Open Access
Add to BibSonomyAbstract
Abstract: We experimentally demonstrate the simultaneous detection of 10-Gbit/s quadrature phase shift keying (QPSK) × 2-channel Fourier-encoded synchronous optical code division multiplexing (FE-SOCDM) signals using a digital coherent receiver, for the first time. First, we analytically verify that simultaneous detection can be achieved with an N-point discrete Fourier transform (DFT) using digital signal processing (DSP) because the N-channel Fourier encoding corresponds to an N × N inverse DFT, then the operation is experimentally confirmed. Simultaneous detection of 10-Gbit/s QPSK × 2-channel FE-SOCDM signals is evaluated. The proposed scheme dramatically expands the capability of OCDM systems.
©2013 Optical Society of America
1. Introduction
Upstream transmissions of passive optical networks (PONs) will require spectral efficiency and high capability in the near future to handle the arrival of new applications. Several novel schemes have been studied for next-generation PONs, and one of the candidates is an orthogonal frequency division multiplexing (OFDM)-PON [1–3]. OFDM signals have high spectral efficiency because the spectrum of an OFDM signal has a near-rectangular shape, and a multi-level modulation format is applicable to the OFDM signals. However, problems that must be solved are the circuit size and the requirements of the transmitters. Necessary components in each transmitter are a large number of digital signal processing (DSP) resources and a high-speed digital-to-analog converter (DAC) for generation of OFDM signals. In contrast, optical code division multiplexing (OCDM) signals are simply generated using passive optical components. In an OCDM-PON that uses fiber-Bragg-grating-based optical correlators (FBG-OCs) for en/decoding [4–6], encoded signals from each transmitter are combined by an optical coupler and OCDM signals are generated. In a conventional OCDM-PON, the received OCDM signals are split by a splitter in an optical line terminal (OLT) and divided into uplink receivers of every channel for decoding. Then, the splitter in the OLT leads significant optical signal-to-noise ratio (OSNR) degradation, which results in shorter reach or more power consumption. In addition, when a multi-level modulation format is applied to the conventional OCDM-PON for improved transmission capacity, a complex digital coherent receiver, including an optical 90° hybrid, balanced photodetectors (BPDs), high-speed analog-to-digital converters (ADCs) and a digital signal processor, are required for every channel. These two issues, the serious OSNR degradation by the splitter in the OLT and necessity of multiple digital coherent receivers, are not acceptable for PON implementation. Employing multi-port optical en/decoders [7–9] may solve the first issue but does not solve the second one.
In this study, we propose and demonstrate the simultaneous detection of multi-channel Fourier-encoded synchronous OCDM (FE-SOCDM) signals with a digital coherent receiver to solve the above two issues of the conventional OCDM-PON at the same time. In the proposed configuration, an OLT is drastically simplified comparing to the conventional OCDM-PONs, because only a digital coherent receiver is employed to decode and de-multiplex OCDM signals in the OLT instead of a splitter and multiple OCDM receivers including FBG-OCs. In what follows, first, it is analytically verified that simultaneous detection can be achieved with an N-point discrete Fourier transform (DFT) using DSP because the N-channel Fourier encoding corresponds to an N × N IDFT. Then, 2-channel FE-SOCDM signals are generated by FBG-OCs and detected with a digital coherent receiver. The 2-channel FE-SOCDM signals are successfully detected simultaneously because quadrature phase-shift keying (QPSK) signals are obtained at the received channels that are equivalent to the transmitted channels.
2. Simultaneous detection of Fourier-encoded synchronous OCDM signals with digital coherent receiver
In this section, it is analytically shown that simultaneous detection can be achieved with an N-point DFT using DSP because the N-channel FE-SOCDM signals are represented with an N × N IDFT matrix. Figure 1 shows the system configuration of the proposed scheme.
In this system, FE-SOCDM signals are generated by the synchronous multiplexing of Fourier-encoded signals from each optical network unit (ONU) and simultaneously detected by homodyne reception and a DFT in the DSP. In each ONU, a short optical pulse is modulated by a baseband multi-level modulated signal sn that is a symbol at time t in channel n.
Here, N indicates the number of channels, and En and θn are the signal amplitude and signal phase, respectively. The modulated short optical pulse is encoded by an FBG-OC with an N-chip Fourier code [10]. Codewords of the Fourier code are taken from each row of the Fourier matrix, which is also known as the IDFT matrix. The Fourier matrix FN is shown in Eq. (2).Here, j is the imaginary unit, and n and k are the positive integer indices for rows and columns, respectively. Equation (3) shows an example of the Fourier matrix for N = 4.The Fourier-encoded signal gn with the Fourier code for N = 4 is represented by using Eq. (4).When the symbols among the channels are synchronized at an optical coupler, the FE-SOCDM signal x is represented as follows:Here, sT shows the transpose of vector s. Equation (5) shows that the FE-SOCDM signal x is obtained by the IDFT of the signal vector s using parallel-to-serial conversion in the optical domain.The transmitted FE-SOCDM signal is detected in a digital coherent receiver and shown in Eq. (6).
D is a 4 × 4 distortion matrix of a transmission line. The detected signal y is transposed by serial-to-parallel (S/P) conversion in the DSP.Then, yT is multiplied by a DFT matrix in the DSP, and the received signal r is obtained.When there is no distortion in the transmission line, D = I4 and Eq. (8) becomes r = s. As a result, the FE-SOCDM signals are simultaneously detected with the proposed scheme.Numerical simulation was carried out to verify the proposed scheme. Figure 2 shows the simulation model and insets (a)–(c) are waveforms of the encoded signal g1, the FE-SOCDM signal x, and the sampled signal in the DSP.
In this simulation, ideal conditions were assumed because this is a fundamental study. For example, the state of polarization was linear and singular, the lasers had no phase noise, and the electric equipments had no band limitations. An equivalent baseband model with a sampling frequency of 2 THz was used. In the transmitter, a S/P converter divided a pseudo-random bit sequence into 2 sequences that were used for modulation of both I and Q components and 5-Gbaud baseband QPSK signals modulated a short optical pulse train that had a 30-ps pulse width and a 1550-nm center wavelength. The generated optical QPSK signals were encoded by an FBG-OC that had a 20-GHz chip rate, and the encoded signal is shown in the inset (a) of Fig. 2. Then, a 4-channel FE-SOCDM signal, as depicted in the inset (b) of Fig. 2, was generated by synchronously combining the Fourier-encoded signals from each channel. After back-to-back transmission, the FE-SOCDM signals were received by a digital coherent receiver. The optical local oscillator had a 1550-nm center wavelength. The ADCs had a 20-GSample/s (GSa/s) sampling rate and no quantization. In the DSP, the sampled signal shown in the inset (c) of Fig. 2 underwent a S/P conversion, and a 4-point fast Fourier transform (FFT) was used to detect the parallel signals simultaneously.
Figures 3(a) –3(c) shows the constellation maps of the received signals of the original QPSK signals, Fourier-encoded signals with codeword c1, and 4-ch. FE-SOCDM signals, respectively. In Fig. 3(c), QPSK signals were demodulated in all the received channels with negligible degradation from Figs. 3(a) and 3(b), and the numerical simulation confirmed the possibility of the efficacy of the proposed scheme.
Fig. 3 Constellation maps of received signals. (a) Original QPSK signal. (b) Fourier-encoded signal with codeword c1. (c) 4-ch. FE-SOCDM signal.
3. Experimental demonstration of the proposed scheme with 10-Gbit/s QPSK × 2-channel Fourier-encoded synchronous OCDM signals
Two experimental demonstrations were carried out for the fundamental study. In the first experiment, 10-Gbit/s QPSK signals were encoded with a 4-chip Fourier code and the Fourier-encoded signals were detected using the proposed scheme. In the second experiment, 10-Gbit/s × 2-channel FE-SOCDM signals were generated and simultaneously detected. Figure 4 shows the experimental setup.
In this experiment, a single light source was used for the signal light and the reference light to avoid an optical frequency offset. The output light of an external cavity laser (ECL) with a 500-kHz linewidth was divided into two parts using an optical splitter. One part was used for the generation of FE-SOCDM signals, and the other part was employed as a reference light. One part of the divided continuous wave (CW) light was modulated using a cascaded electric absorption modulator (EAM) to generate short optical pulses. The generated short optical pulse, which had a 5-GHz repetition rate and a 27.8-ps full width at half maximum (FWHM), was fed into an I/Q modulator for 10-Gbit/s QPSK modulation. The QPSK-modulated signal was split into 4 parts using an optical splitter, and each split signal was encoded using FBG-OCs with the 4-chip Fourier code. Figure 5(a) indicates the structure of the FBG-OCs used for these experiments, and Fig. 5(b) shows the reflection spectra of the FBG-OCs fabricated and used for the experiments.
Fig. 5 Structure and reflection spectra of FBG-OCs. (a) Configuration of the FBG-OC. (b) Reflection spectra of FBG-OCs used for the experiments. Solid and dashed lines show the measured and theoretical reflection spectra, respectively.
FBG-OC n (n = 1, 2, 3, 4) assigned a codeword cn to the incoming signal. Multiple reflections were expected to affect the Fourier-encoded signals because the FBG-OCs had a relatively high reflectivity (over −5 dB). This led to impairment of the decoded signals. In addition, degradation of the orthogonality among the codewords is clearly found in Fig. 5(b) by comparing the measured spectra with the ideal ones. The degradation is because of phase errors in the lab-made FBG-OCs and should be improved in future. Figures 6(a) –6(d) show the waveforms of the Fourier-encoded signals.
Fig. 6 Waveforms of the Fourier-encoded signals. (a) Encoded with codeword c1. (b) Encoded with codeword c2. (c) Encoded with codeword c3. (d) Encoded with codeword c4.
It was confirmed that the Fourier-encoded signals occurred every 200 ps, and each chip pulse in one cycle had a different intensity because of multiple reflections of the FBG-OCs. It was also suggested that the decoded signals were degraded. Each Fourier-encoded signal was synchronized by an optical delay line (ODL) and inputted into an optical coupler through a variable optical attenuator (VOA) and a polarization controller (PC). In the first experiment, one of the Fourier-encoded signals was chosen and adjusted to have an optical power of −17 dBm at the output of the optical splitter. The other Fourier-encoded signals were interrupted with the VOAs and the corresponding channels were disabled. In the second experiment, two of the Fourier-encoded signals were selected and had their powers tuned with the VOAs, and the others were shut off. As a result, 2-channel FE-SOCDM signals were generated. The states of polarization between the reference light and the signal light were adjusted by PCs, and the signal light was launched into the optical 90° hybrid module and mixed with the reference light. The mixed signals were converted to electric signals with two BPDs with 42-GHz bandwidths. Each electric signal was sampled with an ADC and was acquired using a Tektronix DSA71604, which had a 50-GSa/s sampling rate and a 16-GHz bandwidth. The acquired data were processed using a personal computer, represented with complex numbers, and resampled from 50 GSa/s to 20 GSa/s for detection of each chip pulse. The sampled data were S/P-converted, and a 4-point FFT was used for detection.
Figures 7(a) and 7(b) show the results of the first experiment. The numerical simulation results are shown in Fig. 7(a). The simulation conditions were identical to those of the previous section, except for the configurations of the FBG-OCs. Imitated reflection spectra of the actual FBG-OCs depicted in Fig. 5(b) were used in the simulations in Section 3. Each sub-FBG in the FBG-OCs had a −20-dB reflectivity to avoid multiple reflections. In addition, structure identification [11] was carried out to find phase errors in the fabricated FBG-OCs, and the obtained parameters were used for this simulation. From Fig. 7(a), QPSK signals with relatively large amplitudes have been confirmed to be identical in the received and transmitted channels. However, fractional QPSK signals, which are called crosstalk in this paper, were also appeared in the disabled channels. The experimental results shown in Fig. 7(b) had the same tendency as the numerical results. The reason of the crosstalk is the phase errors in the lab-made FBG-OCs. Because of the limited accuracy of our FBG fabrication setup, the fabricated FBG-OCs had certain degree of phase errors while phase tuning by irradiating UV beams between adjacent FBGs was carried out. As clearly seen in Fig. 5(b), because of the phase errors, the fabricated FBG-OCs had fractional reflection at wavelengths where it is null in the ideal spectra and those fractional reflection resulted in the crosstalk. In addition to the crosstalk, degradation of the QPSK constellations is recognized. It was due to the poor phase noise characteristic of the light source used in the experiments, and it is expected to be improved by using a different ECL with a narrower linewidth or a kind of digital compensation algorithm.
Figure 8 shows the constellation maps of the simultaneously detected signals of the 10-Gbit/s QPSK × 2-channel FE-SOCDM signals. Ideally, demodulated QPSK signals are observed in the magenta-colored constellation maps. From the simulation results depicted in Fig. 8(a), it is found that the constellations were spread over all the received channels because the crosstalk signals were superimposed. However, 4 bundled constellations of QPSK were observed in the same received channels as the transmitted channels.
Fig. 8 Constellation maps of detected signals of the 10-Gbit/s QPSK × 2-channel FE-SOCDM signals. (a) Simulation results. (b) Experimental results.
In Fig. 8(b), the constellation maps in the received channels have different phase rotations because there was no phase synchronization between the Fourier-encoded signals. The phase rotations in Fig. 8(b) can be compensated using the digital phase estimation technique [12]. Meanwhile, for transmitted channels (2, 3), the demodulated QPSK signals were particularly degraded at the received channel 2 in the experimental results, as shown in Fig. 8(b). It is expected that the reasons for the degradation were not only the phase noise of the ECL and the phase errors and high reflectivity of the FBG-OCs but also misalignment of the center wavelength of an FBG-OC due to temperature fluctuations. Even though the FBG-OCs included significant phase errors, the demodulated QPSK signals have been confirmed to be identical in the received and transmitted channels (1, 2), (1, 3), (1, 4), (2, 4), and (3, 4). When FBG-OCs, precisely fabricated by a professional manufacturer, are employed, the crosstalk would be significantly reduced and it is expected to obtain better constellation maps. A kind of digital signal processing may be also utilized to compensate the crosstalk. Those investigations have been continued.
4. Summary
In this paper, we proposed and experimentally demonstrated the simultaneous detection of FE-SOCDM signals with a digital coherent receiver. First, it was numerically confirmed that simultaneous detection of 10-Gbit/s QPSK × 4-channel FE-SOCDM signals could be achieved by phase diversity homodyne detection and a 4-point FFT in a DSP. Then, we experimentally demonstrated simultaneous detection of 10-Gbit/s Fourier-encoded QPSK signals and 10-Gbit/s QPSK × 2-channel FE-SOCDM signals. The results of this study indicated that the proposed scheme accomplished simultaneous detection of FE-SOCDM signals and dramatically expanded the capability of OCDM systems.
References and links
1. C. Zhang, C. Chen, Y. Feng, and K. Qiu, “Experimental demonstration of novel source-free ONUs in bidirectional RF up-converted optical OFDM-PON utilizing polarization multiplexing,” Opt. Express 20(6), 6230–6235 (2012). [PubMed]
2. N. Cvijetic, M. F. Huang, E. Ip, Y. Shao, Y. K. Huang, M. Cvijetic, and T. Wang, “1.92 Tb/s coherent DWDM-OFDMA-PON with no high-speed ONU-side electronics over 100 km SSMF and 1:64 passive split,” Opt. Express 19(24), 24540–24545 (2011). [PubMed]
3. N. Cvijetic, M. F. Huang, E. Ip, Y. Shao, Y. K. Huang, M. Cvijetic, and T. Wang, “1.92 Tb/s coherent DWDM-OFDMA-PON with no high-speed ONU-side electronics over 100 km SSMF and 1:64 passive split,” Opt. Express 19(24), 24540–24545 (2011). [CrossRef] [PubMed]
4. M. Hanawa, Y. Okamura, S. Nozaki, and K. Hosoya, “Experimental demonstration of optical concatenated coding enabling channel grouping on OCDM networks,” in International Conference on Optical Internet 2010 (COIN 2010), Korea, 383–385 (2010).
5. M. Hanawa, S. Nozaki, K. Hosoya, Y. Okamura, and K. Nonaka, “BER characteristics of 2ch OOK-OCDM using 16-chip concatenated Fourier code,” in Opto-Electronics and Communications Conference (OECC 2011), Taiwan, 24–25 (2011).
6. Y. K. Choi, K. Hosoya, C. G. Lee, M. Hanawa, and C. S. Park, “A hybrid WDM/OCDMA ring with a dynamic add/drop function based on Fourier code for local area networks,” Opt. Express 19(7), 6243–6252 (2011). [CrossRef] [PubMed]
7. N. Kataoka, N. Wada, X. Wang, G. Cincotti, A. Sakamoto, Y. Terada, T. Miyazaki, and K. Kitayama, “Field trial of duplex, 10 Gbps × 8-user DPSK-OCDMA system using a single 16 × 16 multi-port encoder/decoder and 16-level phase-shifted SSFBG encoder/decoders,” J. Lightwave Technol. 27(3), 299–305 (2009). [CrossRef]
8. T. Kodama, N. Kataoka, N. Wada, G. Cincotti, X. Wang, T. Miyazaki, and K. Kitayama, “High-security 2.5 Gbps, polarization multiplexed 256-ary OCDM using a single multi-port encoder/decoder,” Opt. Express 18(20), 21376–21385 (2010). [CrossRef] [PubMed]
9. N. Kataoka, G. Cincotti, N. Wada, and K. Kitayama, “Demonstration of asynchronous, 40 Gbps x 4-user DPSK-OCDMA transmission using a multi-port encoder/decoder,” Opt. Express 19(26), B965–B970 (2011). [CrossRef] [PubMed]
10. M. Hanawa, “Fourier code: A novel orthogonal code for OCDM systems,” in Opto-Electronics and Communications Conference and Australian Conference on Optical Fibre Technology (OECC/ACOFT’ 2008), Sydney, 1–2 (2008).
11. M. Hanawa, K. Nakamura, and K. Osada, “Structure identification of superstructure fiber Bragg gratings by the least mean square algorithm,” in Opto-Electronics and Communications Conference (OECC 2005), Korea, 820–821 (2005).
12. D.-S. Ly-Gagnon, S. Tsukamoto, K. Katoh, and K. Kikuchi, “Coherent detection of optical quadrature phase-shift keying signals with carrier phase estimation,” J. Lightwave Technol. 24(1), 12–21 (2006). [CrossRef]
