Relative to homodyne coherent detection, heterodyne coherent detection has simple architecture because no 90° hybrid and only half number of photodiodes and analog-to-digital convertor (ADC) chips are required. We experimentally demonstrate that the architecture of heterodyne coherent receivers can be further simplified. When the frequency offset is one half of the channel frequency spacing, one local oscillator (LO) laser can be used for two neighboring wavelength-division-multiplexing (WDM) channels, and therefore the number of LO lasers can be reduced into half compared to homodyne detection. We experimentally demonstrate simplified heterodyne coherent detection of 4 × 196.8-Gb/s polarization-division-multiplexing carrier-suppressed return-to-zero quadrature-phase-shift-keying (PDM-CSRZ-QPSK) modulation after transmission over 1040-km single-mode fiber (SMF)-28 on a 50-GHz grid with bit-error ratio (BER) smaller than pre-forward-error-correction (pre-FEC) limit of 3.8 × 10−3. To our best knowledge, 196.8 Gb/s is the highest bit rate per channel for heterodyne coherent WDM transmission system. An arrayed waveguide grating (AWG) instead of wavelength selective switch (WSS) is used at the transmitter to spectrally shape and multiplex the WDM signal. We also experimentally demonstrate that heterodyne detection causes 3-dB optical signal-to-noise ratio (OSNR) penalty at the BER of 3.8 × 10−3 for a certain single channel compared to homodyne detection.
©2012 Optical Society of America
With the development of large-bandwidth and high-speed electronic analog-to-digital converters (ADCs) and photo detectors (PDs), very recently, coherent detection with digital signal processing (DSP) has been attracting a great deal of interest in research community once again [1–5]. It’s well known that coherent detection includes homodyne detection and heterodyne detection. Compared to homodyne detection, by simultaneously down-converting in-phase (I) and quadrature (Q) components to the intermediate frequency (IF), not only can heterodyne detection reduce the number of balanced PDs and ADCs of coherent receiver into half , but also there is no need to consider the delay between I and Q components in the polarization-division-multiplexing (PDM) signal. The conventional dual-hybrid structure also becomes unnecessary, and therefore heterodyne detection is much more hardware-efficient than homodyne detection. We have experimentally demonstrated a high-speed simplified coherent receiver with heterodyne detection of 8 × 50-Gb/s PDM quadrature-phase-shift-keying (PDM-QPSK) wavelength-division-multiplexing (WDM) after 1040-km single-mode fiber (SMF)-28 transmission, based on IF down conversion in digital frequency domain .
In the paper, we will demonstrate that the architecture of heterodyne coherent detection can be further simplified. In our proposed architecture, one local oscillator (LO) laser can be used for two neighboring WDM channels, and therefore the number of LO lasers can be reduced into half compared to homodyne detection. We will also experimentally demonstrate 4 × 196.8-Gb/s ultra-density WDM transmission over 1040-km SMF-28 on a 50-GHz grid adopting PDM carrier-suppressed return-to-zero QPSK (PDM-CSRZ-QPSK) modulation and heterodyne coherent detection. To our knowledge, this is the highest bit rate per channel for heterodyne coherent detection. In our experiment, an arrayed waveguide grating (AWG) instead of wavelength selective switch (WSS) is used at the transmitter to spectrally shape and multiplex the WDM signal. A digital post filter combined with 1-bit maximum likelihood sequence estimation (MLSE) is adopted at the receiver after carrier phase estimation (CPE) in the conventional DSP process, in order to overcome the severe filtering effect and crosstalk caused by the AWG and ADC bandwidth.
2. Principle of optimal frequency offset
Figure 1 shows the principle of optimal frequency offset and spectral shaping for heterodyne coherent detection. Due to the ADC-bandwidth limit in practice, the frequency offset between the LO source and the received optical signal should not be too large. As shown in Fig. 1(a), the signal spectrum out of the ADC bandwidth is cut off when a large frequency offset is used. On the other hand, the signal spectrum should also be shaped. As shown in Fig. 1(b), spectral overlap and cut-off both exist when the signal spectrum is not spectrally shaped and larger than the ADC bandwidth as well. Thus, in our scheme, we use the technique of spectral shaping with optimal frequency offset to achieve high speed heterodyne coherent detection. Figure 1(c) shows the good case with optimal frequency offset and spectrally shaped signal.
If the choice of the optimal frequency offset also takes into consideration the channel frequency spacing of the optical WDM system at the same time, it can bring out an advantage of heterodyne detection. It’s well known that, for homodyne detection, the frequency of the LO laser is exactly identical to the carrier frequency of the received optical signal, and therefore the number of the LO lasers must be equal to the number of all the WDM channels. On the other hand, for heterodyne detection, when the frequency offset is one half of the channel frequency spacing, one LO laser can be used for two neighboring WDM channels, and therefore the number of the LO lasers can be reduced into half compared to homodyne detection. Figure 2 illustrates this advantage of heterodyne detection. Take a four-channel density WDM (DWDM) signal as an example. Four individual LO lasers are needed for homodyne detection (LO 1 for Ch 1, LO 2 for Ch 2, LO 3 for Ch 3 and LO 4 for Ch 4), while only two LO lasers for heterodyne detection (LO 1 for Ch 1 and Ch 2 as well as LO 2 for Ch 3 and Ch 4). The advantage of the reduction of the LO number can make the heterodyne coherent receiver much more hardware-efficient.
3. 4 × 196.8-Gb/s PDM-CSRZ-QPSK ultra-density WDM on a 50-GHz grid
Figure 3 shows the experimental setup for the generation, transmission and heterodyne coherent detection of a 4 × 196.8-Gb/s PDM-CSRZ-QPSK ultra-density WDM signal on a 50-GHz grid with post filter and 1-bit MLSE. At the transmitter, four external cavity lasers (ECLs), with linewidth less than 100 kHz and maximum output power of 14.5 dBm, are divided into two groups. The continuous-wavelength (CW) lightwaves at 1549.0 nm from ECL 1 and at 1549.4 nm from ECL 2 are combined by polarization-maintaining optical coupler (OC) and then modulated by I/Q modulator for the upper path, while the CW lightwaves at 1549.8 nm from ECL 3 and at 1550.2 nm from ECL 4 for the lower path. The neighboring frequency spacing is 50GHz. Each I/Q modulator is driven by a 49.2-Gbaud electrical binary signal, which, with pseudo-random binary sequence (PRBS) length of (213-1) × 4, is generated from a 4 × 1 electrical multiplexer. The 4 × 1 electrical multiplexer multiplexes four 12.3-Gb/s binary signals generated from pulse pattern generator (PPG). For optical QPSK modulation, the two parallel Mach-Zehnder modulators (MZMs) in each I/Q modulator are both biased at the null point and driven at the full swing to achieve zero-chirp 0- and π-phase modulation. The phase difference between the upper and lower branches of each I/Q modulator is controlled at π/2. Next, after passing through polarization-maintaining Erbium-doped fiber amplifier (EDFA), each single-arm Mach-Zehnder intensity modulator (IM) is driven by a 24.6-GHz sinusoidal radio-frequency (RF) signal and direct current (DC)-biased at the null point, in order to generate optical CSRZ-QPSK signal with 67% duty cycle. The subsequent polarization multiplexing for each path is realized by polarization multiplexer, comprising a polarization-maintaining OC to halve the signal into two branches, an optical delay line (DL) to provide a 150-symbol delay, an optical attenuator to balance the power of two branches and a polarization beam combiner (PBC) to recombine the signal. After an OC for each path, four 49.2-Gbaud optical PDM-CSRZ-QPSK signals are spectrally filtered and combined using an AWG on the 50-GHz grid. Different from WSS, AWG is adopted in the real optical systems and networks. However, compared to WSS, the bandwidth of AWG is narrower and the top of its passband tones is less flat as well. As a result, the filtering effect due to AWG is severer than that due to WSS.
The generated signal is launched into the straight line of five/thirteen spans of 80-km SMF-28. Each span has 18-dB average loss and 17-ps/km/nm chromatic dispersion (CD) at 1550 nm without inline CD compensation. EDFA is used to compensate the loss of each span. The total launched power (after EDFA) into each span is 0dBm. Note that a programmable WSS on the 50-GHz grid is inserted after each 320 km to remove the amplified spontaneous emission (ASE) noise. Inset (a) shows the optical spectrum (0.1-nm resolution) of all the WDM channels after 50-GHz AWG, while inset (b) and (c) after 400- and 1040-km SMF-28 transmission, respectively. The required optical signal-to-noise ratio (OSNR) is 21 and 30 dB for 400- and 1040-km SMF-28 transmission, respectively. At the receiver, a tunable optical filter (TOF) with 3-dB bandwidth of 0.4 nm is used to choose the desired channel. An ECL with linewidth less than 100 kHz is used as the LO, which has 25-GHz frequency offset relative to the received signal. In our experiment, we only use two LOs, one at 1549.2 nm for channel 1 (1549.0 nm) and channel 2 (1549.4 nm) while the other at 1550.0 nm for channel 3 (1549.8 nm) and channel 4 (1550.2 nm). Two polarization beam splitters (PBSs) and two OCs are used to implement polarization diversity of the received signal and the LO in optical domain before balanced PDs each with 50-GHz bandwidth . The analog-to-digital conversion is realized in the real-time oscilloscope (OSC) with 120-GSa/s sampling rate and 45-GHz electrical bandwidth. Two ADC channels are enough for offline DSP. The detailed DSP after analog-to-digital conversion is given as follows. Firstly, the clock is extracted using the “square and filter” method, and then the digital signal is re-sampled at twice of the baud rate based on the recovered clock. Secondly, the received signals are down-converted to the baseband by multiplying synchronous cosine and sine functions, which are generated from a digital LO for down conversion . Thirdly, a T/2-spaced time-domain FIR filter is used for CD compensation, where the filter coefficients are calculated from the known fiber CD transfer function using the frequency-domain truncation method. Fourthly, two complex-valued, 13-tap, T/2-spaced adaptive FIR filters, based on classic CMA, are used to retrieve the modulus of the PDM-QPSK signal and realize polarization de-multiplexing. The subsequent step is carrier recovery, which includes residual frequency-offset estimation and CPE. The former is based on fast Fourier transform (FFT) method while the latter fourth-power Viterbi-Viterbi algorithm. A post filter is then adopted to convert the binary signal to the duo-binary one. The final symbol decision is based on 1-bit MSLE [8–10]. Finally, differential decoding is used to eliminate π/2 phase ambiguity before BER counting. In this experiment, the BER is counted over 10 × 106 bits (10 data sets, and each set contains 106 bits).
Channel 2 at 1549.4 nm is chosen when one specific channel is required for experimental measurements. For the WDM-channel experiment, the neighboring channels of channel 2 are all turned on, while for the single-channel experiment, the neighboring channels of channel 2 are all turned off. We respectively use ch2 and CH2 to denote channel 2 at 1549.4nm in the single-channel and WDM-channel experiments in the following part. All experimental measurements are carried out in the case of 25-GHz frequency offset. Figure 4(a) gives the transfer function (0.02-nm resolution) of the 50-GHz AWG, while Fig. 4(b) the optical spectra (0.02-nm resolution) for ch2 before and after the 50-GHz AWG, respectively. The 3-, 10-, and 20-dB bandwidth of the 50-GHz AWG is 26.7, 44.9 and 59.6 GHz, respectively. It can be seen from Fig. 4(b) that the optical spectrum of the single channel becomes much narrower after the 50-GHz AWG. The black curve corresponds to the non-return-to-zero QPSK (NRZ-QPSK) case, which shows that the adoption of CSRZ can broaden the spectrum of the QPSK signal. It’s also worth noting that, compared to NRZ, the adoption of CSRZ can reduce the requirement for ADC bandwidth. Figure 5(a) gives the transfer function (0.02-nm resolution) of the TOF, while Fig. 5(b) the optical spectrum (0.1-nm resolution) for ch2 after the TOF. The 3- and 10-dB bandwidth of the TOF is 49 and 61.3 GHz, respectively.
Figure 6(a) shows back-to-back (BTB) BER versus OSNR for ch2 and CH2, respectively. For ch2, the adoption of heterodyne detection causes 3-dB OSNR penalty at the BER of 3.8 × 10−3 compared to homodyne detection. Moreover, in the case of heterodyne detection, the OSNR penalty at the BER of 3.8 × 10−3 is 1.5 dB for CH2 compared to ch2. Inset (I) and (II) show the X-polarization BTB constellations (at the 30-dB OSNR) for CH2 after CPE and further post filtering, respectively. Figure 6(b) gives BER versus total launched power into fiber for CH2 after 400- and 1040-km SMF-28 transmission, respectively. The optimum BER performance is attained when the launched power is 0dBm/channel (corresponding to ~6dBm/total channels). Figure 7(a) shows BER of all channels after 1040-km SMF-28 transmission. The BER for each channel with optimum launched power of 0dBm/channel and spectral efficiency of 3.67b/s/Hz is less than 3.8 × 10−3 . Inset (I) and (II) respectively show the X- and Y-polarization constellations (at the 21-dB OSNR) for CH2 over 1040-km SMF-28 transmission after CPE, while inset (III) and (IV) after further post filtering. Figure 7(b) shows the required OSNR at the BER of 3.8 × 10−3 when the channel spacing varies. As the channel spacing decreases from 50 to 42GHz, the required OSNR increases from 21.2 to 25.7dB due to the increase of inter-channel crosstalk.
Based on a 4 × 196.8-Gb/s WDM PDM-CSRZ-QPSK transmission over 1040-km SMF-28 on a 50-GHz grid adopting simplified heterodyne coherent detection, we experimentally demonstrate that, when the frequency offset is one half of the channel frequency spacing, one LO laser can be used for two neighboring WDM channels, and therefore the number of LO lasers can be reduced into half compared to homodyne detection. An AWG instead of WSS is used at the transmitter to spectrally shape and multiplex the WDM signal. The BER for all channels is under the pre-FEC limit of 3.8 × 10−3 after 1040-km SMF-28 transmission. This work was partially supported by NNSF of China (61107064, 61177071, 60837004, 61250018), NHTRDP (863 Program) of China (2011AA010302, 2012AA011302), NKTR&DP of China (2012BAH18B00) and ICPSSTA of Shanghai (12510705600).
References and links
1. G. Colavolpe, T. Foggi, A. Modenini, and A. Piemontese, “Faster-than-Nyquist and beyond: how to improve spectral efficiency by accepting interference,” Opt. Express 19(27), 26600–26609 (2011). [CrossRef] [PubMed]
3. C. Liu, J. Pan, T. Detwiler, A. Stark, and Y.-T. Hsueh, G.-K. Chang1, and S. E. Ralph, “Joint Digital Signal Processing for Superchannel Coherent Optical Systems: Joint CD Compensation for Joint ICI Cancellation,” in Proc. ECOC2012, Amsterdam, Netherlands, paper Th.1.A.4.
4. J. Li, E. Tipsuwannakul, T. Eriksson, M. Karlsson, and P. A. Andrekson, “Approaching Nyquist limit in WDM systems by low-complexity receiver-side duobinary shaping,” J. Lightwave Technol. 30(11), 1664–1676 (2012). [CrossRef]
5. X. Zhou, J. Yu, M. F. Huang, Y. Shao, T. Wang, L. Nelson, P. Magill, M. Birk, P. I. Borel, D. W. Peckham, R. Lingle, and B. Zhu, “64-Tb/s, 8b/s/Hz, PDM-36QAM transmission over 320km using both pre- and post- transmission digital signal processing,” J. Lightwave Technol. 29(4), 571–577 (2011). [CrossRef]
6. L. G. Kazovsky, “Decision-driven phase-locked loop for optical homodyne receivers: Performance analysis and laser linewidth requirements,” IEEE Trans. Electron. Dev. 32(12), 2630–2639 (1985). [CrossRef]
7. J. Zhang, Z. Dong, J. Yu, N. Chi, L. Tao, X. Li, and Y. Shao, “Simplified coherent receiver with heterodyne detection of eight-channel 50 Gb/s PDM-QPSK WDM signal after 1040 km SMF-28 transmission,” Opt. Lett. 37(19), 4050–4052 (2012). [CrossRef] [PubMed]
8. J.-X. Cai, C. R. Davidson, A. Lucero, H. Zhang, D. G. Foursa, O. V. Sinkin, W. W. Patterson, A. N. Pilipetskii, G. Mohs, and N. S. Bergano, “20 Tbit/s transmission over 6860 km with sub-Nyquist channel spacing,” J. Lightwave Technol. 30(4), 651–657 (2012). [CrossRef]
9. J. Yu, Z. Dong, H. C. Chien, Z. Jia, M. Gunke, and A. Schippel, “Field trial Nyquist-WDM transmission of 8×216.4Gb/s PDM-CSRZ-QPSK exceeding 4b/s/Hz spectral efficiency,” in Proc. OFC/NFOEC2012, Los Angeles, CA, paper PDP5D.3.
10. Z. Dong, J. Yu, Z. Jia, H. C. Chien, X. Li, and G. K. Chang, “7x224 Gb/s/ch Nyquist-WDM transmission over 1600-km SMF-28 using PDM-CSRZ-QPSK modulation,” Photon. Technol. Lett. 24(13), 1157–1159 (2012). [CrossRef]
11. ITU-T Recommendation G.975.1, “Forward error correction for high bit-rate DWDM submarine system,” 2004.