to be scanned and calculated to find out the optimum point in the conventional dither frequency monitoring method.Considering the orthogonality of trigonometric function, Eq. (5) could be simplified as
where and (n = 1,2,3) are constant values which are affected by the parameters of the dither signals, such as , and .As we all known, the optimum bias point of an optical IQ modulator means that the bias of two child MZMs are set to be at their null transmission point ( and ), and the bias of the parent MZM is set to be at its point. As shown in Eq. (6), it is easy to find out the optimum bias point of the parent MZM just by scanning to make. However, doesn’t mean two-child MZMs are at their optimum bias points due to the influence of . To solve this problem, the bias of parent MZM should be adjusted preferentially. Assuming the bias of parent MZM is set to be around the optimum point (), Eq. (6) could be rewritten as
Figures 3(a) and 3(b) show the simulated correlation integral coefficient curves of one child MZM and the parent MZM versus bias voltage. It could be clearly observed that, when or is equal to zero, and is equal to or 0. and correspond to the null transmission point (which is the optimum bias point), and correspond to the peak transmission point. When is equal to zero, , and both of these two points are the optimum bias points. Therefore, the optimum bias point of the parent MZM could be found out by scanning the bias voltage and monitoring . As same as the averaged optical power curves in the traditional ABC method, it could be clearly shown in Fig. 4(a) that the () curves in the proposed scheme will also be reversed when the modulation index is really high such as in large-driver single carrier QPSK (SC-QPSK) application [16]. However, the zero points of these curves are not changed with different modulation format and modulation index as shown in Figs. 4(a) and 4(b), and the curves are not reversed with different modulation format and modulation index [17, 18] as shown in Figs. 4(c) and 4(d). To avoid this reversion, the modulation index is set to relative low value by turning off RF signal or reducing the gain of RF driver amplifier [16] before the ABC operation. Subsequently, to distinguish the null and the peak transmission point of two child MZMs, a coarse adjustment process [14, 15] which could adjust the bias voltages to around the optimum point () is adopted in our scheme. In this coarse adjustment, the average optical power at the PD is reduced as much as possible by adjusting BiasI and BiasQ in turn. After that, the gain of the RF driver amplifier could be set to the normal value.
As mentioned above, the proposed ABC algorithm could be performed with three steps. Firstly, the RF signal is disabled, and the bias voltage of two child MZMs are scanned in turn to find out the minimum average optical power at the input of PD. In the second step, the RF signal is enabled, and the bias voltage of parent MZM is scanned to decrease . Finally, the bias voltage of two child MZMs and parent MZM are all adjusted to decrease , and as much as possible by an iterative algorithm. To stabilize the bias of optical IQ modulator at the optimum point, an iterative algorithm based on Newton-iterative method is proposed. For example, assuming that is the slope of the curve near the optimum point, the th iteration of can be written as
where the is the correlation integral coefficient corresponding to , and is calculated in real time to eliminate the interference caused by modulation format or modulation index changing. , and could be calculated by numerical integration, which are defined aswhere is the sampling points, and . According to Eq. (9), higher sampling frequency means smaller and more samples can be acquired in every dither signal period, so more information of dither signals can be acquired by increasing the sampling frequency. Figure 5 shows the slope of curve versus different sampling frequency, and larger slope coefficient means better sensitivity.It can be easily obtained from Eq. (9) that the computation complexity of correlation integral operation is . However, in the conventional dither signal monitoring based ABC methods, FFT based frequency spectrum analysis is adopted and the complexity of FFT is . Obviously, the proposed scheme has the lower algorithm complexity.
Figure 6 shows the monitored signals as a function of errors of and around the optimum bias point by using dither-free based ABC approach, dither signal monitoring techniques and the proposed ABC method. In all cases, the monitored signal are normalized to the value when we get the biggest error of or , and the value of monitored signal in this simulation can be defined as distinguish ratio. Moreover, in our simulation, SC-16QAM RF signal with and white Gaussian noise is used. It could be clearly observed that the proposed ABC method and the FFT-based dither signal monitoring techniques have much higher sensitivity than the dither-free based ABC approach. However, the proposed ABC method outperform the dither-free based ABC approach in the bias control of the parent MZM, especially when the amplitude of dither signals is lower as shown in Fig. 6 [12].
3. Experimental setup and results
Figure 7 shows the experimental setup of coherent optical back-to-back transmission system to verify the proposed ABC scheme. In our experiment, both single carrier modulation and multicarrier modulation are evaluated. For single carrier modulation, 5 Gbaud QPSK signals are produced by an arbitrary waveform generator (AWG, TekAWG7122C), and the net rate of the generated QPSK signal is 10 Gb/s. For multicarrier modulation, 16QAM-OFDM signal is produced by this AWG operated at 10 GSa/s. The 16QAM-OFDM baseband signal consists of 138 subcarriers, and 10 subcarriers are used to estimate the phase noise. The used inverse fast Fourier transform (IFFT) length is 256, and the cyclic prefix is 10% of the IFFT length. The length of a frame is set to 139, and in each frame 2 training symbols are inserted to implement time synchronization and channel estimation. Thus, the net rate of the generated 16QAM-OFDM signal is 17.9 Gb/s (). The photograph of the used ABC module which is designed by ourselves are shown in the inset of Fig. 7. This ABC module consists of MPU, ADC and DAC chips, and the frequency of the dither sine wave signals are set to 3.9 kHz and 4.9 kHz in the experiment. The generated baseband signal is used to modulate a 100 kHz-linewidth continuous-wave from an external cavity laser in an optical IQ modulator (Fujitsu FTM7962EP) with a half-wave voltage of 9V at the bias port. At the receiver, the coherent detector and 100 GSa/s digital sampling oscillator (DSO, Tektronix DSA72504DX) is used to capture the QPSK or 16QAM-OFDM signals. At last, offline signal demodulation is performed by a DSP-based receiver. In our experiment, 99520 bits in QPSK case and 267264 bits in 16QAM-OFDM case are calculated for BER counting.
Figures 8(a)-8(c) show the measured correlation integral coefficient versus different bias voltages in the experiment. In the test, each correlation integral coefficient curve is measured when two other bias voltage is set to the optimum point value. It could be clearly observed that, the measured curves are accordance with the Eq. (7) and Fig. 3. The measured half-wave voltage of the child MZM and the parent MZM are not strictly equal, and the deviation is about 0.5V. Figure 9(a) presents the measured BER performance of 17.9 Gb/s optical 16QAM-OFDM signal as a function of optical signal noise rate (OSNR) when different modulation depth (MD) of dither signal is used, and the results by using manual bias control method are also plotted for comparison. Negligible power penalty is observed between the proposed ABC method and manual bias control method. The MD of dither signal is defined by , where is the amplitude of the dither signal. Obviously, MD has significant influence on auto bias control. Figure 9(b) shows the measured curves as a function of the deviation of the current bias voltage and the optimum bias voltage when different MD are used, and Fig. 10 shows the distinguish ratio curves of in different MD value. It could be clearly observed that smaller MD lead to higher sensitivity and better ABC performance in child MZM’s adjusting. However, too small MD will significantly reduce the distinguish ratio of as shown in Fig. 6(b), so that MD is set to be 6.6% in our experiment.
Figure 11 shows the measured BER performance of both 10 Gb/s optical QPSK signal and 17.9 Gb/s optical 16QAM-OFDM signal as a function of OSNR when the proposed ABC scheme and the manual bias control method are used. MD is set to in the test. It could be clearly observed that the proposed ABC scheme is modulation format free and has no power penalty compared to the manual bias control method. The constellation of the received optical QPSK signal and optical 16QAM-OFDM signal are both plotted in the insets of Fig. 11. Furthermore, to evaluate the long-term and stable performance of the proposed scheme, both the error vector magnitude (EVM) performance of 17.9 Gb/s optical 16QAM-OFDM signal and the corresponding environment temperature of the optical IQ modulator are measured and shown in Fig. 12. The OSNR is maintained as 20.69 dB, and a hot air blower is used to vary the temperature of the optical IQ modulator indirectly. In the test, not only the temperature is varied but also mechanical vibration is induced by the wind, and this mechanical vibration will cause optical signal polarization jitter, resulting in the fluctuation of the measured EVM performance. Therefore, it could be clearly observed that when the air blower is set to strong wind mode (between 5th and 15th minutes), EVM performance become litter worse in ABC-on case. However, when the temperature changes within 20 to 36°C, the measured EVM performance fluctuates wildly if no ABC is adopted, and the measured EVM performance remains basically below 10% when the proposed ABC method is used.
4. Conclusions
In this paper, we propose a novel ABC method for coherent optical communication system based on two different low frequency sine wave dither signals and dither-correlation detection. The theoretical analysis shows that no FFT-based frequency spectral analysis is needed, and the directional bias adjustment could be easily realized, resulting in the lower algorithm complexity and faster convergence rate. This ABC scheme is proved to have other advantages such as higher sensitivity and modulation format independence. The experimental results show that the power penalty caused by the proposed ABC is negligible for both single carrier modulation and optical orthogonal frequency division multiplexing modulation, which make the proposed scheme very suitable for high speed coherent optical communication systems.
Funding
National “863” Program of China (No. 2015AA016904); National Natural Science Foundation of China (NSFC) (Nos. 61675083, 61307091, 61575071 and 61331010).
References and links
1. Y. Wang, K. Kasai, M. Yoshida, and M. Nakazawa, “320 Gbit/s, 20 Gsymbol/s 256 QAM coherent transmission over 160 km by using injection-locked local oscillator,” Opt. Express 24(19), 22088–22096 (2016). [CrossRef] [PubMed]
2. 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]
3. G. W. Lu, T. Sakamoto, and T. Kawanishi, “Flexible high-order QAM transmitter using tandem IQ modulators for generating 16/32/36/64-QAM with balanced complexity in electronics and optics,” Opt. Express 21(5), 6213–6223 (2013). [CrossRef] [PubMed]
4. S. Yan, X. Weng, Y. Gao, C. Lu, A. P. T. Lau, Y. Ji, L. Liu, and X. Xu, “Generation of square or hexagonal 16-QAM signals using a dual-drive IQ modulator driven by binary signals,” Opt. Express 20(27), 29023–29034 (2012). [CrossRef] [PubMed]
5. L. Tao, J. Yu, and N. Chi, “Generation of flat and stable multi-carriers based on only integrated IQ modulator and its implementation for 112Gb/s PM-QPSK transmitter,” in Optical Fiber Communication Conference (Optical Society of America, 2012), paper JW2A.86. [CrossRef]
6. F. A. Gutiérrez, P. Perry, F. Smyth, A. D. Ellis, and L. P. Barry, “Optimum bias point in broadband subcarrier multiplexing with optical IQ modulators,” J. Lightwave Technol. 33(1), 258–266 (2015). [CrossRef]
7. J. Švarný, “Analysis of quadrature bias-point drift of Mach-Zehnder electro-optic modulator,” in Proceedings of IEEE Biennial Baltic Electronics Conference (IEEE, 2010), pp. 231–234. [CrossRef]
8. H. Nagata, Y. Li, D. R. Maack, and W. R. Bosenberg, “Reliability estimation from zero-failure LiNbO3 modulator bias drift data,” IEEE Photonics Technol. Lett. 16(6), 1477–1479 (2004). [CrossRef]
9. P. S. Cho and M. Nazarathy, “Bias control for optical OFDM transmitters,” IEEE Photonics Technol. Lett. 22(14), 1030–1032 (2010). [CrossRef]
10. H. S. Chung, S. H. Chang, J. C. Lee, J. H. Lee, and K. Kim, “Field experiment of 112 Gb/s dual-carrier DQPSK signal transmission with automatic bias control of optical IQ modulator,” in Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2013 (Optical Society of America, 2013), paper NW4E.5. [CrossRef]
11. T. Sakamoto, I. Morohashi, and T. Kawanishi, “Mach-Zehnder-modulator-based flat comb generator with auto bias control,” in Proceedings of IEEE International Meeting on Microwave Photonics jointly held with the 2008 Asia-Pacific Microwave Photonics Conference(IEEE2008), pp. 154–157. [CrossRef]
12. X. Zhu, Z. Zheng, C. Zhang, L. Zhu, Z. Tao, and Z. Chen, “Coherent detection-based automatic bias control of Mach-Zehnder modulators for various modulation formats,” J. Lightwave Technol. 32(14), 2502–2509 (2014). [CrossRef]
13. X. Zhang, Y. Wang, X. Xiao, C. Li, C. Li, Z. Li, Q. Yang, and S. Yu, “Real-time bias control for optical OFDM transmitter,” in Asia Communications and Photonics Conference 2013 (Optical Society of America, 2013), paper AF1E.7. [CrossRef]
14. T. Gui, C. Li, Q. Yang, X. Xiao, L. Meng, C. Li, X. Yi, C. Jin, and Z. Li, “Auto bias control technique for optical OFDM transmitter with bias dithering,” Opt. Express 21(5), 5833–5841 (2013). [CrossRef] [PubMed]
15. P. S. Cho, J. B. Khurgin, and I. Shpantzer, “Closed-loop bias control of optical quadrature modulator,” IEEE Photonics Technol. Lett. 18(21), 2209–2211 (2006). [CrossRef]
16. T. Yoshida, T. Sugihara, K. Sawada, K. Uto, and K. Shimizu, “Automatic Bias Control for Arbitrary Optical Signal Generation by Dual-Parallel MZM,” in Proceedings of OptoElectronics and Communications Conference (OECC2010), paper 8B2–5.
17. H. Kawakami, T. Kobayashi, E. Yoshida, and Y. Miyamoto, “Auto bias control technique for optical 16-QAM transmitter with asymmetric bias dithering,” Opt. Express 19(26), B308–B312 (2011). [CrossRef] [PubMed]
18. H. Kawakami, T. Kobayashi, M. Yoshida, T. Kataoka, and Y. Miyamoto, “Auto bias control and bias hold circuit for IQ-modulator in flexible optical QAM transmitter with Nyquist filtering,” Opt. Express 22(23), 28163–28168 (2014). [CrossRef] [PubMed]